Short Course on

Crop Weather Modeling

(Sponsored by ICAR)

13-22 December 2011

Lecture Notes

AICRP on Agrometeorology
Central Research Institute for Dryland Agriculture
Santoshnagar, HYDERABAD – 500 059
BASIC ASPECTS OF MODELING SOIL-PLANT-ATMOSPHERE CONTINUUM

P.S.N. Sastry
17-1-391/17, Subrahmanyanagar
Saidabad colony, HYDERABAD-500 059.
e-mail: [email protected].

What is a Model?

It is a simplified description
(often, a mathematical representation)
of a system to assist
calculations and predictions.

In the present context, ‘model’ is expressed as a computer program that can be repeatedly
run several times for computing several designed mathematical or statistical expressions (equations)
governing crop growth-environment relations, given appropriate input data.

Simulation: This is the reproduction of an observed phenomenon (e.g., growth of biomass with
time; water use by a growing crop etc.,) by developing a model and a computer programme written
for it. Such a programme usually is comprised of mathematical, statistical, physical, graphical or
empirical expressions relating the various parameters given as input information or data.
Model is a concept; simulation helps reproduction of a system in the laboratory using the concept;;
could contain measurable or estimated parameter values or both. Most often, the computer
programme written for any particular purpose is itself called a model.
Crop-environment interactions are unlimited in number. They can be studied from several points of
view (physical, physiological, chemical, biochemical, bio-technological, agronomical,
entomological or pathological, economic benefit angles etc.,). We have the roots growing with
passage of time and interacting with soil, taking up water and nutrients for transport to the above-
ground parts of a plant. The stem, branches, leaves as they grow interact with environment (both
individually and together), under the influence of solar radiation to produce flowers and pods / oils,
1
grains— ultimately yield. Evapotranspiration, Leaf-air interactions, Photosynthesis, respiration,
carbon dioxide assimilation, are the other processes involved in crop growth. Crop is also affected
by pest/disease incidence. Thus crop growth is usually viewed as a “complex system” which
comprises of “sub-systems” in which several processes take place. One process leads to the other
and so, individual processes (water or nutrient uptake by roots, biomass accumulation, grain growth
etc.,) are considered as “sub-systems”. All the processes which interact among themselves (since the
start of growth of plant from seeding to final yield) and put together are considered as a “system”.
Thus one can have “sub-models” as part of a “model”, “sub-systems” and a “System”. One can
simulate water uptake, branching pattern and growth, leaf development, pod growth, etc. and their
interaction with soil and aerial environment, as individual models. The point to note is that there is
no limit to the items that can be taken up to develop a simulation model.
Systems analysis models:
Modeling several of the soil-plant-atmosphere-water interactions which are mutually
dependent on each other resulting in crop growth, popularly known as the SPAW-system, is a linked
single entity of sub-systems. System models could also include economic factors such as operating
costs, cost-benefit ratios from the time land is prepared, till transport and marketing of the produce.
Examples of systems-model are the Oryza model for rice, CERES maize model, DISSAT models
etc., which have several component sub-systems.

Subsystems: These are parts of a complex ‘whole’ which themselves could be viewed
independently where needed. Rainfall-yield model, Soil moisture distribution model, rainfall-run off
model, root growth model etc., are all sub-systems. Interaction among leaf-atmospheric vapour,
stomatal resistance, air stream adjoining the leaf surface, net radiation could be the parameters of a
theoretical (mechanistic) subsystem model development. Each such objective can become a sub-
model material. In modeling crop-weather interactions, possibilities are immense and limitless.
Ultimately, such subsystems can be appropriately connected to evolve them into a single ‘System-
model’.

Mechanistic process models:

A process model is an elaborate and practically complete description of the mechanism
involved in a process—e.g., photosynthesis, green or dry matter production, partitioning of
2
photosynthates, soil water uptake and transport by the root system etc., Such models for crop
growth usually are designed to compute the products daily to simulate growth of a plant including
all known processes (the underlying mechanisms) in the soil-plant-water-environment system.
These could include water-fertilizer uptake and their transport, effect of flood and water logging,
effect of pest-disease incidence etc., popularly known as the dynamic crop growth simulation
models, since, given the relevant data input, they are designed to compute day-to-day expected crop
growth as a result of several growth related phenomena that ultimately influence the yield. .

Operational models:
On the other hand, operational models(for day-to-day field operations) in relation to the
SPAW system can be developed to simulate crop growth using known relations (statistical,
empirical, mathematical or graphical models) depending on data availability, regional and local
crop-environmental conditions for growth, including or bypassing some of the mechanistic details
involved in the system. Also, different models need to be developed for space and time variations
involved. In this exercise, area of operation (could start with a village), time-duration of individual
crop growth phases and seasonal factors characteristic of crop species are involved. After gaining
experience with such operational models at a few diverse locations, they can be modified or
integrated to extend them from local (a village) to a district or agroclimatic regional level. In this
process, several modifications may be needed. They can be “user targeted” to find an answer to
several day-to-day crop-weather related problems encountered in the field. For example, an
operational model can be developed to answer a question such as: How many days would it take for
the field to be free from water logging after a heavy rainfall for a couple of days? How much
reduction in yield could one expect due to continuous high temperature period for four to five days
at pod formation stage? Is the crop suffering from agricultural drought or atmospheric drought?
What would be expected reduction in yield? Such individual models not forming a systems model,
ultimately lead to development of “EXPERT” systems.
It should be recognized that unforeseen contingencies cannot be modeled and local on-the-
spot decision has to be taken. Flooding due to breeches in bunds, water logging due to very heavy
local rainfall, gale strength winds, thunderbolts are some of the examples. After immediate remedial
action is taken against these calamities, if the crop gets destroyed, crop-weather contingency models
can then be used for alternate contingency crop planning for the remaining part of the season.
3
Statistical models and dynamic simulation models
Crop-weather modeling has two approaches (i) statistical (ii) dynamic simulation modeling.
Statistical approach has found wide application since the early 20th century but it has several
limitations for application in operational agriculture at the present time. Dynamic simulation
approach similarly has both advantages and disadvantages. Process oriented approaches are
considered desirable for establishing rate-processes and linkages in the soil-plant-atmosphere-water
flow system. They have their own role to play more as research tools and for yield forecasting rather
than for field operations.
Statistical models are developed using long term (say 20-30 year series) average values over
a long period between two or more parameters—say rainfall and crop yield. Statistical functions like
linear, curvilinear, multiple regression, orthogonal polynomials (depending on the number of
parameters and data availability) etc., are developed between these two or a few more parameters.
Their variability and significance are tested using rigorous procedures and ultimately a regression
function is finalized. These could assist in making long-term assessment of crop performance on an
average over a couple of decades but given the vagaries of monsoon rainfall, such regressions, more
often than not, fail in an individual year. As an example, in semi-arid regions, rainfall variability
being high (>100%), applicability of such regressions may fail in an individual season. So, in
practice, it becomes unusable except to understand the extent of association between rainfall,
temperature etc., and yield in general in a locality over a long period. This is a limitation of such
regression models in the tropical or sub-tropical region like ours. Often the experience in the All
India Coordinated field trials, is that one year the crop-weather parameter association comes out as
significant while the very next year it could be non-significant association leading to erroneous
interpretation.

Dynamic simulation models:

On the other hand, dynamic simulation models seek to compute such growth values on a day
to day basis using the relations between crop growth parameters and weather parameters. It seeks to
rebuild the day to day crop growth in mathematical or mechanistic terms (simulation) depending on
the magnitude of rainfall (or any other parameters) on a particular day and magnitude of a crop
parameter (or other parameters like physiological, soil, biological parameters) representing crop
growth till that day. i.e., daily simulation is done depending on the parameter values obtaining on a
4
day and cumulated over the growth period.. Such simulation is continued till harvest time.
“Growing the crop on the computer” is a popular expression.
It is not essential to run a mechanistic model for all purposes. In subsystems of such models, several
assumptions are made with boundary conditions. One need not always look to full-scale
mechanistic models and systems approach for finding solutions to day-to-day problems arising in
agriculture operations imposed by short-period adverse weather conditions. Individual subroutines
can be utilized profitably.
Development of a model:
Objective: Clarity of objective or purpose is significant and essential.
Objective of developing a particular model is the first step. It should be derived from intended end-
use of any model and clearly identified. The objective could be (i) for academic
understanding(research purpose) of crop growth dynamics (ii) for monitoring crop growth for any
possible field action including prediction of crop growth and yield (iii) for solving field level
(extension) problems (iv) for crop planning in relation to climate change or climate variation, for
introduction and assessment of new varieties etc.,. Parameters used may remain same but depending
on the purpose of developing a model, the structure of the model differs. After the objective is
decided, it is customary to prepare a flow chart. It is a framework depicting the different steps to be
followed like reading the data, computing different components, repetition of calculations if any,
print format desired etc., in achieving the objective of the model. Sample flow chart is shown in
fig.1.
Field level problems need an immediate answer for decision-making where both time and space
scales are involved. It could range from a single village to a region, or a single day to a week or
even a season, depending on the nature of the day’s weather (a heavy shower of rain) or movement
of a weather system (depression, cyclone, heat wave etc.,) over a period of time. So, models need to
cater to several such widely varying specific objectives. A thermometer developed for macro scale
use (Stevenson screen) obviously is not suitable for measurement of micro-climate. Similar is the
case with crop-weather models. Models cannot be expected to have universal application.
To illustrate, one can ask the question: Do we know how the crop growth in its various facets (water
redistribution in the cropped soil, pest/disease development, nutrient distribution, biomass
accumulation, partitioning of dry matter etc.), is affected by say a rainfall of 20mm received on a
single day at different growth stages of the crop? Or if the same rain occurs in three or four
5
consecutive days or falls intermittently, what would be the effect on crop growth or agricultural
operations? If there is a rise or fall of temperature during a heat or cold wave at a particular growth
stage how is crop growth rate affected and when would the temperatures return to the normal?
These questions look simple. Linear equations built in a model do not serve such purpose. More “if”
“go to” statements are needed in the computer programme. As yet we do not have immediate
answer in quantitative terms for these contingencies as to their effect on crop growth. These are of
immediate practical utility. Models can be developed for finding solution to such problems. They
would be different from the models developed for research / academic purposes.

Fig.1. A simplified flowchart of ‘BRASSICA’ model

6
Sub-models:
Each sub-model is geared to provide quantitative relationship between the parameters involved. For
example, root growth subroutine provides information on root growth rates with time, soil depth and
moisture for a particular crop and soil type, which are of practical utility in working out water
balance or irrigation depth and needs of a growing crop. Rainfall-runoff sub-model can provide
information on how much of the rain received on a day (a heavy shower) would infiltrate into the
soil and get redistributed depending on the rainfall intensity, antecedent soil wetness and root
growth.

Most of the time our concentration is on testing readymade models using single plot approach
without nutrient or moisture constraint.. While this is essential to generate data under controlled
conditions, to date, we do not know how to translate these plot level results to regional level even if
it is as small as a taluk / mandal. Hence one cannot expect homogeneity in any parameter and one
has to deal with applying results derived under homogeneity to a heterogeneous larger region with
high spatial variability. This heterogeneity cannot be avoided as farmer’ practices and capabilities
vary widely and exhibited both in time and space. It is also not practicable to generate data from
every plot in the region.
Graphical and Checklist models:
Besides simulation models, graphical, parametric or checklist models are also useful in day-to-day
work in field operational decisions. These are developed from thumb rules from past experience and
simple relations between crop growth and related environmental parameters For example, at a
particular growth stage of crop, afternoon humidity more than 60percent, a brief rainfall of 3mm or
more, temperature between 25 to 30°C is known to initiate a pest/disease development, Such
information can be displayed in a graphical form everyday and marked ‘favourable’ or
‘unfavourable’ using weather data. Normally three or four such favourable days would be needed
for the pest/disease symptoms to appear on the crop. Since organisms are viable, even when one or
two days are not favourable in-between, such favourable days can be counted to make a prediction
about the onset of pest/disease over the crop. A mere glance at the chart would reveal the situation.
No computer model need be run. The country needs such simple models, easy to develop into
“EXPERT” systems without much sophistication.

7
CROP-ENVIRONMENT MODELS.

A pertinent question in the above context is: Is it always essential to use mechanistic models? The
answer is “NO”. Mechanistic models are more geared towards research and need several data inputs
which are always not measurable or experimented at every location. This is a physiological
approach, and such models also involve several approximations and estimates often resulting in
deviations from the actual. Crop-weather models should preferably be designed as operational
models needing weather and agronomic data with no genetic coefficients involved, or not always
requiring potential conditions of moisture or nutrients etc., Rainfed agriculture being a dominant
practice in the country, with potential conditions being absent in several seasons, rainfall driven
models are our need. A few models are listed below that can be written as statistical or dynamic
simulation models or both and are depicted in fig.2.which shows different pathways to design a
crop-weather model. These are only a few examples and not an exhaustive list.

Rainfall- yield model (atmospheric drought, flood)

ET-biomass-yield model ---(Yield potential)
Rainfall-soil moisture distribution model – (ET, Water balance)
Rainfall-soil water balance-yield model.
Solar radiation interception, LAI, green/dry matter production (Remote sensing)
Rainfall intensity / surface run-off model ----- (water harvesting)
Root zone moisture linked to growth phase ---- water availability to crop
Yield potential models without any constraints (for water, soil, pest disease, nutrient etc.,)
Water –nutrient uptake ---yield models
Yield potential models with constraints like drought, water logging, pest/disease etc.,
Yield correction models for catastrophic or disastrous events linked to growth phase
Storms and cyclones
Heat and cold waves

8
Advection effects

Fig.2. Crop – Environment Models

Concentration should be on
Models relevant to weather based agro-advisories
User targeted models
Phenology as driver—phenology based models
Minimal mechanistic processes—more crop-weather parameters

If you get to successfully write a flow chart and computer software for “How to make tea” or “How
to cook rice”, you would have learnt the logic of programming and model development. Training
should provide capacity to write computer programmes ourselves and not be a slave to readymade
programs that we may not care to thoroughly scrutinize and understand before applying them.

9
 Phenology, Biomass and its partitioning in crops
P.Vijaya kumar
Senior Scientist (Ag.Met),
CRIDA,Hyderabad

1. Introduction:

Estimation of the timing of the Phenological stages and partitioning of biomass are two
important components of any crop simulation model. Accurate representation of phenology is
essential when using crop simulation models to study the effect of changes in climate or
management practices (e.g. planting date, cultivar selection or irrigation scheduling) on crop yield.
Unfortunately, the timing of phenologiocal stages is not always well simulated when models are
used in environment or management conditions differing significantly from those for which they
were developed (White et al., 2003). Allocation (Partitioning) of total drymatter to different organs
of a plant is of great importance in crop growth, development and yield. There is a great diversity in
the way crops partition assimilates and simulation models developed so far are species–specific.
Within species, genotype, development stages of plant, growth conditions and internal regulations
by the plant may also affect drymatter allocation (Marcelis 1996).The simulation of biomass
partitioning is one of the weakest features of current crop growth models.
Phenology or stages of crop development, crop growth rate, partitioning of biomass in to
growing plant organs accomplish the state of a crop at any time. All these processes are dynamic
and are affected by weather, soil and cultivar specific factors. The objective of the presentation is to
describe the processes, their estimation and importance to crop simulation modeling.
2.Basic principle of crop simulation:
In simplest form, total biomass (Bt) of a crop can be written as the product of average
growth rate (g) and growth duration (d) i.e
Bt = g × d
Simulation of yield in process based models must predict these two important variables. The
duration of plant growth also has two district features viz., phasic and morphological development.
Phasic development involves changes in the stage of growth and is always associated with major

10
changes in biomass partitioning. Phasic development of crop in the models quantifies the
physiological age of the plant and describes duration of various growth phases.

3.Phenology:
Phenology is the study of periodic biological phenomena. It qualitatively describes the
successive stages in the development of plants, from seed germination to flowering to maturity. The
crop growth stages, also known as phenophases differ from crop to crop. Only main stages viz.,
germination, differentiation of flower, flowering seed formation, seed filling and maturity are
common to almost all flowering plants.
Phenology can be modeled based on vernalization, photoperiod, thermal response and
intrinsic earliness (Cao and Moss 1997), most of which are plant specific. The environmental and
cultural factors directly or indirectly influencing crop phenology are : Atmospheric Carbon dioxide
(indirect ), Solar radiation (indirect ) , photo period (direct on flowering), Temperature (direct) ,
water (Indirect) ,wind (indirect), nutrients(N,P,K) ( both direct and indirect) and growth
regulators (indirect).
Temperature and photoperiod are the two main environmental factors that determine flowering in
young and established plants.
3.1. Effect of temperature on phenology:
Temperature plays an observable effect on rate of development of plants and the effect is significant
not only in temperate countries but also in tropical countries (Malhood, 1997)
Growing degree-day Concept;
Heat units, expressed in growing degree days (GDD) are widely used to describe the timing of
biological process. Growing degree-days (GDD), also called heat units, effective heat units, or
growth units, are a simple means of relating plant growth, development, and maturity to air
temperature. The concept is widely accepted as a basis for building phenology and population
dynamic models. Degree-day units are often used in agronomy, essentially to estimate or predict the
lengths of the different phases of development in crop plants (Bonhomme, 2000).
The GDD concept assumes a direct and linear relationship between growth and temperature.
It starts with the assumption that the growth of a plant is dependent on the total amount of heat to
which it is subjected during its lifetime. A degree-day, or a heat unit, is the departure from the mean
daily temperature above the minimum threshold (base) temperature. This minimum threshold is the
11
temperature below which no growth takes place. The threshold varies with different plants, and for
the majority of plants it ranges from 4. 5 to 12.5°C, with higher values for tropical plants and lower
values for temperate plants.
3.2 Methods of Degree-Day Estimation
Many methods for estimating degree-days are available in the literature (Perry et al., 1997; Vittume
et al., 1995). The three most dependable and commonly used methods are the standard method, the
maximum instead of mean method, and the reduced ceiling method. Numerous other methods have
been proposed, majority of them are modifications of one of these three. An exhaustive review of
degree-day methods was reported by Zalom et al. (1993).
1. Standard degree-day method:
GDD =∑ [(Tmax + Tmin)/2] – Tbase (1)
where (Tmax + Tmin)/2 is the average daily temperature and Tbase is
the minimum threshold temperature for a crop.
2. Maximum instead of means method:
GDD = ∑ (Tmax – Tbase) (2)
3. Reduced ceiling method: where Tmax ≤ Tceiling, then
GDD = ∑ (Tmax – Tbase), or (3)
where Tmax > Tceiling, then
GDD = ∑[( Tceiling - (Tmax - Tceiling )) - Tbase] (4)
If maximum temperature (Tmax) is greater than the ceiling temperature (Tceiling), then set Tmax
equal to Tceiling minus the difference between Tmax and Tceiling.

Over the years, many equations have been proposed to substitute the GDD method.

The equation below, recently published by Xinyou Yin et al. (1995) seems to give reasonable results
under field conditions (which does not mean that it would yield useful results at a regional scale).

The graph below (figure 1) illustrates the behaviour of a more sophisticated model proposed by
Xinyou Yin et al. (1995), based on the beta distribution. The development rate DR is given
by

12
 DR= еµ (T–Tb)α (Tu - T) β (5)

where Tb and Tu are the base and upper values. µ is a size parameter, while α and β are
shape parameters.
For a given phenological interval (planting to emergence, of heading to maturity, etc.), the
development rate is the reciprocal of the time in days to complete the phase.

The curve culminates at To (o for optimum). To is a function of the other so-called cardinal
temperatures:
To= (αTu + β Tb )/ ( α + β) (6)
The maximum development rate Ro is computed by substituting To in the equation for DR above.
figure 1 : Variation of development rate (Rice 1R8, from sowing to flowering) as
a function of temperature using the Beta model of Xinyou Yin et al. (1995). T
b is taken as 8 C and Tc is 42 C. The other parameters are
µ= -15.6721; α= 2.5670; β= 1.3726.
Development rate

0.014

0.012

0.01 Tu

0.008

0.006
T b
0.004
To
0.002

0
0 5 10 15 20 25 30 35 40 45

Temperature (C)
13
4. Photoperiod

Plants can be categorized as long-day plants, short-day plants and day-neutral plants. F lower
differentiation is initiated in long-day plants by a threshold of day length below which the plants
will not flower. Above the threshold, there is an optimum daylength. Similarly, short-day plants
will not flower if the day exceeds a threshold. To some extent, the photoperiodic response is
independent of growth: if plants are grown outside the optimum time of the year, they may flower in
very early stages (millet) or never flower at all if the proper daylength is not available.
Duration of the life cycle and, therefore, rate of development, may be influenced by photoperiod
during one or more phases of development (Fig. 2). Duration of the life cycle in short-day plants can
be increased when plants are growing in environments with a day length longer than 12 hours (or
when the duration of the dark period is less than 12 hours).
Effects of photoperiod on duration of the life cycle vary with phase of development, i.e., at least
three phases can be distinguished:
Phase 1: A photoperiod-insensitive phase. Photoperiod does not influence time to flower
initiation during the juvenile phase, a phase that starts at sowing.
Phase 2: A photoperiod-sensitive inductive phase. This phase extends from the end of the
juvenile phase to flower initiation. Flower initiation is advanced or postponed by the photoperiod
during this phase.
Phase 3: A photoperiod post-inductive phase. This phase may comprise photoperiod sensitive
and/or photoperiod insensitive phases; photoperiod during this phase may influence the duration
from flower initiation to physiological maturity.

14
Fig 2. Duration of time to flowering or maturity in response to photoperiod in a short-day or long-
day plant

5 Vernalization.
Vernalization can be seen as the need for seeds or plants to be exposed to a cold threshold
between T1 and T2 (T1 < T2). It also constitutes a mechanism to avoid frost damage.
Temperatures below T1 will kill the plant, while if temperature stays above T2, plants will not
develop. This may be combined with the duration of exposure: a shorter exposure is sometimes
sufficient close to T1, while the vernalization duration is much longer close to T2.
Other environmental factors can have a "photoperiod like" trigger response in plants. For instance,
some cereal crops (e.g., winter wheat and barley cultivars originating in northern Europe and
Canada) have a specific requirement for a period of low temperature for floral initiation (the
vernalization requirement). In other crop species (such as tulip), seeds, bulbs, seedlings, and/or
plants have to be exposed to a specific temperature regime for floral initiation to occur.
Temperature, solar irradiance, soil nitrogen, and soil moisture can also have a small effect on final
leaf number in maize, thereby changing time to flowering and maturity. The photoperiod and
15
photoperiod like responses appear to influence the duration of the life cycle by affecting translation
and/or transcription of the genetic code that triggers differentiation.
Wheat and barley varieties usually require relatively low temperatures before spikelet formation can
begin. This low temperature requirement for flowering, called vernalization , begins at germination .
The optimum temperature for vernilization is assumed to be in the range of O to 70 C ,with
temperatures between 7 and 180 C having decreasing influence on the process.
6.Biomas growth
There are several approaches to the estimation of biomass production, which range from the
estimations of photosynthesis and respiration to direct estimation of biomass production from
radiation interception or transpiration of crops. Simple and most common method is the estimation
of biomass from radiation interception and radiation use. In CERES models biomass production is
estimated as follows.
Potential biomass production (PCARB)=RUE × IPAR
Where RUE is the radiation use efficiency and IPAR is the fraction of PAR intercepted by plants.
The actual daily biomass production (CARBO) may be less than PCARB because of non-optimal
temperature or deficits of water or nitrogen. The equation to calculate CARBO uses the law of
limiting concept to reduce PCARB.
CARBO=PCARB×MIN (PRFT, SWDF1, NDEF1,1)
Where PRFT, SWDF1 and NDEF1 are the temperature, water deficit, and nitrogen deficit factors,
respectively, varying between 0 and 1, and min indicates the minimum value of the parenthesis is
used. Another approach used in the APSIM suit of models is the estimation of biomass production
by two methods each day, one limited by available water for transpiration, and the other limited by
radiant energy. The minimum of these is the actual biomass production for the day.
Biomass production = transpiration × transpiration efficiency
Biomoss production = RUE ×radiation interception
RUE incorporates the temperature, water-logging (oxygen deficit), and nitrogen stress effects.
7. Biomass partitioning and translocation
On the day of emergence, biomass (and nitrogen) allocation to leaves, stems, and roots is initialized.
After this biomass produced is allocated to different plant parts as function of growth stage and
evaluates each day the sink capacity of the above ground biomass to determine if the crop is sink-
limited or source-limited (or the supply –demand limited).
16
The following procedures are followed in the models.
Allocation to roots
Roots are grown as per the root-shoot ratio specified for each growth stage.
Above ground biomass allocation in legumes
Emergence to flowering: A proportion of biomass produced during this phase is partitioned to leaf
and stem. If leaf demand for assimilates is less than supply, the residual is partitioned to stems.
Likewise, if supply is less than leaf demand, the rate of leaf area increase is reduced.
Flowering to start of grain fill: The same procedure is used for determining leaf biomass as for
the emergence to flowering phase. Of the remaining carbon, a proportion goes to stem and pod in
the specific ratio.
Start to grain –fill to maturity: The biomass produced during this phase is portioned among pod
including grain, and stem. If grain demand is lower than supply, the remaining goes to leaf and stem
as per amounts specified for the growth phase.
Above ground biomass allocation in cereals
Emergence to terminal spikelet: Of the daily biomass produced , 65% is portioned to leaves and
the rest goes to stems. This may vary among different cereals.
Terminal spikelet to flag leaf: After terminal spikelet leaf biomass fraction is linearly decreased
with the fraction of thermal time to zero at flag leaf. On the day the estimates of specific leaf area
(SLA) goes below the minimum SLA, the extra biomass is diverted to roots.
Flag leaf to beginning grain fill: Leaf growth ceases and all the above ground biomass increase is
assumed to go to the functional stems.
Beginning grain fill to end of grain fill: Biomass increase is used to meet grain demand first and
the rest is put in to stems.
Reference
Bonhomme, R. (2000). Beware of comparing RUE values calculated from PAR vs solar radiation or
absorbed vs intercepted radiation. Field Crops Research 68: 247-252.
Perry, K.B., Wu, Y., Sanders, D.C., Garrett, J.T., Decoteau, D.R., Nagatta, R.T., Dufault, R.J.,
Batal, K.D., Granberry, D.M., and Mclaurin, W.J. (1997). Heat units to predict tomato harvest in the
southeast USA. Agricultural and Forest Meteorology 84: 249-254.
Cao, W. And D.N. Moss, 1997. Modelling phasic development in wheat: a conceptual
integration of physiological components. J. Agric. Science, 129(2):163-172.
17
Vittum, M.T., Dethier, B.E., and Lesser, R.C. (1995). Estimating growing degree days. Proceedings
of the American Society for Horticultural Science 87: 449-452.
Xiniou Yin, M.J. Kropff, G. McLaren and R.M. Visperas, 1995. A nonlinear model for
crop development as a function of temperature. Agric. Eor. Meteorol., 77:1-16.
Zalom, F.G., Goodell, P.B., Wilson, L.T., Barnett, W.W., and Bentley, W.J. (1993). Degree-Day:
The Calculation and Use of Heat Units in Pest Management. University of California, Division of
Agricultural Sciences Leaflet 21373.

18
 Interception of Solar Radiation, Radiation and Water-use Efficiency

Dr. A.S. Rao

(Principal Scientist (Agrometeorology), Div. of Resource Management

Central Research Institute for Dryland Agriculture, Hyderabad – 500 059

01. Solar Radiation and Solar Constant

Introduction:
Plants are the nature’s converters of solar energy into useful form of the energy, which is
biomass and sustain biological life providing food. fodder, fiber, fuel etc. Solar energy fulfils
two essentials need of the plants –
1. Light for Photosynthesis
2. Thermal Conditions required for all normal physiological functions of the plants
Radiation increases evapotranspiration. Transpiration rate increases almost in proportion to the
intensity of the solar radiation, while in many crops, the rate of the photosynthesis increases less
rapidly, Crop production in agriculture is basically dependant on quality and quantity of
radiation.
Radiation:
The transfer of thermal energy in the form of electromagnetic waves from one place to another
through the vacuum with speed of light is called radiation. Every material in our vicinity--soil,
water, plant, animal etc. with temperature greater than absolute zero emits characteristics
radiation specific to it’s own body temperature. Thus, all bodies are in interaction with other
bodies through radiation process .
“Solar radiation is defined as “The flux of radiant energy from the sun. The variations
of the total radiation flux from one site to another on the surface of the earth are enormous
and the distribution of plants and animals responds to this variation”.

19
Solar Constant:

The flux of solar radiation that is received on a unit area of surface held perpendicular to the
sun’s direction at 0

the mean distance between the sun and the earth is referred to as solar constant.

The solar constant is 56 x 1026 cal energy min-1.The mean distance of the earth from the sun
is

1.496 x 1013 cm. The energy per unit area incident on the spherical shell of the earth
concentric with the sun with this radius is :

Total radiation = 56 x 1025 cal cm-2min-1

Total area of spherical shell 4π(1.5 x 1013 )2

= 1.979 cal cm-2 min -1 OR 2 Langley min -1

02. Important Terms and Definition

 Forms of Solar Radiation  Transfer of heat can occur in 3 ways

 Direct Solar Radiation  Conduction: transfer of heat through solids
 Sky radiation or diffused radiation  Convection: transfer of heat by circulation
 Global radiation  Radiation: transfer of heat by electromagnetic
 Reflected radiation waves
 Thermal radiation
 Net Radiation

20
 Electromagnetic Energy: Solar energy is an electromagnetic phenomenon. In terms of
electric energy, the radiation from the sun attenuated by atmosphere as incident upon the earth is

equivalent to 1390 watts m -2 or roughly, 1.4 kilowatt m-2

 Electromagnetic Spectrum: It is total range of radiation from short to microwave.
In other words, it is the continuous sequence of electromagnetic energy arranged
according to wavelength or frequency
 Solar Spectrum: A very broad part of the electromagnetic spectrum occupied
by wavelength from which the radiant energy received from the sun is spread.
 Global Radiation: Sum of the short wave radiation directly from the sun and
indirectly from the sky.
 Insolation: Incoming solar radiation from the sun that strikes the earth‘s surface over
a finite time period such as day.
 Infrared Radiation: Solar radiation higher than visible wavelength segment
wavelength between 07 and 50 microns
 Ultraviolet radiation: Part of solar spectrum having wavelength shorter than visible
segment and range from 0.0005 to 0.4 microns.
 Light: Middle part of the solar radiation in the wavelength between 0.4 to 0.7 microns,
It is sensitive to human eye and most important for plant life.

 Net Solar Radiation: It is the difference between the extra terrestrial radiation received on
outer boundaries of atmosphere and actual radiation absorbed by the earth.
 Net radiation: It is the balance of the energy after gain of solar radiation and loss of long
wave terrestrial radiation flux. The net radiation responses the amount of energy, which is
used for various kind of activities of biosphere.
 Albedo: The proportion of incident energy that is reflected or the reflection coefficient of
incoming solar radiation . A high albedo indicates that much of the incident solar
radiation is reflected than absorbed.
 Atmospheric Window: This refers to the outgoing long wave radiation from earth in the
wave length of 8.5 to 11 microns, which goes into space without absorption.
 Radiant Flux : The amount of radiant energy emitted , received or transmitted across a
particular area per unit time.

21
 03. Electromagnetic Spectrum

Electromagnetic Radiation

22
Electromagnetic Spectrum:

Light forms a small part of a large family of electromagnetic waves. We know how light splits
into the colors of the rainbow. The scientific term for this is a spectrum. You can see that the
colours run into each other. There are no distinct boundaries.
The properties of electromagnetic waves are related to their length and frequencies. When
arranged according to length, they form a continuous arrangement, known a Electromagnetic
Spectrum. Here is a picture that sums up the electromagnetic spectrum

Figure 03. EMR Spectrum

23
 Physiological Response to plant to different band of Incident Radiation:

04. Light Interception and Crop Growth

Photosynthetically Active Radiation (PAR) is the amount of light available for photosynthesis,
which is light in the 400 to 700 nanometer wavelength range. PAR changes seasonally and varies
depending on the latitude and time of day.Levels are greatest during the summer at mid-day.
Factors that reduce the amount of PAR available to plants include anything that reduces sunlight,
such as cloud cover, shading by trees, and buildings. Air pollution also affects PAR by filtering
out the amount of sunlight that can reach plants. Usually measured in Einsteins (Einstein = 6.02 x
1023 photons or one mole of photons). At night, PAR is zero. During mid-day in the summer,
PAR often reaches 2,000 to 3,000 millimoles per square meter.
How is Photosynthetically Active Radiation measured?
Photosynthetically Active Radiation (PAR) is reported as millimoles of light energy per square
meter. Photosynthetically Active Radiation (PAR) is measured by a silicon photovoltaic detector.

24
This detector measures light in the 400 nanometer to 700 nanometer range. Some PAR sensors
measure the PPFD of photosynthetically active radiation.

PAR Spectral Characteristics of the Crop:

Spectral characteristics explains the reflectance , transmittance, absorption of surface under

different wavelength . spectral properties of a vegetation or canopy or it’s components like leaves,
stems, ear head in the wavelength visible (0.4 -0.7µm) , NIR(0.76-1.0 µm ),mid IRC (2.0-
4.0 µm) can be measured with spectral radiometer (LICOR-1800-10) with cosine correction.
Diurnal Pattern and Seasonal variation : PAR, albedo increases from PI (panicle initiation) to
flag leaf as LAI and chlorophyll concentration increases further till boot stage as LAI becomes
highest .PAR albedo at boot to 50% flowering. Green reflectance decreases with senescence.
Moisture stress increases blue and red reflectance.
Changes in Spectral Composition in Plant Canopy : The spectral composition of the radiation
after transmission changes. This may be partly due to the factor that most of the visible (about
80% in maize ) is absorbed by leaf however the leaf allows the transmission of IR radiation. Kyle
(1971) found that 30-40% IR transmitted to the ground while only 5-10 % of visible part
transmitted to the ground in corn crop.
Light Distribution in canopy: There is a exponential relationship between light intensity and
height in the canopy which is know as Beer’s Law. In double plant density of a crop, less
incident radiations travel to the ground as compare to single planting. In closer spacing also, less
penetration of radiation takes place.

Let F be the average cumulative leaf area index, F is zero at the top of the canopy and is
maximum at ground level. PAR (Q) in the horizontal plane immediately above the top of the
canopy is defined by Q0. At any level F within the canopy, the rate of change of Q is
dQ/dF = -kQF
where dimensionless parameter k represents the fraction of incident photons absorbed per unit
area and is referred to as foliar absorption coefficient or extinction coefficient

After integration, Q at level F (Q F) is QF=Q0 e-kF which upon taking logarithms and
rearranging becomes
25
 kF= ln (Q0/QF)

The foliar absorption coefficient ranges from 0.3 to 1.3 for the majority of leaf canopies. In
canopies where the leaves are nearly vertical e.g many grasses, light penetrating in the lower
layers, k often low, typically 0.4. For such a canopy the cumulative F needed to absorb 95% of
PAR incident at the top of the canopy may be determined using the above equation. Then
F= ln (Q0/QF)/k
= ln [Q0/(0.05Q0]/0.4
= 7.5
The value of k becomes 1 for a high foliar absorption coefficient having horizontal leaves with

high chlorophyll levels e.g 0.5 g chlorophyll/m2 which can be found in crop plants such as
potato, soybean, sunflower etc. Canopies with most leaves in horizontal plane are termed
planophile, whereas canopies in which leaves are close to vertical are termed erectophile.
At solar noon, vertical leaves absorb less Q per unit leaf area than do horizontal leaves. This
accounts for low values of k for grasses, because their leaves are generally erect. Moreover,
leaves tend to be vertical near the top of certain plants e.g sugar beat, pineapple having more
horizontal leaves towards ground. This canopy architecture reduces the foliar absorption
coefficient of the upper layer leaves. In fact, optimal Q utilization for canopy photosynthesis
generally occurs when the incident Q is distributed as uniformly as possible over the leaves,
because the fraction of the leaves that are exposed to Q levels above light saturation or below
light compensation is then minimized.

26
 05. Duration of Light or Photoperiodism

Photosynthesis : Photosynthesis is the process by which plant convert light energy into
chemical energy in the form of reducing power as NADPH and ATP . This reducing power used to
fix Carbon dioxide as carbohydrates . In oxygenic photosynthetic organisms, inducing higher
plants, the source of reducing equivalents is H2O , releasing O2 as a by product.

27
 Figure 04. Process of Photosynthesis
Duration of Light : Duration of light (photoperiod ) influence time of flowering of many crops .
Response of plants to day light period is known as photoperiodism. Based on flowering
behaviour to photoperiod , plants are grouped in to four broad type –

No. Plant Type Characteristics

01 Short Day Photoperiod < 10 hr flowering does not occur if the day length is
Strict Short day above or below a critical value as the case may be short –day
Short Day /long day plants requiring an induction period of ling days.
02 Long day Photoperiod < 14 hr, effects of photoperiod additive . long day
Facultative plants will flower in a subsequently unfavorable period , long day
Long day/Short day plant requiring short day induction
03 Intermediate day Photoperiod of 12-14 hours with inhibition of reproduction either
Day Neutral below or above these levels unaffected by variation in day.
Importance of intensity of light : light intensity measures it’s quantity or brightness.
 Light intensity is indispensable for photosynthesis
 It controls leaf growth
 It influence sturdiness of stem and duration of vegetation growth specially of bulbous and
tuberous crops.
 Nodulation of legumes
 Light intensity yield attributes and yield.

28
Day Degree Days or Heat Units

 Heat Unit = days or hours of accumulated temperature above some threshold (but below
max. limits)
 Measured in degree-days or degree hours
 Lower temperature is called the threshold or base temperature
Heat Unit Calculation :
d
DD = ∑ Tmean –
Tbase Minimum temp for Plant

i=1
Sum over no. of Degree above base temp
days

29
06. Radiation use efficiency of crops

Crop growth results from photosynthesis and is subject to modification by both abiotic and
biotic factors. Early in the growing season the rate of dry matter production by a crop is
proportional to the amount of radiation intercepted, a function of leaf area index. Crop
growth depends on the quantity of incident light (Q0), the proportion of that light
intercepted (Qt) by the photosynthesizing organs of the plant, its efficiency of conversion
of light into dry matter (e) and respiratory losses. The amount of light intercepted at any
height z within the canopy is obtained by the difference between the incident radiation and
that transmitted and may be calculated from the equation

Qi = Qt/Q0 = e-kL
where Qi is the radiation transmitted at a height z in the canopy and L is the leaf area
index between the points of measurement of Q0 and Qt. Efficiency of conversion of light
into dry matter is also termed as radiation use efficiency or crop growth efficiency.
Crop growth efficiency tends to be higher in crops grown with adequate light, water and
nutrients and disease and pest free environment. For most field crops e tends to be high
during the vegetative phase when the photosynthetic rate is high and decreases when the
area decreases because of age and physiological maturity. The values of e is often
calculated as the slope of the best fit line on a plot of cumulative dry matter against
cumulative intercepted radiation or absorbed PAR. The values of e for a range of species
and environments are given in the following Table. The crop growth efficiency e varies
with several factors that affect cro growth such as air and soil temperatures, radiation
levels, soil moisture levels, nutrition etc.,.

30
 Shoot dry matter production per unit of intercepted PAR

Crop Site e (g/MJ)

Maize(Zea mays) Davis, CA, USA 3.0
Sorghum (S. bicolor) Pearl ICRISAT, India 2.4
millet (P. glaucum) Pearl ICRISAT, India 3.0
millet (P. glaucum) Niamey, Niger 2.4
Groundnut (A. hypogen) ICRISAT, India 0.8
Pegionpea (Ciamus cajan) Trinidad, West indies 1.2
Chickpea (Cicer arietinum) ICRISAT, India 1.36
Chickpea (Cicer arietinum) ICRISAT, Inida 1.24

07. Water Use Efficiency in Agriculture

The Food and Agriculture Organization (FAO) predicts a net expansion of irrigated
land of some 45 million hectares in 93 developing countries (for a total of 242 million
hectares in
2030) and project that agricultural water withdrawals will increase by approximately 14%
during
2000-2030 to meet food demand. of the water used to grow crops. Rainfed non-
irrigated agriculture accounts for some 60% of production in
developing countries. Although irrigation provides only 10% of agricultural
water use and covers just around 20% of the cropland, it can vastly increase crop yields,
improve food security and contribute 40% of total food production since the productivity of
irrigated land is three times higher than that of rainfed land.

31
 Figure 05. Water needed for food production

Competition among different sectors for scarce water resources and increasing public concern
on water quality for human, animal and industrial consumption and recreational activities
have
focused more attention on water management in agriculture. As water resources shrink
and
competition from other sectors grows, agriculture faces a dual challenge: to produce more
food with less water and to prevent the deterioration of water quality through contamination
with soil runoff, nutrients and agrochemicals. Current response measures, including
policies and regulations, consist of a combination of ways to ensure
adequate and more equitable
allocation of water for different sectors. Measures include improving water use
efficiency, pricing policies and privatization. Similarly there is an emphasis on integrated
water resources management, which takes into account all the potential stakeholders
in the planning, development and management.
Water use efficiency (WUE) is a broad concept that can be defined in many ways. For
farmers and land managers, WUE is the yield of harvested crop product achieved from
the water available to the crop through rainfall, irrigation and the contribution of soil
water storage. Improving WUE in agriculture will require an increase in crop water
productivity (an increase in
marketable crop yield per unit of water removed by plant) and a reduction in water losses
from
32
the plant rooting zone, a critical zone where adequate storage of moisture and
nutrients are required for optimizing crop production.
The amount of water required for food production depends on the agricultural
commodities produced. Improving WUE by 40% on rainfed and irrigated lands
would be needed to counterbalance the need for additional withdrawals for irrigation
over the next 25 years from additional demand for food. However, this is a big
challenge for many countries. Increasing WUE is a paramount objective, particularly
in arid and semi-arid areas with erratic rainfall patterns. Under rainfed conditions,
soil water can be lost from the soil surface through evaporation (termed soil
evaporation) or through plant uptake and subsequently lost via openings on plant leaves
(termed plant transpiration). It can also be lost through runoff and deep
infiltration through the soil. Total amount of soil water losses associated with
both soil evaporation and plant transpiration is referred to as evapotranspiration.
Evapotranspiration, grain yield and water use efficiency of crops

Crop Evapotranspiration (mm) Grain yield Water-use efficiency

(kg/ha) (kg/ha/mm)

Pearl millet 362 2049 5.66

Greengram 310 672 217

Mustard 552 2557 4.64

(Source: CAZRI,
Jodhpur)

Many promising strategies for raising WUE are available. These include appropriate
integrated land-water management practices such as –
(i) Adequate soil fertility to remove nutrient constraints on crop production
for every drop of water available through either rainfall or irrigation,
33
 (ii) Efficient recycling of agricultural wastewater,
(iii) Soil-water conservation measures through crop residue
incorporation,

adequate land preparation for crop establishment and rainwater harvesting

(iv) Conservation tillage to increase water infiltration, reduce runoff and improve soil
moisture storage.
(v) Fertigation, which combines irrigation and fertilization, maximizes the
synergy between these two agricultural inputs increasing their efficiencies.
Overall, improving irrigation WUE
(vi) Novel irrigation technologies such as supplementary irrigation (some
irrigation inputs to supplement inadequate rainfall), deficit irrigation
(eliminating irrigation at times that have little impact on yield) and drip irrigation
(targeting irrigation water to plant rooting zones) can also minimize soil
evaporation thus making more water available for plant transpiration.
Related literature:
1. Monteith, J.L. and M.H. Unsworth. 1990. Principles of Environmental Physics. Edward
Arnold (Publishers) Ltd., London UK.
2. Rao, A.S. and Vamadevan, V.K. 1985. Insolation in monsoonal wet tropics and
transmission/reflection coefficients of leaves of some plantation crops. Mausam,
36(3):347-350.
3. Subramaniam, A.R. and Rao, A.S. 1985. Prediction of ET of some crops under semi-arid
and dry sub-humid climates of Maharashtra. Mausam, 36(1):67-70.
4. Rao, A.S. and Alexander, D.1988. Prediction of evapotranspiration and grain yield of
rice. Intl. J. Biometeorology, 32 (2):81-86.
5. Rao, A.S. 1989. Water requirements of young coconut palms in a humid tropical climate.
Irrigation Science, 10:245-249
6. Singh, R.S., Rao, A.S. and Ramakrishna, Y.S. 1991. Growth characteristics of mustard
crop in response to thermal environment under arid conditions. Mausam. 42(4): 409-410.
7. Rao, A.S., Ramakrishna, Y.S. and Singh, R.S. and Chopra, N.K.1993. Water and energy
use efficiencies of sorghum inter-cropped with Accacia nilotica. Annals of Arid Zone.
32(2): 99-101.
8. Rao, A.S., Singh, K.C. and Ross weight, J.R. 1996. Productivity of Cenchrus ciliaris in
relation to rainfall and fertilization. J. Range Management. 49(2):145-146.9. Singh, R.S.,
34
 Rao, A.S., Joshi, N.L. and Ramakrishna, Y.S.2000. Evapotranspiration, water and
radiation-utilization in mothbean under two moisture conditions. Annals of Arid Zone,
39(1):21-28.
10. Rao, A.S, Singh, R.S, Joshi, N.L. and Ramakrishna, Y.S.2000. Evapotranspiration, water
and radiation-utilization of clusterbean (Cyamopsis tetragonolaba). Indian Journal of
Agricultural Sciences, 70(3):149-153.
11. Rao, A.S and Singh, R.S.2003. Evapotranspiration, water-use efficiency and thermal time
requirements of greengram (Phaseolus radiatus). Indian Journal of Agricultural
Sciences, 73(1):18-22.
12. Rao, A.S and Singh, R.S.2004. Water and thermal characteristics of cowpea (Vigna
unguiculata L. Walp.). J. of Agrometeorology,6(1):39-46.
13. Rao, A.S. and Singh, R.S.2004. 2004. Evapotranspiration rates, growing degree days and
water-use efficiency of some arid legumes. J. Arid Legumes 1(2): 166-170.
14. Singh, R.S. and Rao, A.S. 2007.Water and heat-use efficiency of mustard (Brassica
juncea L. Czern. & Coss) and its yield response to evapotranspiration rates under arid
conditions. J. of Agrometeorology. 9(2): 236-241.
15. Rao, A.S. and Singh, R.S. 2007.Evapotranspiration rates and water-use efficiency of
Pearl millet (Pennisetum glaucum L cv. HHB 67) under arid climatic conditions. Indian
J. Agril. Sciences 77(12): 810-813
16. Rao, A.S. 2008. Micro-meteorological variations and Rain water-use efficiency of a
Silvi-pastoral system. J. of Agrometeorology 10(2):137-140
17. Rao, A.S. and Poonia, S. 2011. Climate change impact on crop water requirement in arid
Rajasthan. Journal of Agrometeorology, 13(1):17-24.
18. Rao, A.S. 2011. Growing degree days, evapotranspiration and water-use-efficiency of
chilli crop in arid environment. J. of Agrometeorology 13(2) 128-130..
19. 64 Rao. A.S. 2011. Evapotranspiration and Water-use efficiency of Sesame crop
(Sesamum indicum L. cv. RT-127). Annals of Arid Zone. Accepted.

35
 Crop Specific Water balance models-SPAW (Soil-Plant-Air-Water) Model

(A.S. RAO)
Division of Natural Resources Management, Central Research Institute for Dryland
Agriculture, Hyderabad-500059

Simulation of crop growth/forage production from drylands had several difficulties

due to temporal and spatial variability in rainfall, high evapotranspiration rates and
heterogeneity in land use or crop acreage. Because of our limited knowledge in understanding
the assimilation of carbon by various dryland crops, trees or grasses and proceeds to allocate
this material to other organs which will undergo respiration, by and large, we have to depend
on regression models or soil moisture simulation models for explaining the crop/forage yield.
Studies on finding quantitative relationships between growth of crop/tree/grasses in relation
to environmental variables will help to know in advance about crop yield/forage production
through continuous monitoring of environmental variables. Some of the earlier studies were
based on rainfall relating crop production using regression equations. Later studies involved
use of dynamic simulation models for simulation of soil moisture and estimation of
production.
Soil water and its role in agriculture:
The availability of moisture in the soil is a prerequisite for the survival and growth of plants.
The capacity of soil for storing available water can be expressed conveniently in terms of
maximum available soil moisture in the root zone. Soil acts as a store-house for moisture,
where moisture accumulates upto a certain value, which is called field capacity during
periods of excess precipitation. The water contained in the soil above field capacity is not
available to crops due to quick drainage to lower layers. The water contained below
permanent wilting point (WP) is also not available to crops as it is tightly held by the soil
particles. Thus, it is generally considered that the amount of water held in the soil, between
field capacity and permanent wilting point is water available for crop use. This is a widely
accepted view and is followed by most of the workers for irrigation purposes.
The rate of availability of moisture to crops within field capacity and permanent WP
and its effect on crop growth is an important factor in determining crop growth. Veihmeyer
and Hendrickson (1955) have made extensive field tests on perennial fruit crop and stated
that water is equally available for plant growth at all levels, between field capacity and
36
permanent WP. Their experiments show that favorable conditions of soil moisture extend
from the field capacity to about the permanent WP. This view is still being accepted and
adopted by agricultural scientists and in irrigation scheduling.
Soil moisture retention
The moisture content of a sample of soil is usually defined as the amount of water lost

when dried at 105OC, expressed either as the weight of water per unit weight of dry soil or as
the volume of water per unit volume of bilk soil. Although such information may not give a
clear indication of the availability of water for the plants, differences exist because the water
retention characteristics are generally different for different soils.
The forces that keep soil and water together are based on the attraction between the
individual molecules, both between water and soil molecules (adhesion) and among water
forces, while in the dry range absorption is the main factor. The factors influencing the
relations of plants and thus their growth and yield response may be grouped into the
following;
1. Soil-factor-Soil moisture content, texture, structure, density, salinity, minerate
processes, fertility, aeration, temperature and drainage
2. Plant factors-Type of crop, density and depth of rooting, rate of root growth,
aerodynamics, roughness of the crop, drought tolerance and varietal effects
3. Weather factors-Sunshine, temperature, humidity, wind and rainfall
4. Miscellaneous factors-Soil volume and plant spacing, soil fertility and crop water,
soil and agronomic practices management
Water stress and plant development:
Plant growth and development depends
1. Upon a continuous process of cell division
2. On the progressive initiation of tissue and organ primordial and
3. On the differentiation and expansion of the component cells.
Along with this is an inter-connected chain of metabolic events which involves the uptake
of nutrients from both soil and air, the synthesis of metabolites and structural materials and
also from the flow of substances within the plant body. Because all these plants processes
take place in the aqueous medium and water being a transporting agent as well as a reactant
in the majority of these processes, any shortfall in water uptake and dehydration results in
negative effects on most of the physiological processes.

37
 Stress affects those tissues most which are in rapid stages of development. Primordial
initiation and cell enlargement are inhibited by moisture stress.. After moisture stress is over,
developing tissues are rejuvenated and growth rates of many plants are more rapid than those
which remained unstressed. This is due to the continued but slow cell division as well as the
availability of nutrients released by the old tissues.
Soil moisture stress and grain yield:

There are three key stages when moisture stress affects the grain yield in cereals and
these are
1. Stage of floral initiation and inflorescence development
2. Stage of anthesis and fertilization
3. Grain filling stage
Soil-Plant-Air Model:
The SPAW model is a daily hydrologic budget model for agricultural fields. It also
includes a second routine for daily water budgets of ponds and wetlands, which utilizes the
field hydrology as the watershed. The field budget utilizes a one-dimensional vertical system
beginning above the plant canopy and proceeding downward into the soil profile a sufficient
depth to represent the complete root penetration and subsurface hydrologic processes. The
following schematics describe the field and pond hydrologic systems and each major
hydrologic process impacting water movement across the system boundaries and within the
systems.

38
 For drought and crop surveillance over the region, the SPAW (Soil-Plant-Air-Water)
was tested and modified for Indian region by Rao and Saxton (1995) and Rao et al., (2000).
They used the SPAW model for assessment of soil moisture and crop stress conditions under
pearl millet which explained 89% of pearl millet yields of Jodhpur district. The relationship
between pearl millet grain yield (kg/ha) and the water stress index (WSI) of SPAW model
was Y = -45.38 W
SI + 526.18 (r=0.9427; Figures 2&3).
5000
4500
4000
Pearl millet grain yield (kg/ha)

3500
3000
2500
2000
1500
1000
500
0

0 2 4 6 8 10 12 14
Water Stress Index (SPAW model)
Fig. 2. Relationships between Water Stress Index and Pearl millet
yield (Source: Rao and Saxton,
1995)
5000 Observed Estimated
4500
4000
Grain yield (kg ha-1)

3500
3000
2500
2000
1500
1000
50
0

39
1990 1991 1992 1993 1994 1995 1996 1990 1991 1992 1993 1994
1995 1996

Year
Irrigated Rainfed

Fig.3 Observed and estimated pearl millet grain yield using the Soil-Plant-Air Water Model
(Source: Rao et al., 2000)
References:

1. Rao, A.S., Joshi, N.L. and Saxton, K.E.2000.Monitoring of productivity and crop
water stress in pearl millet using the SPAW model. Annals of Arid Zone, 39(2):151-
161.
2. Rao, A.S. and Saxton, K.E. 1995. Analysis of soil water and water stress under pearl
millet in an Indian arid region using the SPAW model. J. Arid Environments,
29(2):155-167.

40
 Modeling the effect of weather or weather based indices on performance of crops

P. Vijaya Kumar
Senior Scientist (Agro met)
Central Research Institute for Dryland Agriculture,Hyderabad

Introduction:
Crop weather modeling refers to the techniques that are used to determine the effect of weather
on yield.Although influence of weather conditions on yield is well established, the quantitative
assessment is not always straight forward and simple.Inter-annual variability,in a long time
series yield data can be due to either or all of these components:Trend,Direct weather effect and
Indirect weather effects viz.,pest,diseases,weed competition etc.In developed weather
countries,trend due to improved technology and management accounts for 80% of variability
while in developing countries trend is substantially less or negligible.The quantitative
assessment of the effect of weather on crops is no doubt is a most important application of
agrometeorology in both developed and developing countries.The application of crop weather
modeling extends from the scale of farmers field to entire countries or continents.The aim of
developing crop weather models is to ensure better utilization of resources and hence a more
environmental friendly and sustainable agriculture.
Importance of weather:
Weather directly or indirectly influences the crops in their growth cycle. The growth,
development and productivity of crops are the resultant of many physical and physiological
processes, each of which are affected individually or jointly by weather parameters. Though
weather or climate is the least manageable natural resource, understanding of its interaction with
agricultural parameters was found to be a powerful tool to develop weather based management
strategies in agriculture that will enhance benefits from positive interactions and minimize the
losses from negative interactions (Virmani, 1994).
The principal weather parameters which affect crop growth and yield are: Precipitation
(amount and distribution), air temperature (Maximum and Minimum), moisture content of the
air (Relative humidity, SVPD), solar radiation or sunshine hours and wind speed.
Developments in crop-weather relationship studies
A great deal of research on crop-weather relationships in respect of important rainfed
crops was reported in India and elsewhere. Pioneering crop-weather relationship studies can be
traced back to Fisher (1924). In India, research on the crop-weather relationships was initiated
by Prof. L.A. Ramdas, considered the ‘Father of Agrometeorology in India’ in 1926. These

41
efforts led to the initiation of research on crop-weather relationship by IMD in 1932 and later by
ICAR in 1948 through a Coordinated Crop Weather Scheme. Later, some Agricultural
Universities and Agricultural Research Institutes started conducting crop-weather relationship
studies. Considerable work on crop-weather relationships was carried out over years by Sarkar
(1965) from IMD and Bhargava et al (1978) and Agarwal et al (1986) from IASRI. The multi-
location crop-weather relationships in various crops were geared up after launching of the All
India Coordinated Research Project on Agricultural Meteorology in 1983. A review on crop-
weather relationship studies in important crops in India and other countries till the late sixties
was made earlier by Venkataraman (1972) and on rice by Rao and Das (1971). A brief review
of crop-weather regression models in dryland agricultural research was also made later (Rao and
Rao, 1992).
Changes in concept of crop-weather relationship studies
Over the years, a lot of changes in the concepts of crop-weather relationships have been
evolved. The crop-weather relationship studies in earlier years were based on statistical
techniques like correlation, simple and multiple regression, step-wise regression, etc. It was
believed by Agrometeorologists working on dryland agriculture earlier that rainfall is main
factor for variation in yields of dryland crops. Crop yields were related with rainfall during
different stages of the crop growth to identify the critical stages of the crop. Experience and
logic prompted them to look for some more parameters other than rainfall for accurate prediction
of crop yields. Though total amount of seasonal rainfall showed some amount of relation with
final yield, it was not representing the actual water available for plant growth as it does not
account the losses through drainage, runoff and also the influence of the water holding capacity
of the soil. The proposition of potential Evapotranspiration concept simultaneously by Penman
(1948) and Thornthwaite (1948) and the introduction of water budgeting by Thornthwaite (1948)
and the modification of the same by Thornthwaite and Mather (1955) brought in an appropriate
independent variable, i.e., water use or Evapotranspiration for prediction of crop yields. Later,
de Wit (1958) developed an equation to relate dry matter yield (Y) to transpiration as:
Y = m T/Eo
Where T is transpiration in cm, Eo is average free water evaporation rate (cm/day) and m is a
crop factor.
The ratio of actual evapotranspiration to potential Evapotranspiration (AE/PE) known as
Index of Moisture Adequacy (IMA) has found its use, later, in crop-weather relationship studies.
The non-accountancy of crop factor in water balance models was corrected by introduction of
models by researchers like Frere and Popov (1979), Ritchie (1972), etc. The simple (FAO)

42
water balance model developed by Frere and Popov (1979) introduced an index called Water
Requirement Satisfaction Index (WRSI) for predicting crop yields.
Although water supply plays a dominant role in agriculture, other climatic factors also
influence the performance of crops and to understand the effect of more weather parameters,
multi-variate crop-weather relationships were developed.
All these statistical models developed with data from a given place though have higher
predictability suffer from location-specific bias. To overcome the site-specific problem,
concerted scientific effort for development of dynamic crop simulation models which are
generic in nature and are applicable universally was initiated across the globe.
Effects of different weather parameters:
Solar radiation:
Crop production is in fact exploitation of solar energy. Solar energy(solar radiation) is the
driving force and only source of energy for photosynthesis(Monteith 1973).It is one of the main
factors influencing biomass, yield and its quality. When water and nutrients, diseases and insects
are not limiting factors, crop growth is determined by the amount of solar radiation intercepted
and carbon dioxide assimilated. Three aspects of solar radiation are important for plant
processes: Intensity, duration (i.e., photoperiod or day length), and quality. Low intensity of
solar radiation during grain filling phase negatively influences grain yield of cereal crops.
The length of the day or photoperiod determines flowering and has a profound effect on the
content of soluble carbohydrates present. A majority of plants flower only when exposed to
certain specific photoperiods. It is on the basis of this response that the plants have been
classified as short day plants, long day plants and day neutral plants. When any other
environmental factors is not limiting, the longer duration of photoperiod increases
photosynthesis.
Temperature
Temperature is very important not only to plants but also to all the biological species
because of following factors:
ú Physical and chemical processes within the plants are governed by temperature.
ú The diffusion rate of gases and liquids in soil-plant-atmospheric system changes with
temperature.
Temperature affects crops by causing (i) variations in duration of phenological events or
crop development. (ii) variation in magnitude and time of occurrence of peak in biomass, (iii)
significant increase / decrease in growth rates, (iv) variation in growth pattern deviating from
sigmoidal curve and ultimately affecting grain yield or harvest index.
Combined influence of temperature and photo-period

43
 Though development of crops is mainly driven by temperature, some plant species
respond to photo-period or day length. The photo thermal effects on phenology in many crops
were reported. For all tropical and sub-tropical species, the warmest temperature combined with
shortest photo period hastened flowering and fruit maturity (Keating et al, 1998). However, all
temperate species both flowered and matured sooner at the warmest temperature combined with
longest photo period.
Cardinal points or temperature thresholds (°C) for some major crops are listed in Table-
3.
Table-3.Temperature thresholds (°C) during growing season for some major crops
Crop Minimum Optimum Maximum
Sugarcane 13 35-37 >40
Wheat 0 17-23 30-35
Rice 7-12 25-30 35-38
Maize 8-13 25-30 32-37
Potato 5-10 15-20 25
Sorghum 8-10 32-35 40
(Rötter and van de Geijn, 1999)

Low temperature affect

Low temperature affects several aspects of crop growth, viz., survival, cell division,
photosynthesis, water transport, growth and finally yield.

High temperature affects

High temperature adversely affects mineral nutrition, shoot growth and pollen
development resulting in low yield. Adverse effects of high temperature during critical growth
stages of some major crops were mentioned in Table-4.
Table-4. High temperature effects on key development stages of five major arable crops

Crop Effect
Wheat T > 30°C for > 8 hrs can reverse vernalisation
Rice T > 35°C for > 1 hr at anthesis causes spike let sterility
Maize T > 36 °C reduces pollen viability
Potato T > 20°C reduces tuber initiation and bulking
Cotton T> 40ºC for more than 6 hours causes bolls to abort
Source: Acock and Acock (1993)

44
Rainfall or water use:
Rainfall is an important parameter in agriculture. All plants need water to survive and rainfall is
the main source providing water to plants. While normal rainfall is vital to healthy plants, too
much or too little rainfall can be harmful to crops. Plants need varying amounts of rainfall to
survive. Desert plants require small amounts of water while tropical plants need much higher
rainfall. Water is an essential component in the process of photosynthesis. The movement of
water out of the plant stomata, known as transpiration is an inevitable consequence of
assimilation of carbon dioxide. As transpiration or water use and photosynthesis are inter related,
a linear relation between crop yield and seasonal transpiration was established by Hanks (1974)
as follows:
Y=m*(T/E0)
Where Y=Yield, T=Seasonal transpiration and E0 is average seasonal free water evaporation and
m is a crop factor. This equation gave very good fit for several crops grown in different years in
different locations.
Agroclimatic indices:
Some important agroclimatic indices formed by combining two or more weather parameters, are
given below:.
1.Photo thermal units (PTU):
It is the product of Degree days (0C) and Day length (hours) and expressed in units 0C hrs
2. Helio thermal units(HTU):
It is the product of Degree day (0C) and actual bright sunshine hours and it is expressed in units
0
C hrs
3.Thermal Interception Rate(TIR):
TIR=PARI/n(Tm-Ta)
Where PARI=Photosynthetically active radiation intercepted by the crop, n=No. of plants/m2,
Tm=Mean daily temp and Ta is base temp
4.Water Requirement Satisfaction Index(WRSI):
The WRSI is an indicator of crop performance based on the availability of water to the crop
during growing season. WRSI for a growing season is calculated as the ratio of seasonal actual
evapotranspiration(AET) to the seasonal crop water requirement (WR).The water requirement of
the crop at a given time of the growing season is calculated by multiflying the
reference(potential) evapotranspiration with a crop coefficient,whose values are published by
FAO(FAO,1998).
5. Moisture Availability Index(MAI):

45
It is defined as the ratio of rainfall at 75% probability (PD) and Potential
Evapotranspiration(PET).
MAI=PD/PET
6.Moisture Adequacy Index(MAI):
It is defined as the ratio of actual evapotranspiration and potential evapotranspiration and can be
written as MAI=AET/PET
Methods to evaluate Crop-weather Relationship
The three commonly used approaches in crop weather modeling studies are are:
§ Correlation techniques
§ Crop weather analysis model
§ Crop growth simulation models
Correlation analysis provides a measure of the degree of association between variables
Regression analysis describes the effect of one or more variables (independent variables) on a
single variable (dependent variable)
Regression and correlation procedures can be classified according to the number of variables
involved
§ Simple (If only 2 variables, one independent and another dependent)
§ Multiple (If more than 2 variables)
The procedure is termed linear, if underlying relationship is linear or non-linear, if otherwise
Regression equations are broadly of four types
Ø Simple linear regression
Ø Multiple regression
Ø Simple non-linear regression
Ø Multiple non-linear regression
Simple and multiple regressions widely used for crop weather relationship studies can be written
as
Y=a+b*X
Y is the dependent variable, example-Yield
X is the independent variable, example-Rainfall, temperature etc
Y=a+b1*X1+b2*X2+……..+bk*Xk
k =Number of independent variables
R2 is coefficient of determination
There must be enough observations to make n greater than (k+1)

46
Multi Collinearity:
Multi collinearity in regression equations occurs when predictor variables (independent
variables) in the regression model are more highly correlated with other predictor variables than
with the dependent variable.
It commonly occurs when a large number of independent variables are incorporated in a
regression model.
Searching for best regression
There are two ways in which relationship between dependent variable and k independent
variables be specified
• Based on accepted biological concepts, secondary data, past experience etc.
• Based on the data collected
Four procedures commonly used for specification of appropriate relationship between X and Y
are
1. Scatter Diagram(for simple regression)
2. Analysis of variance technique(not relevant for CWR studies)
3. Test of significance technique(for elimination of unnecessary variables)
4. Step-wise regression technique(for identifying the sequence of importance of each
variable)
Standardizing variables
The following standardization procedures help to reduce experimental error and biases
§ Yields from different varieties to be adjusted to a “standard” “base” variety
§ Weather variables are to be measured within specific stages of plant development rather
than within specified weeks or months
§ Yields are to be culled to remove those reduced by disease,hail,pests and other factors
§ Reduction in experimental error can be accomplished through use of simulated
evapotranspiration amounts rather than precipitation, to measure effects of droughts
Models relating yield with different weather parameters or indices:
Yield and water use:
Doorenbos et al.(1979) proposed the following equation relating yield and water use.
(1-Ya/Ym ) = ky(1-ETa/ETm)
Where Ya is actual yield, Ym is maximum potential yield, ky is a yield response factor, Eta is
actual crop evapotranspiration and ETm is maximum or potential evapotranspiration.
Yield response to temperature:

47
The response of grain yield of wheat to minimum temperature was different under different
phenological stages. The linear regression of yield with minimum temperature was positive
during crown root initiation stage while it was negative during anthesis stage(Fig- 1 and 2 )
Relationship between wheat yield and minimum temperature at
the stage of anthesis at New Delhi.
70
y = -5.6x + 112
R2 = 0.66
65

60
Yield(q/ha)

55

50

45

40
8 8.5 9 9.5 10 10.5 11 11.5 12
0
Minimum Temperature( C)

R elationship between yield an d minimum temperatu re d uring

C RI stag e at Ludh iana

55

50

45
Yield(q/ha)

40

35
y = 2. 9x + 24.05
30 R2 = 0.44**
25

20
3 4 5 6 7 8 9 10
Mini mum tem perature( 0C)

Figure 1 & 2
Yield and rainfall:
Yield usually shows curvilinear relationship with rainfall in most of the crops. The yield
response to rainfall varies with the phenological stage of the crop. In most of the field crops
rainfall during reproductive stage is critical for the grain yield achieved. The yield and weather

48
relationship of soybean, illustrated below is showing significant curvilinear relationship with
rainfall during reproductive period (Fig 3 ).

Fig. 3. Relationship between rainfall during reproductive period and yield of soybean
at Jabalpur
Yield and canopy temperature:
Yield showed significant inverse linear relationship with canopy temperature in chickpea at
Jabalpur(Fig 4).The stress degree day ,which was worked out as the difference between canopy
and air temperature also showed highly significant inverse relationship with yield of pearl millet
at Solapur (Fig 5)

Fig 4. Relationship of canopy temperature and yield in chickpea at Jabalpur

49
Fig 5. Relation between stress degree days and grain yield of pearlmillet at Solapur

Modeling the effect of cold and heat waves:

Modeling the effect heat and cold waves during critical stages of wheat were evaluated by
relating yield with number of days with above and below normal temperature during the critical
periods. The results showed that the cold wave conditions during jointing stage at Ludhiana(Fig
6 ) and hot wave conditions during flag leaf stage at Hisar (Fig 7) adversely affected the yield.

Fig 6. Relationship between wheat yield and number of days with less than base temperature
during Jointing stage at Ludhiana

50
Fig 7. Relationship between wheat yield and number of days with hyper optimal temperature
during flag leaf to milking stage at Ludhiana
Effect of de-trending the yield on predictability:
The grain yield of wheat showed significant positive relationship with diurnal temperature
range(Fig 8 ). The coefficient of determination (R2) of of this relationship improved very much
after de-trending of yield, as revealed by the Figure 9.

Figure 8 : Relation between yield and Diurnal Temperature Range (DTR)

51
Figure 9: Relation between wheat yield (detrend) and Diurnal Temperature Range (DTR)
Conclusion
Weather is one of the important natural resources influencing crop growth and productivity
across different agro-ecological regions of the country. Crop weather models help in assessing
growth and yield of crops at different crop stages and also in quantifying the stress-yield
relations in respect of moisture, thermal and radiation regimes. Though much work has been
done in developing crop weather models and generating knowledge on the physical processes
influencing plant growth and productivity, there is still a need to generate similar information in
future for the upcoming promising genotypes and new crops that replace existing crops or
cropping systems due to economical considerations and technological innovations.
References
Acock, B. and Acock, M.C.1993.Modelling approaches for predicting crop ecosystem responses
to climate change, In: International crop science, vol.1. pp.299-306, Crop Science society of
America, Madison,Wisconsin,USA.
Agarwal, R., Jain, R.C. and Jha, M.P. 1986. Models for studying rice crop-weather relationship.
Mausam 37(1): 67-70.
Bhargava, P.N., Aneja, K.G. and Ghai, R.K. 1978. Influence of moist days on crop production.
Mausam, 29(2): 111-118.
de Wit, C.T. 1958. Transpiration and crop yields. Versl. Landbouwkd. Onderz, No. 64.6,
Pudoc, Wageningen, The Netherlands, 88 p.
Doorenbos, J.and Kasam, A.H.1979.Yield response to water.FAO Irrigation and Drainage
Paper no.33.Rome, Italy, FAO.
Fisher, R.A. 1924. The influence of rainfall on the yield of wheat at Rothamsted. Roy. Soc.
(London), Phil. Trans. Ser. B., 213: 89-142, Illus.
Frere, M. and Popov, G.F. 1979. Agrometeorological crop monitoring and forecasting. FAO
Plant Production and Protection Paper 17. Food and Agricultural Organization, Rome.
Hanks, R.J. 1974. Model for predicting plant growth as influenced by Evapotranspiration
and soil water. Agronomy Journal 66: 660-665.

Keating ,J.D.H.,Qi,A.,Wheeler,T.R.,Ellis,R.H.and Summerfield,R.J.1998.Effects of temperature

and photoperiod on phenology as a guide to the selection of annual legume cover and green
manure crops for hillside farming systems. Field Crops Research 57(2):139-152.
Monteith, J.L. 1973.Principles of environmental physics.London, UK: Longman.
Penman, H.C. 1948. Natural evaporation from open water, bare soil and grass. Proceedings of
Royal Society of London, Series A. 193: 120-145.

52
Rao, K.N. and Das, J.C. 1971. Weather and crop yields. Rice (Survey). Pre Publication
Science Report 137, Indian Meteorology Department.
Rao, D.G. and Rao, U.M.B. 1992. Simulation and regression models in dryland agricultural
research. In: Dryland Agriculture in India – State-of-art-of-research in India, pp. 57-77.
Somani, L.L., Vittal, K.P.R. and Venkateswarlu, B. (Eds), Scientific Publishers, Jodhpur.
Ritchie, J.T. 1972. Model for predicting evaporation from a row crop with incomplete cover.
Water Resources Research 8: 1204-1213.
Rötter, R. and Van de Geijn, S.C. 1999.Climate change effects on plant growth, crop yield and
livestock, Climatic Change 43:651-681.
Sarkar, R.P. 1965. A curvilinear study of yield with reference to weather – Sugarcane. Indian
Journal of Meteorology and Geophysics 16:103-110.
Thornthwaite, C.W. 1948. An approach towards a rational classification of climate.
Geographical Review 38: 85-94
Venkataraman, S. 1972. Weather relations of crops. In: Crops and Weather, pp. 302-
457.Venkataraman, S. and Krishnan, A. (eds.), Publication of Indian Council of Agricultural
Research, New Delhi.
Virmani, S.M. 1994. Climatic resource characterization in stressed tropical environment:
Constraints and opportunities for sustainable agriculture. In: Stressed Ecosystems and
Sustainable Agriculture, pp. 149-160. Virmani, S.M., Katyal, J.C., Eswaran, H. and Abrol, I.P.
(eds.), Oxford & IBH Publishing Co. Pvt. Ltd., New Delhi.

53
 Water Balance by Thornthwaite & Mather
And FAO Methods
GGSN Rao
Retd. Project Coordinator (Ag. Met)
CRIDA, Hyderabad – 500 059
Introduction
Estimations of water balance components, viz., actual evapotranspiration(AET), water
surplus (WS)and water deficit(WD) over a region are extremely important in the field of
Hydrology, Agriculture, Ecology, etc. in identifying the regions suitable for different crops.
Water balance computation is one of the important tools in applied climatology that has
innumerable applications, viz., climatic classification, agricultural crop planning, water
harvesting potentials, and in climate change studies, Thornthwaite (1948) developed the
procedure to compute the water balance by considering the monthly rainfall and potential
evapotranspiration (PET) a new terminology introduce by him.

The procedure was slightly modified in 1955 by Thorthwaite and Mather by introducing
the soil moisture retention tables for different types / depth of soils. Due to its wide
applicability, the water balance computational procedures are in great demand. FAO (1979) also
brought out a monograph to compute crop-specific water balance by considering the weekly
rainfall and the corresponding crop water requirements instead of potential evapotranspiration.
The required crop water requirements are computed by multiplying the PET with crop
coefficient (Kc) values.
All water balance models attempt to determine what happens to water that is applied to
or fall on a given area. The water balance of a system is the difference between the inputs to the
system and the flow of water out of the system or storage of water within the system. The inputs
are generally precipitation or in some cases, irrigation although, depending upon the boundaries
of the system, it could also include water brought into the system through run off. An equation
describing the water balance may be written as:

P + I = R + D + ET+ ∆SM
Where,
P = Precipitation
I = Irrigation
R = Surface runoff
D = Deep drainage

54
 ET = Evapotranspiration
∆SM= Change in soil moisture storage

A. Calculation of Water Balance by FAO Method (1979)

In order to compute the weekly water balance according to FAO method, it is necessary to
have the following information at a place.

ú Weekly rainfall (mm)

ú Weekly potential evapotranspiration (mm)
ú Weekly crop coefficients
ú Available water holding capacity of the soil (mm)
Procedure for calculating weekly water balance

The different steps involved in the computation of the cumulative weekly water balance for
the specific crop are detailed below:

Step-1:

ú Enter weekly rainfall (PPT)

Step-2:

ú Enter weekly potential evapotranspiration in the same units as rainfall (PET)

Step-3:

ú Enter crop coefficients (KCR)

Step-4:

ú Compute water requirements of the crop (WR)

ú Water requirements of the crops are worked out by multiplying the potential
evapotranspiration of the week by the crop coefficient of that week and also calculate
total water requirement of the crop for the season by adding the successive water
requirements week by week.
Step-5:

ú Compute (PPT-WR) for different weeks

ú The difference between actual rainfall and water requirement expresses whether the
rainfall is adequate to meet the demand of the crop, without, however, taking into
account the water stored in the soil The negative value of the (PPT-WR) indicates, the
water demand of the crop is not met by rainfall. In this case, the crop takes the water
from the soil if water is available in the soil. The positive value of (PPT-WR) indicates,

55
 the water supply is more than the water requirement of the crop. The excess water goes
to recharge the soil upto its capacity.
Step-6:

ú Surplus (SPL)
ú Surplus refers to quantity of water whenever the soil moisture reserve exceeds water-
holding capacity of the soil
Surplus = (PPT-WR) + Previous week’s soil moisture reserve – Available water
holding capacity

Note: Whenever surplus occurs, the value of water holding capacity itself is the soil
moisture reserve

Step-7:

ú Deficit (DEF)
ú When the difference between (PPT-WR) is negative, it indicates deficit. This refers to
the short falls in the water requirement after taking soil moisture reserve into
consideration.
ú Deficit = (PPT-WR) (without sign) – Previous week’s soil moisture Reserve
Step-8:

ú Water requirement satisfaction index (WRSI)

ú It is assumed that sowing takes place when at least 75 mm of rainfall has been accumulated.
So index is assumed to be 100 at the beginning of crop growing season. This index will
remain at 100 for the successive weeks until either a surplus of more than 100 mm or a
deficit occurs. If a surplus of more than 100 mm occurs during a week and the rainfall
during the same week has fallen in less than 3 days, the index is reduced by 2.1 units during
this week and remaining at the level until a further stress period occurs. If the deficit occurs,
the index is calculated by subtracting the percentage reduction during the week from the
preceding week’s index. The percentage reduction during the week is calculated as the ratio
of deficit during the same week and total water requirement of the crop expressed as
percentage. The calculation is pursued to the end of the growing season taking into account
the fact that the index starts in the first week at 100 and thereafter can only remain at 100 or
goes down. The index at the end of the growing season will reflect the cumulative stress
endured by the crop through excesses and deficits of water and will usually be closely linked
with the final yield of the crop, unless some other harmful factors (eg. pests and diseases,
strong winds etc.) have predominant affects.

56
B. Calculation of Weekly Water Balance by Thronthwaite and Mather’s method (1955):
To compute the weekly water balance according to Thronthwaite and Mather’s method
(1955), following information at a place is required.
ú Weekly rainfall in mm
ú Weekly potential evapotranspiration in mm
ú Available water holding capacity of the soil in mm
Procedure to calculate weekly water balance:
The different steps involved in the calculation of weekly water balance are given below:
Step-1: Enter weekly rainfall (P)
Step-2: Enter weekly potential evapotranspiration (PET)
Step-3: Enter available water holding capacity of the soil (AWC)
Step-4: Compute (P-PET) for different weeks
The difference between actual rainfall and potential evapotranspiration expresses whether the
rainfall is adequate to meet the atmospheric demand, without, however, taking into account the
soil moisture stored in the soil. The negative value indicates that the atmospheric demand is not
met by the rainfall. In this case, soil moisture is taken from the soil, if it is available in the soil.
The positive value indicates that the water supply (rainfall) is more than the atmospheric
demand. The excess water goes to recharge the soil up to its capacity.

Step-5: Surplus (SPL)

Surplus refers to quantity of water wherever the soil moisture reserve

exceeds water holding capacity of the soil
SMi = SMi-1 + (Pi-PETi)
If SM i is greater than AWC, surplus occurs
SPLi = SMi – AWC
SMi = AWC
AETi = PETi
Step-6: Deficit (D)
When the difference between (P-PET) is negative, it indicates deficit. This refers
to short falls in the atmospheric demand.
Compute accumulated potential water loss (APWL)
APWLi = å ( P - PET )
i i

æ - APWL ö
SMi = AWC * exp( ç ÷
è AWC ø

57
 AETi = Pi + ∆SMi Di = PETi - AETi

Where, ∆SM is change in the soil moisture storage

∆SMi = SMi-1 - SMi
AET is actual evapotranspiration
SM is the soil moisture reserve
References

Frere, M., and Popov, G.F. 1979. Agrometeorological crop monitoring and forecasting, FAO
Plant Production and Protection paper No. 17, Food and Agriculture Organization.
Rome, Italy, 64 pages.

Thronthwaite, C.W. (1948). An approach toward a rational classification of climate, Georg.

Rev., vol.38, No1, PP 55-94
Thronthwaite, C.W. and Mather, J.R. 1955. Water Balance: Publications in Climatology No. 8,
Drexel Institute of Technology, Centerton, N.J. 104 pp.

58
 Computation of water balance in crop growth models (DSSAT)
B. Bapuji Rao, Principal Scientist (Agromet)

The widely evaluated CROPGRO and CERES models use the one dimensional
tipping bucket soil water balance of Ritchie (1985, 1998) and furt her mo dified by Porter et
al. (2004). Soil water balance processes envisaged in these models include infiltration of
rainfall and irrigation, runoff, soil evaporation, crop transpiration, distribution of root water
uptake from soil layers, and drainage of water through the profile and below the root zone. The
soil is divided into a number of computational layers, up to a maximum of 20. Water content in
each layer varies between the lower limit of plant extractable soil water [LL(J)], the drained
upper limit [DUL(J)], and the saturated soil water content [SAT(J)]. If water content of a
given layer is above the DUL, then water is drained to the next layer with the “tipping bucket
approach, using a profile wide drain- age coefficient (SWCON). If available, saturated
hydraulic conductivity (K sat) for water flow of each specific soil layer can be entered to control

vertical drainage from one layer to the next. This allows the soil to retain water above the DUL
for layers that have sufficiently low K sat for water flow, and in this case, soil layers may

become saturated for sufficient time to cause root death, reduced root water uptake, Water
between SAT and DUL is available for root uptake subject to the anoxia induced problem that
is triggered when air filled pore space falls below 2% of total volumetric pore space. Infiltration
and runoff of rainfall and applied irrigation water depends on the Soil Conservation Service
runoff curve number.
Reference crop evapotranspiration

Three options are provided in these models for computing climatic potential
evapotranspiration (PET, equivalent to ETo): (i) Priestley Taylor method (Priestley and Taylor,
1972) also described by Ritchie (1985), (ii) FAO Penman 24 method described by Jensen et al.
(1990), and (iii) the FAO 56 described by Allen et al. (1998). The default PET option is the
Priestley Taylor option, primarily because it is the less demanding of weather data (it is the
only one that does not require daily windspeed or dewpoint temperature as input). The other
methods additionally require wind speed and humidity (actually dewpoint temperature).

Through the model has ETPHOT, which is a more mechanistic model, but this
component was not discussed here.
Partitioning of PET into transpiration and Soil Evaporation

The DSSAT crop models partition the PET to potential plant transpiration (EPO) and
potential soil evaporation (ESO), following the Ritchie (1972, 1985) approach, which considers
59
the portion of net radiation that reaches the soil and that can be spent as latent energy to
evaporate water from the soil surface if the soil is wet. The climatic EPO is computed by
multiplying the PET of Options 1, 2, or 3 by the exponential function of LAI shown in Eq. [1]
using a KEP that is smaller than the extinction coefficient for photosynthetically active
radiation. Energy not absorbed by the crop is transmitted to the soil surface (Eq. [2]) and is
available to drive ESO.

EPO = PET x [1.0 — exp(—KEP x LAI)] [1]

ESO = PET x exp(—0.40 x LAI) [2]

The actual soil evaporation (ES) and plant transpiration (EP) depend on the avail- ability
of water to meet these potential rates. The current DSSAT models compute the ES following the
two stage soil evaporation method of Ritchie (1985): (i) the energy limited stage (Stage 1); and
(ii) the falling rate stage (Stage 2) that begins after the first stage loss has been met, after which
ES declines as the square root of time. The Stage 1 soil water evaporation limit is a value
defined for each soil pro- file; once this amount is achieved, the evaporation follows the falling
rate stage. In addition, ES is allowed to have access only to the soil water in the top 5 cm of soil
(for all soils) that limits the function. If the potential ES O energy is not used for ES during

Stage 2, it is presumed absorbed by the canopy as follows:

IF EPO + ES < PET, THEN EPO = PET - ES [3]

Root Water Uptake and Water Stress Factors

Root water uptake must be computed before the actual canopy transpiration (EP) is
computed. Potential root water uptake per soil layer, RWU(L), is a function of root length
density (RLD) and soil water content within each soil layer using a simplified computation of
radial flow to roots (Ritchie, 1985) as shown in Eq. [4] where Cl, C2, and C3 are constants.

The total potential root water uptake (TRWU) is integrated over RWU of root length in all soil
layers, TRWU = ∑RWU(L), and is then compared with climatic EPO. If EPO is less than
TRWU, then actual root water uptake is limited to EPO (Eq. [5] and [6]). The TRWU is
usually larger than EPO, until the soil water reaches a given level of depletion.

RWU(L) = Ci x EXP[MIN({C2(L) x [SW(L) - LL(L)]},40)] {C3 - ALOG[RLD(L)]} [4]

IF TRWU > EPO , THEN EP = EPO [5]
IF TRWU < EPO , THEN EP = TRWU [6]
The ratio of TRWU to EPO is used to compute water deficit effects, via two different
ratios (SWFAC and TURFAC) per Eq. [7] and Eq. [8]. These are signals that regulate crop
processes. Figure 1 shows graphically the meaning of SWFAC and TURFAC. TURFAC and an
60
“early signal are also computed from TRWU and EPO, but are scaled by 1.5 or Ki to act
before photosynthesis is reduced. SWDF3 in Eq. [9] is a hypothetical “early signal with a Ki
ranging from 2.5 to 5.0, and is designed to act sooner than TURFAC or SWFAC.

SWFAC = TRWU EPO, limited to a maximum of i.0 [7]

Fig. 1 Four water stress signals, SWFAC

for photosynthesis, TURFAC for expansive
processes, N-Fix-Soybean for nodule
growth and nitrogenase activity, and an
“early” signal, computed from
the ratio of potential root water supply to
transpirational demand for water

TURFAC = TRWU (EPO x 1.5), limited to a maximum of 1.0 [8]

SWDF3 = TRWU (EPO x K1), limited to a maximum of 1.0 [9]

When the SWFAC (Eq. [7]) is less than 1.0, then daily photosynthesis and
transpiration are reduced in proportion to SWFAC. This is a mimic of stomatal action, allowing
CO2 to be fixed in proportion to the stomatal opening to allow for transpiration. CROPGRO
has no vapor pressure deficit effect on photosynthesis or stomatal function. When SWFAC is
less than 1.0, root depth progression is accelerated, leaf senescence is more rapid, crop
phenology may be delayed or accelerated depending on the crop growth phase, and N is
mobilized more rapidly during seed fill. When TURFAC (Eq. [8]) is less than 1.0, the
expansion of new leaves and internode elongation (height and width increase) are reduced. A
TURFAC less than 1.0 reduces rate of leaf appearance (V stage), specific leaf area of new
leaves, the increase in height and width, N fixation, and it shifts allocation from leaf and stem
toward root.

Root Growth function

Root mass and root length in each soil layer are computed on a daily step basis. New root
length produced each day depends on daily assimilate allocated to roots and a constant length
to weight parameter. Fraction partitioning of assimilate to root varies with crop growth stage
and eventually becomes zero when reproductive growth uses all the daily assimilate. The
distribution of the new RLD into respective layers depends on progress of downward root
depth front, a soil rooting preference function (SRGF) describing the probability of roots

61
growing in each soil layer, and the soil water content of each layer. The SRGF defines the
hospitality of the soil (soil impedance, soil pH, soil nutrient effect, and organic matter effect)
to root proliferation. Rate of root depth progression (RTPROG) in Eq. [10] is a function of
thermal time accumulation (DTX), a species potential root depth progression rate in
centimeters per thermal day (RFAC2), is accelerated as much as 15% by SWFAC, and is
dependent on soil water status of the rooting front layer (reduced only when fraction available
water is less than 0.25 or reduced if water content is very high, within 2% of saturation). The
SWDF(L) in Eq. [10, 11, 12] is not fraction available water and is not SWFAC, rather
SWDF(L) = {SW(L) - LL(L)] (0.25 x [DUL(L) - LL(L)]} and mimics increased soil impedance,
occurring when available water in layer L is less than 25% of (DUL - LL). SWEXF(L) is
anaerobic stress computed from fraction of pore space filled with water for the rooting front
layer (L). While CROPGRO accelerates root depth with water stress, CERES Maize is
different, and uses SWFAC as a third factor in the MIN part of the equation to limit root depth
progression with plant water stress.

RTPROG = DTX x RFAC2 x MIN[SWDF(L), SWEXF(L)]

x {1.0 + 0.25 x [1.0 - MAX(SWFAC, 0.40)]} [10]

The fraction of new root length increase allocated to a given layer of soil, RLDF (L),
is dependent on the Soil Root Growth preference Function [SRGF(L)], layer thickness
[DLAYR(L)], the presence of the rooting front in that layer, and the soil water status of that
layer [SWDF(L) and SWEXF(L)] defined the same as for root depth progression (Eq. [11]). It is
assumed that roots will grow into a soil layer if its water content is above 25% of (DUL - LL).
RLDF(L) = SRGF(L) x DLAYR(L) x MIN[SWDF(L), SWEXF(L)] [11]
There is senescence of RLD in each layer as a function of thermal time, accelerated
when available soil water content is below a critical fraction (0.25) or near saturation. Root
senescence is influenced by soil water status and water excess of each soil layer. In Eq. [12],
RTSURV is fraction root survival, RTSDF is fraction root death under zero water availability
and RTEXF is fraction root death under fully saturated soil.
RTSURV(L) = MIN(1.0,{1.0 - RTSDF x [1.0 - SWDF(L)]},

{1.0 - RTEXF x [1.0 - SWEXF(L)]}) [12]

In addition, Eq. [13] provides a thermal time dependent rate of root length senescence in
each layer, where RTSEN (0.02) is the fraction senesced per day if at optimum temperature:
RLSEN(L)=RLV(L) x RTSEN x DTX [13]
As a result of these equations, roots tend to grow and accumulate in moist soil layers

62
and diminish in the saturated or drying soil layers (less than 25% available water). Typically, the
simulated RLD of the top 5 cm layer is less than that in the 5 to 15 cm layer because of more
frequent soil drying, despite both layers having the same soil hospitality factor.
Conclusions

Most widely used DSSAT models use the Ritchie tipping bucket soil water balance
model that works satisfactorily, when the soil water holding traits (DUL, LL) are estimated
properly for the soil in question and when root growth is adequately predicted. There are four
PET options used with CROPGRO of which three (FAO 24, FAO 56, and a prototype hourly
energy balance) require additional inputs of wind run and dew-point temperature.

References

Allen, R.G., L.S. Pereira, D. Raes, and M. Smith. 1998. Crop evapotranspiration. Guidelines for
computing crop water requirements. FAO Irrig. And Drainage Paper no. 56. FAO, Rome,
Italy.
Jensen, M.E., R.D. Burmaan, and R.G. Allen (ed.) 1990. Evapotranspiration and irrigation
water requirements: A manual. Am. Soc. Of Civil Engineers, New York.
Priestley, C.H.B., and R.J. Taylor. 1972. On the assessment of surface heat and evaporation
using large scale parameters. Mon. Weather Rev. 100:81-92.
Porter, C.H., J.W. Jones, G. Hooogenboom, P.W. Wilkens, J.T. Ritchie, N.B. Pickering, K.J.
Boote, and B. Baer. 2004. DSSAT v4 soil water balance module. P. 1-23. In J.W. Jones et
al. (ed.) Decision support system for agrotechnology transfer Version 4.0, Vol. 4 DSSAT
v4: Crop model documentation. Univ. of Hawali, Honolulu, HI.
Ritchie, J.T. 1972. Model for predicting evaporation from a row crop with incomplete cover.
Water Resour. Res. 8:1204 – 1213.
Ritchie, J.T. 1985. A user-oriented model of the soil water balance in wheat. p. 293-305. In E.
Fry and T.K. Atkin (ed.) Wheat growth and modeling. NATO-ASI Series, Plenum Press.
Ritchie, J.T. 1998. Soil water balance and plant water stress. p. 41-54. In G.Y. Tsuji et al. (ed.)
Understanding options for agricultural production. Kluwer Academic Publ., Dordrecht,
The Netherlands.

63
 M i n i m u m D a t a s e t fo r s i m u l a t i o n m o d e ls
D r. V . U . M . R a o a n d N . M a n i k a n d a n
P r o j e c t c o o r d i n a t o r ( A g me t )
C R I D A , S a id a ba d , H yd e r a ba d – 5 9
E ma i l : vu mr a o @ c r i d a . e r n e t . i n

I n t ro d u c t i o n

Crop growth simulation models are process oriented and intended to have wider
applications, and work independent of location, season, crop cultivar, and management system.
The models simulate the effects of weather, soil water, genotype, and soil water balance, and
crop photosynthetic, nitrogene dynamics on crop growth and yield. In simple, these models /
software integrate crop, weather, soil and management practices to simulate growth and
development of various crops. For effective utilization of simulation models, good quality data
are important and these models require exhaustive data on weather, soil, cultivar and
management aspects. Practically, it is very difficult to have all required data for simulation.
However, with the available data source it is possible to run the simulation models for various
purposes. This is called as minimum dataset and it refers to a minimum set of data required to
run the crop models and validate the outputs.

Types of data

Crop growth simulation models have Data Base Management System (DBMS) and are
used to organize and store the Minimum Data Set (MDS). The minimum data required to run
crop models are different for different crop growth models (INFOCROP, DSSAT, EPIC,
APSIM etc.,). But, in general following datasets are required to run the crop models.

Weather Data This data base contains daily data of temperature (Maximum and minimum),
sunshine hours / solar radiation, rainfall. In addition to this, data on wind speed, relative
humidity, soil moisture at different depths if available, it is added advantage.

Soil Data

This data base is comprised principally soil physical, chemical and biological properties
of the experimental site.

64
Crop management data

This data includes information on planting date, dates when initial soil conditions were
measured before planting/sowing, plating density, row spacing, variety, irrigation and fertiliser
practices.

Observed / measured crop data

This data comprises time series (phenology-wise or at preferred intervals) data on dry
matter production, plant height, number of seeds/pods per plant, leaf area index, by product yield
and grain yield etc., The minimum dataset required for running crop simulation for rice crop
and soil profile data for DSSAT model are given in appendix 1 and 2.

Data requirements for agricultural crops are closely associated with the level of analysis
(Nix, 1984; Bouman and Lansigan, 1994; Eswaran et al., 1996) and to the purpose, domain or
level of analysis considered in the study. Each level of analysis of crop production has its own
data requirements based on the details and the resolution required for the crop models.
Alternatively, data requirements for crops may be specified based on the level of crop
production (Lovenstein et al., 1993; Kropff et al., 1994) which considers the factors that define
and/or limit crop growth and development.

Collaboration between organisations/research institutes needed

Efficient interchange of data among researchers, especially for use in simulation models
and other decision support tools, requires use of a common terminology and approach for
organizing data. The agricultural research community increasingly encounters research problems
that require interdisciplinary collaboration. Agrometeorologists wish to study the impact of
climate change of crop production/productivity. Physiologists and molecular biologists work
together to develop a better understanding of the genetic control of productivity-related traits.
Agronomists, soil scientists and irrigation specialists combine efforts in order to increase the
efficiency of crop water use. In such collaborations, ready data interchange is possible. Genomic
data are widely available through publicly accessible databases (Blanchard, 2004). Daily
weather records and soil profile data are increasingly available through the Internet. The
International Research Institute for Climate Prediction recently developed two daily weather
data download options in ICASA format that can be accessed from the ICASA web site
(www.icasa.net/weather_data). Efforts are also underway to make the “World Inventory of Soil
Emission Potentials” (WISE) database developed by the International Soil Reference and

65
 Information Centre in The Netherlands available for crop model applications. However getting
field research data is very difficult through public databases. Although there have been various
initiatives (Like INARIS) to develop systems for reporting and storing data from field research,
to date, no system is available for perfect.

References

Nix, H.A., 1984. Minimum data sets for agrotechnology transfer. In: Proceedings of the
International Symposium on Minimum Data Sets for Agrotechnology Transfer, March 21-
26, 1983, ICRISAT, Patencheru, India, pp. 181-188.

Bouman,B.A.M. and F.P. Lansigan, 1994. Agroecological zonation and characterization. In:
Bouman, B.A.M et al.(Eds.) Agroecological zonation, characterization and optimization of
rice-based cropping systems. SARP Research Proceedings, Wageningen and Los Baños,
pp.1-8.

Eswaran, H., F. Beinroth, and P. Reich, 1996. Biophysical consideration in developing resource
management domains., Proceedings of the International Workshop on Resource
Management Domains, Kuala Lumpur, Malaysia, August 26-29,1996.

Lovenstein, H., H.A. Lantinga, H. van Keulen, 1993. Principles of Production Ecology,
Wageningen Agricultural University, Wageningen, The Netherlands.

Kropff, M.J., H.H. van Laar, and R.B. Matthews (Eds.), 1994. ORYZA1: An ecophysiological
model for irrigated rice production, SARP Research Proceedings, AB-DLO/WAU and IRRI,
Wageningen and Los Banos, 110 pp.

MINIMUM DATASET REQUIRED FOR CROP WEATHER RELATIONS AND CROP

SIMULATION MODELS – RICE

I. Station details and weather data

Station details (Latitude, Longitude and Altitude) and daily weather data viz., Tmax &
Tmin (°C), Rainfall (mm), sunshine hours (hrs), Solar radiation (MJ / m2), Wind speed (km /hr),
RH (% - I & II), Pan evaporation (mm) are needed throughout year.

66
II. Soil data

S. No Parameters Value / Text

1 Soil colour
2 Soil texture
3 Drainage
(Very excessive / Excessive / Some
what excessive / Well / Moderately
well / Some what poorly / poorly /
very poorly
4 Soil series name
5 Soil classification
6 Slope (%)
7 Electrical conductivity (dS / m)
Layer wise / depth wise information (Minimum 3 layers)
8 Clay (%)
9 Silt (%)
10 Sand (%)
11 pH
12 Cation exchange capacity (Cmol/kg)
13 Total nitrogen (%)
14 Field capacity (Volume basis)
15 Wilting Point (Volume basis)
16 Bulk density (Mg / m3)
17 Run off potential
(Lowest / Moderately low /
Moderately high/Highest)
18 Organic carbon (%)
19 Soil moisture at sowing (Volume
basis)
20 NO3 content at sowing (Kg / ha)
21 NH4 content at sowing (Kg /ha)

67
III. Experimental data
S. No Parameters Date / Value / Text
1 Variety name
2 Type of variety (Short / medium / long duration)
3 Duration of variety (in days)
4 Previous season crop
5 Experimental design
6 Sowing method (Transplanted / direct seeded)
7 Sowing depth (If, direct seeded)
8 Puddling (Yes / No)
9 Seed rate
10 Row to row spacing
11 Age of seedlings at transplanting day
12 No of seedlings / hill
13 Plant population (No of hill / m2)
12 Sowing date
13 Transplanting date
14 Panicle Initiation stage
15 Heading stage
16 50% flowering / Anthesis
17 Beginning of grain filling
18 End of grain filling
19 Physiological maturity
20 Harvesting
21 *No of effective tillers / m2
22 *No of grains / ear
23 *No of grains / m2
24 *Single grain weight (g)
25 *Straw yield (Kg / ha)
26 *Biomass yield (Kg/ha)
27 *Grain yield (Kg / ha)
28 Harvest index
Irrigation details

68
 29 Irrigation amount (mm)
30 Date of irrigation (in DAT)
31 Irrigation method
Fertiliser management
32 Date of fertilizer (N, P and K) application (DAT)
33 Amount of fertilizer (N, P and K) application
34 Type of fertilizer (N, P and K)
35 Depth of fertliser application
36 Fertiliser application method
Farmyard manure / green manure applied, if
any
37 Name of green manure crop
38 Application date
39 Amount
*Data needs replication-wise
IV. Periodic measurements of crop growth parameters

Periodic measurements of Leaf area index, Dry matter production (leaf, stem, root, ear and
total), Plant height, Specific leaf area, Relative growth rate and PAR inside & outside
canopy are to be taken through out crop season. Measurements should be started at 30 DAS and
continued at 15 days interval up to physiological maturity.

In addition to these observations, any biotic stress (Heavy weed infestation, insect and
disease) and abiotic stress should be noted down during crop season.

Appendix-2: Inputs required for creating a new soil profile for DSSAT Crop Model

I. General Information

1. Country : 2. Site Name:

3. Latitude: 4. Longitude:
5. Soil Data source 6. Soil Series name
7. Soil Classification

69
II. Surface Information
1. Colour (a) Brown (b) Red (c) Black (d) Gray (e) Yellow

2. Drainage (a) Very excessive (b) Excessive (c) Some what excessive
(d) Well (e) Moderate well
(f) Some what poorly (g) Poorly (h) Very poorly
3. % slope
4. Runoff potential (a) Lowest (b) Moderately low (c) Moderately high (d) Highest
5. Fertility factor (0 to 1)
6. Runoff Curve Number
7. Albedo
8. Drainage rate

III. Layer-wise soil information: No. of layers depends on the location. Here layers up to 120 cm
depth are shown as a sample.

Depth Master Clay Silt Stones Organic pH in Cation Total

(bottom) Horizon % % % Carbon water Exchange Nitrogen
Cm % Capacity %
C mol/kg

0-5

5-15

15-30

30-45

45-60

60-90

90-120

70
 Table continued….

Depth Lower Drainage Saturation Bulk Saturated Root growth

(bottom) limit Upper Density Hydraulic Factor 0.0-1.0
Cm Limit g/cm3 Conductivity
Cm/hr

0-5

5-15

15-30

30-45

45-60

60-90

90-120

71
 Agro-climatic analysis: Weathercock Software
Dr. VUM Rao
Project Coordinator (Agromet)
CRIDA, Saidabad, Hyderabad – 500 059
Email: [email protected]
1. Introduction

Climate is the primary important factor for agricultural production. Concerning the
potential effects of change in weather parameters on agriculture has motivated important change
of research. The research on climate change concentrates on possible physical effects of climatic
change on agriculture, such as changes in crop and livestock yields as well as the economic
consequences of these potential yield changes. In order to achieve maximum and sustainable
crop production from available farm resources, it is essential to have proper knowledge of the
agro climatic resources of the location/region. Agroclimatological analysis is used to study about
climatic characteristics and crop performance of a particular region and also to know the
climatic variability/climate change and its impact on agriculture. Therefore, a thorough
understanding of the climatic conditions would help in determining the suitable agricultural
management practices for taking advantage of the favorable weather and avoiding or minimizing
risks due to adverse weather conditions.

2. Purpose and scope

This hands-on-training on agroclimatic analysis will help to study location-specific and

spatial based agroclimatic resource characterization especially for the Indian conditions.
Agroclimatic information is necessary in enhancing crop productivity through better agricultural
planning including land use planning, water resources availability, crop suitability, pests and
disease management and also in weather based agro advisories. This particular hands-on-training
is aimed to develop skills of technical persons/scientists towards agroclimatic analysis, which
helps in efficient utilization of available resources, and also to develop the resources for
sustainable agricultural growth. In this connection, agroclimatological analysis is helpful in
understanding change and variability in climatic characteristics so as to get an idea on possible
impacts on crop performance for a particular location / region. The reliance on the change /
variability of climatic parameters at sub-district / block / mandal / tehsil level is increasing as
there is paucity of reports and government agencies are in dire need. This compelled AICRPAM
Unit of CRIDA, to develop a computer software program “WEATHERCOCK” (Fig.1.) for agro-
climatic analysis. Different agro-climatic analysis viz., converting daily weather data on to
weekly, monthly, seasonal and annual data, rainy days analysis, meteorological and agricultural
72
drought analysis, probability analysis, water balance, extreme event analysis for temperature and
rainfall and to estimate length of growing season have been brought out in to one umbrella. This
particular software is based on Visual Basic (VB) and easy to operate even by beginners. Doing
agro-climatic analysis with MS – EXCEL for individual stations is drudgery and may lead to
wrong results. The weathercock software reduces this drudgery and eliminates any mistakes
associated with MS-EXCEL. Moreover, “batch processing” a special provision was made in the
weathercock to facilitate to run the analysis for hundreds of stations at a moment if input files
are prepared in the said format as doing agro-climatic analysis at localized scale have hundreds /
thousands of stations.

Fig.1. Main window of “WEATHERCOCK” software program

3. Determining minimum weather data requirement

To characterize a place/region long-term weather data is basic requirement. Different

type of analysis needs different weather elements either in single or in multiple. For example,
rainfall analysis/drought analysis (meteorological & agricultural) and rainfall trend analysis
require only rainfall data. In case of water balance analysis, seven weather parameters viz.,
temperature (maximum and minimum), relative humidity (morning and evening), wind speed,
sunshine hours and evaporation are necessary to compute potential evapotarnspiration (PET).
Thus, it is desirable to collect all weather parameters (minimum data set for agroclimatic
analysis) at a time. The minimum data set for agroclimatic analysis includes (i) rainfall (ii)
temperature (maximum and minimum) (iii) relative humidity (morning and evening) (iv) wind
speed (v) sunshine hours and (vi) evaporation. In addition to these weather parameters,
coordinates of place (Latitude, Longitude and Altitude), soil information like permanent wilting

73
point and field capacity of the soil (maximum water holding capacity) are also needed for agro
climatic analysis.
4. Description of programs
4.1. Converting daily weather data into weekly, monthly, seasonal and annual values along
with CV and normals
Daily data of weather parameters is base for all agroclimatic analysis. Though it is base,
it is necessary to convert these data in to weekly / monthly / seasonal / annual format in order to
understand distribution of different weather parameters over different periods (weeks / months /
season / annual). These tools (weeks / months / season / annual) are very useful to characterize
the region in relation to weather.
4.1. Number of rainy days along with Coefficient of Variation
The number of rainy day analysis gives an idea on rainy days in a week / month / season /
annual. Information of rainy days of a place over a period of time determine the need and design
both for rainwater harvesting and structure to recharge groundwater aquifers. With the help of
number of rainy days planners may plan cropping pattern/cropping systems.
Rainy day: A day with rainfall amount equal or more than 2.5 mm considered as a rainy day
according to India Meteorological Department for Indian region.
4.2. Initial and conditional probabilities and probability for consecutive wet and dry weeks
Agricultural operations are determined by the certain amount of rainfall received in a
period. There are specific amounts of rainfall required for the activities like land preparation,
sowing and for various agricultural activities. Hence, estimation of probabilities with respect to a
given amount of rainfall is useful for rainfed agricultural planning especially in semiarid region.
Initial probability rainfall analysis will give percentage probability to get certain amount of
rainfall in a given week. Probability of wet week is denoted as P(W) and dry week as P(D).
Conditional probability rainfall analysis will give the percentage probability for wet week
followed by wet week [P(W/W)], wet week followed by dry week [P(W/D)], dry week followed
by dry week [P(D/D)] and dry week followed by wet week [P(D/W)]. Probability of wet and dry
week program is used to find out percentage probability of consecutive wet weeks (2W, 3W,
4W) and consecutive dry weeks (2D, 3D, 4D). For efficient planning, research workers, farmers
and planners stand to gain significantly by using quantified rainfall at different probability levels
called assured rainfall. Incomplete gamma probability model is used for computing the assured
rainfall amount at different probability levels.
Initial probability: It is the probability of receiving a certain amount of rainfall in a given week.
Conditional probability: It is the probability of getting a next week as a wet week, given the
condition that the current week is also a wet week.

74
Consecutive wet and dry weeks: It is the probability of getting two or three or four weeks as a
wet week consecutively for a given amount of rainfall. The probability for getting consecutive
dry weeks refers to probability for getting less than the given amount of rainfall consecutively
for two/three/four weeks.

4.4. Drought analysis

4.4.1. Classification of droughts
Drought is a normal, recurrent climatic feature that occurs in virtually around the world
causing huge loss for the farming community. Drought is universally acknowledged as a
phenomenon associated with deficiency of rainfall. There is no single definition, which is
acceptable universally. Droughts occur at random and there is no periodicity in its occurrence
and cannot be predicted in advance. In semiarid stations, the occurrence of rainfall is seasonal
and is known more for its variability with respect to space and time. Drought is characterized by
moisture deficit resulting either from i) Below normal rainfall ii) erratic rainfall distribution iii)
higher water need iv) a combination of all the three factors. Wilhite and Glantz (1985) analysed
more than 150 such definitions of drought and thus broadly grouped these into four categories
and explained below.
Meteorological drought: A period of prolonged dry weather condition due to below normal
rainfall.
Agricultural drought: Agricultural impacts caused due to short-term precipitation shortages,
temperature anomaly that causes increased evapotransipration and soil water deficits that could
adversely affect crop production.
Hydrological drought: Effect of precipitation shortfall on surface or sub-surface water sources
like rivers, reservoirs and groundwater.
Socio-economic drought: The socio-economic effect of meteorological, agricultural and
hydrological drought in relation to supply and demand of the society.
4.4.2. Analysis of meteorological and agricultural drought
In agroclimatic analysis, meteorological and agricultural drought study is important. The
frequencies of occurrence of different type of meteorological droughts (moderate and severe)
over a period of year would give insight for vulnerability of a particular location/region to
drought on annual basis. Agricultural drought analysis would give idea about susceptibility of a
region to drought on seasonal basis, i.e., main crop growing season.

Meteorological Drought- According to India Meteorological Department

2 types: based on rainfall deficit from normal

75
 · Moderate : 26-50%
· Severe : > 50%
Agricultural drought: According to National Commission on Agriculture, 1976, at least four
consecutive weeks receiving less than half of the normal rainfall during Kharif season and six
such consecutive weeks during Rabi season is considered as agricultural drought peiord.
Normal rainfall: Average rainfall for a location/region over a period of years (preferable 30
years).
4.5. Water balance analysis
a. Water balance analysis
Availability of water in right quantity and in the right time and its management with
suitable agronomic practices are essential for good crop growth, development and yield. To
assess water availability to crops, soil moisture is to be taken into account and the net water
balance through soil moisture can be estimated using the water balance technique. The concepts
of PET (Potential evapotransipiration) and water balance have been extensively applied to
studies such as climatic classification, aridity, humidity and drought.
b. Computation of Potential Evapotranspiration (PET)
Information on PET for a location on a short timescale has great importance in
agricultural water management. Many empirical methods are available viz., Thornthwaite
(1948), Blaney and Criddle (1950), Hargreaves and Christiansen (1973) and others. Guidelines
were developed and published in the FAO Irrigation and Drainage Paper No 24 “Crop Water
Requirements” (Doorenbos and Pruitt 1977) to compute ETo using several methods. FAO
Penman – Monteith method is recommended as the sole standard method. FAO Penman-
Monteith method is selected as the method by which the evepotranspiration of the reference
surface (ETo) can be unambiguously determined, and this method provides consistent ETo
values in all regions and climates.

c. Computation of indices like Aridity index (Ia), Humidity Index (Ih), Moisture Index (Im)
and MAI
These indices are output of water balance analysis. The indices viz., Aridity index (Ia),
Humidity Index (Ih), Moisture Index (Im) are useful in climatic classification and to find
climatic type of a particular place. Moisture Adequacy Index (MAI) provides a good indication
of the moisture status of the soil in relation to the water-need, high values of the index signifying
good moisture availability and vice versa.

76
d. Components like water surplus, water deficit, Actual Evapotranspiration (AET)
Water surplus (WS) and water deficit (WD) occur in different seasons at most places and
both are significant in water balance studies. The information about when the period of water
surplus and deficit occurring in a season or year is helpful to find ideal period for starting of crop
season and stages, which may fall in deficit period. It also helps in flood and drought analysis.
Potential evapotranspiration (PET): It is defined as the maximum quantity of water, which is
transpired and evaporated by a uniform cover of short dense grass (Reference crop) when the
water supply is not limited.
Reference crop: A hypothetical reference crop with an assumed crop height of 0.12m, a fixed
surface resistance of 70s/m and an albedo of 0.23.
Water balance: It refers to the climatic balance obtained, by comparing the rainfall as income
with evapotarnspiration as loss or expenditure, soil being a medium for storing water during
periods of excess rainfall and utilizing or releasing moisture during periods of deficit
precipitation.
Water surplus: It is the excess amount of water remaining after the evaporation needs of the
soil have been met (i.e., when actual evapotranspiration equals potential evapotranspiration) and
soil storage has been returned to the water holding capacity level.
Water deficit: It is the amount by which the available moisture fails to meet the demand for
water and is computed by subtracting the potential evapotranspiration from the actual
evapotranspiration for the period of interest.

Actual Evapotranspiration: It is the actual amount of water lost to the atmosphere by

evaporation and transpiration under existing conditions of moisture availability.

Aridity Index Ia (%) = Water deficit / PET * 100

Humidity Index Ih (%) = Water surplus / PET * 100
Moisture Index Im (%) = Ih – Ia
Moisture Adequacy Index MAI (%) = AET / PET *100
References
Thornthwaite, C. W. 1948. An approach toward a rational classification of climate.
Geographic Review 38:55-94
Blaney, H.F. & Criddle, W.P. 1950 Determining water requirements in irrigated areas from
climatological and irrigation data. U S Department of Agriculture, Soil Conservation Series
(SCS) TP-96, 48

77
Doorenbos. J. And Pruitt, W.O. 1977. Crop water requirements, irrigation and Drainage Paper
No. 24, (rev.) FAO, Rome,
Penman-Monteith method. 1977. Crop Evapotranspiration - Guidelines for computing crop
water requirements, Paper No. 56, (http://www.fao.org/docrep/X0490E/X0490E00.htm)
Wilhite, D.A. and M.H. Glantz. 1985. Understanding the Drought Phenomenon: The Role of
Definitions. Water International 10:111-120.
National Commission on Agriculture. 1976. Rainfall and Cropping Patterns; Vol. XIV,
Government of India, Ministry of Agriculture, New Delhi.
The Fourth Assessment Report-AR4. 2007. IPCC - Intergovernmental Panel on Climate
Change, Working Groups Report.(www.ipcc.ch/ -)

78
 Modeling nitrogen dynamics in the soil-plant system
K. Srinivas. CRIDA

Nitrogen modeling in CERES

The nitrogen dynamics routines of the CERES models were designed to simulate each of the
major N loss processes and the contributions to the N balance made by mineralization. The
routines also describe the uptake of N by the crop and the effects of N deficiency on crop growth
processes. The transformations simulated are mineralization and/or immobilization, nitrification,
denitrification, and urea hydrolysis. Nitrate movement associated with water movement in both
an upward and downward direction is also simulated. Since the rates of transformation of
nitrogen are very much influenced by soil water status, the simulation of nitrogen dynamics
requires that water balance also be simulated. Soil temperature greatly influences many of the
transformation rates. Therefore, a procedure to calculate soil temperature at various depths,
based on the soil temperature is also invoked in the nitrogen component of the model.

The model does not simulate losses by ammonia volatilization or ammonium exchange
equilibria and fixation. Under conditions of good fertilizer practice where fertilizer is either
incorporated or placed beneath the soil surface, volatile ammonia losses should be small.

Initialization

Inputs describing the amount of organic matter and the amount of mineral nitrogen present in the
soil are required to initialize the model. The model requires the organic carbon concentration in
each layer (OC(L)) as an input and using an assumed soil C:N ratio of 10:1 calculates the
amount of organic N associated with this organic matter (HUM(L)). These initializations are
performed in subroutine SOILNI. To determine the contribution of recent crop residues to the
supply of nitrogen in the soil, the model also requires an estimate of the amount of crop residue
(STRAW) which is present. Based on this estimate and the depth of incorporation (SDEP) of the
crop residue, the fresh organic matter content of each layer (FOM(L)) is estimated. An estimate
of the amount of root residue remaining from the previous crop is also required for the
calculation of FOM(L). Initial partitioning of the fresh organic matter into the component pools
of carbohydrate (FPOOL(L,1)), cellulose (FPOOL(L,2)) and lignin (FPOOL(L,3)) is also
performed in subroutine SOILNI.

79
NITRATE FLUX

Leaching of nitrates is probably the most common and best understood N loss process. Nitrates
leaching from soil often become a source of contamination of groundwater and has recently
generated interest in leaching from an environmental standpoint. There have been many
approaches to modelling leaching based on numerical techniques which require solution in a
manner inappropriate for use in a management level model such as CERES. In the CERES
model, leaching is simulated using a simple approach based on the cascading system for
drainage described in the previous chapter. Nitrate N may move between layers of the soil
profile in the CERES models, but the movement of ammonium is not considered. Nitrate flux
calculations are performed in subroutine NFLUX. Nitrate movement in the soil profile is highly
dependent upon water movement. Therefore, the volume of water present in each layer (SW(L)
* DLAYR(L)) and the water draining from each layer ((FLUX(L)) in the profile is used to
calculate the nitrate lost from each layer (NOUT) as follows:

NOUT = SNO3(L) * FLUX(L)/(SW(L) * DLAYR(L) + FLUX(L))

A fraction of the mass of nitrate (SNO3(L)) present in each layer thus moves with each drainage
event. A simple cascading approach is used where the nitrate lost from one layer is added to the
layer below. When the concentration of nitrate in a layer falls below a critical level, no further
leaching from that layer is deemed to occur. The method used may be termed a "reservoir
mixing model", but water movement is controlled by the SWCON variable in the drainage
routine. The implicit assumption is that all the nitrate present in a layer is uniformly and
instantaneously in solution in all of the water in the layer. Thus no attempt is made to separate
nitrate in solution between the retained water and the mobile water. Differences in the relative
volumes of retained water and mobile water between clays and sands occur as a function of the
relative magnitudes of LL(L), DUL(L), and SAT(L). The rate of nitrate flux is also sensitive to
changes in SWCON since this variable determines the rate of drainage. Nitrate is more readily
displaced from sands since the volume of water which can move ((SAT(L) - DUL(L)) *
DLAYR(L)) is large in comparison to the retained water (DUL(L) * DLAYR(L)). Most of the
difference in the simulated leaching rate between soils of different texture is explained by this
difference in proportion of water which is mobile. Some difference is also attributable to the rate
at which the soil profile can drain (SWCON). The upward flow of water in the top four soil
layers will also cause some redistribution of nitrate. A second loop, commencing in the deepest
layer of evaporative water loss (MU), is used to calculate this redistribution. Nitrate moving

80
from a layer (NUP) is calculated as a function of upward movement (FLOW(L)) in a manner
identical to leaching:

NUP = SNO3(L) * FLOW(L)/(SW(L) * DLAYR(L) + FLOW(L)) * 0.5

No upward loss from the top layer occurs by this process. Since there will occasionally be
instances when this slowly moving water can move in a downward direction (negative values of
FLOW(L)) a third loop is set up with calculations commencing in the top layer and running to
lower layers. This is achieved by first reinitializing the array FLUX to 0 and reversing the sign
(to make it positive) at the FLOW array and copying it to FLUX. When this has been done the
normal leaching calculations used in the first loop can be used again. These instances would
occur when a small rainfall wets the top layer of a very dry soil profile. There may have been
insufficient water for drainage to occur but a moisture potential between the top layer and the
second layer initiates this flow. The resultant movement of nitrate will be very small.

SOIL NITROGEN TRANSFORMATIONS

The CERES model simulates the decay of organic matter and the subsequent mineralization
and/or immobilization of N, the nitrification of ammonium and denitrification in subroutine
NTRANS. Fertilizer addition and transformations (assumed to be instantaneous) are also
performed in this subroutine.

Fertilizer Additions

Fertilizer N is partitioned in the model between nitrate and ammonium pools according to the
nature of the fertilizer used. Fertilizer products are specified by a numeric code IFTYPE. In
addition to the numeric code for fertilizer type, inputs required to describe the fertilizer are: the
date of application (FDAY), the amount of N applied (AFERT) and the depth of placement
(DFERT). For any placement depth the assumption is made that the fertilizer is uniformly
incorporated into the layer. Layer thicknesses are supplied as input and are usually based on
natural horizonation in the profile. These must correspond with those used to describe the soil
water inputs. Surface fertilizer applications are treated as being uniformly incorporated into the
top layer. Up to 10 split applications can be accommodated by the model.

Mineralization and Immobilization

Mineralization refers to the net release of mineral nitrogen with the decay of organic matter and
immobilization refers to the transformation of mineral nitrogen to the organic state. Both

81
processes are microbial in origin. Immobilization occurs when soil microorganisms assimilate
inorganic N compounds and utilize them in the synthesis of the organic constituents of their
cells. A balance exists between the two processes. When crop residues with a high C:N ratio are
added to soil, the balance can shift resulting in net immobilization for a period of time. After
some of the soil carbon has been consumed by respiration, net mineralization may resume. N
mineralized from the soil organic pool can often constitute a large part of the nitrogen available
to the crop.

The perceived application for the CERES models in studies examining crop growth and fertilizer
management requires that a mineralization model be simple, require few inputs, and work on a
diversity of soils. Simulation studies examining the affects of crop residues also requires that the
model be capable of simulating the fate of residues of different compositions. Other studies
examining the potential role of nitrification inhibitors require a model wherein the processes of
ammonification and nitrification are separated. The approach used in the CERES-WHEAT
model is based on a modified version of the mineralization and immobilization component of the
PAPRAN model (Seligman and van Keulen, 1981). This model is an attempt at maintaining
some of the functionality of the microbiological level models but doing so at a very simplified
level. The model's modifications have been to simulate nitrification separately and to partition
the simulated fresh organic matter pools differently. Modifications were also made to
temperature and water indices to fit the CERES water balance and soil temperature routines.
Unless otherwise indicated, the coefficients used for the mineralization/immobilization functions
described below were drawn from the PAPRAN model.

The mineralization and immobilization routine simulates the decay of two types of organic
matter: Fresh organic matter (FOM) which comprises crop residues or green manure and a stable
organic or humic pool (HUM). Three pools comprise the FOM pool in each layer (L), vis:

FPOOL(L,1) = carbohydrate

FPOOL(L,2) = cellulose

FPOOL(L,3) = lignin.

In PAPRAN, FOM is simulated as one pool and the decay rate constant is selected according to
the proportion of the initial amount of FOM remaining. The CERES model separates FOM into
three pools giving a better estimate of soluble carbon which is used in the denitrification routine.
These three pools are initialized as a fraction of the FOM(L) pool in subroutine SOILNI.

82
Initially, the FOM(L) contains 20% carbohydrate, 70% cellulose and 10% lignin. The model
requires as input data, the amount of straw added, its C:N ratio and its depth of incorporation (if
any) and an estimate of the amount of root residue from the previous crop. Based upon these
data, initial values of FOM and the N contained within it (FON) for each layer are calculated in
subroutine SOILNI. The soil organic carbon in each layer (OC(L)) is also required by the
mineralization routine. This is used to calculate HUM(L), and together with a simplifying
assumption of a bulk soil C:N ratio of 10, is used to estimate the N associated with this fraction
(NHUM(L)). Each of the three FOM pools (FPOOL (L,1 to 3)) has a different decay rate
(RDECR

(1 to 3)). Under nonlimiting conditions the decay constants as reported by Seligman and van
Keulen (1981) are 0.80, 0.05, and 0.0095 for carbohydrate, cellulose, and lignin, respectively. A
decay constant at 0.20 for the carbohydrate fraction has since been found to be more appropriate.
The decay constant for carbohydrate implies that under nonlimiting conditions 20% of the pool
will decay in one day. Nonlimiting conditions very seldom occur in soils since one or all of soil
temperature, soil moisture, or residue composition will limit the decay process. To quantify
these limits three zero to unity dimensionless factors are calculated. A water factor (MF) is first
determined from the volumetric soil water content (SW(L)) relative to the lower limit (LL), and
drained upper limit (DUL). In accordance with the soil water balance model, provision is made
for the water content of the uppermost layer to be lower than the lower limit. The variable
SWEF determines the lowest possible value the uppermost soil layer water content have. When
the soil is drier than DUL, MF is calculated as:

AD = LL(L)

IF (L.EQ.1) AD = LL(1) *SWEF

MF = (SW(L)-AD)/(DUL(L)-AD)

where:

AD = lowest moisture content for a layer (volume fraction)

When the soil is wetter than DUL, MF is calculated as:

MF = 1.0-(SW(L)-DUL(L))/(SAT(L)-DUL(L))*0.5

The functions follow the observations reported by Myers et al. (1982)

83
and Linn and Doran (1984) on moisture effects on ammonification. Under very wet conditions
(100% of water filled porosity) ammonification proceeds at approximately half of the rate of
ammonification at field capacity (Linn and Doran, 1984). The comparative effects of soil
moisture on the simulated rates of ammonification, nitrification, and denitrification can be seen
in Fig. 5.1. A temperature factor (TF) is calculated directly from soil temperature (ST(L)):

TF = (ST(L)-5.0)/30.0

This approximates the soil temperature effects on ammonification reported by others (Stanford
et al., 1973; Myers, 1975). If the soil temperature (ST(L)) is less than 5o C then TF is set to zero
and no decay occurs. The C:N ratio (CNR) imposes the third limit on decay rate. In this case
C:N ratio is calculated as the C contained in FOM divided by the N "available" for the decay
process. This N available for decay is the sum of the N contained in the FOM, which is FON,
and the extractable mineral N present in the layer (TOTN). Thus,

CNR = (0.4*FOM(l))/(FON(L)+TOTN)

From CNR an index (CNRF) is calculated which has a critical C:N ratio of 25.

CNRF = EXP(-0.693*(CNR-25)/25.0)

Thus, in low N containing residues (e.g., freshly incorporated wheat straw) with a high C:N
ratio, the N available for the decay process will greatly limit the decay rate (Fig. 3.5). For each
of the FOM pools a decay rate (GRCOM) appropriate for that pool (JP) can be calculated. G1 =
TF*MF*CNRF*RDECR(JP)

GRCOM = G1*FPOOL1(L,JP)

The gross mineralization of N associated with this decay (GRNOM) is then calculated according
to the proportion of the pool which is decaying.

GRNOM = G1 * FPOOL(L,JP)/FOM(L) * FON(L)

GRCOM and GRNOM are summed for each of three pools in each layer. The procedure used for
calculating the N released from the humus (RHMIN) also utilizes TF and MF. In this case CNRF
is not used and the potential decay rate constant (DMINR) is very small (8.3E-05). A further
index (DMOD) was added to the RHMIN calculations to adjust the mineralization rate for
certain atypical soils. On soils with chemically protected organic matter, a less than unity value
of DMOD is required so that mineralization is not overestimated. On freshly cultivated virgin
84
soils, a slightly greater than unity value has been found necessary to account for the sudden
increase in mineralization activity. In all other circumstances a value of 1.0 is used for DMOD.
Satisfactory alternatives for estimating DMOD are currently being sought. The procedure for
calculating RHMIN, then is the product of the various indices and the N contained within the
humus (NHUM(L)).

RHMIN=NHUM(L)*DMINR*TF*MF*DMOD

After calculating the gross mineralization rate, HUM(L) and NHUM(L) are updated.

HUM(L)=HUM(L)-RHMIN*10.0+0.2*GRNOM/0.04

NHUM(L)=NHUM(L)-RHMIN+0.2*GRNOM

These calculations also allow for the transfer of 20% of the gross amount of N released by
mineralization of FON(L) (0.2*GRNOM) to be incorporated into NHUM(L). This accounts for
N incorporated into microbial biomass and has a concentration of 4% (0.04) determined as 0.1 g
N/g C (soil C:N ratio of 10) multiplied by 0.4 g C/g OM (40% of OM is C). As organic matter
decomposes some N is required by the decay process and may be incorporated into microbial
biomass. The N which is immobilized in this way (RNAC) is calculated as the minimum of the
soil extractable mineral N (TOTN) and the demand for N by the decaying FOM(L).

RNAC=AMIN1(TOTN,GRCOM*(0.02-FON(L)/FOM(L))

where 0.02 is the N requirement for microbial decay of a unit of FOM(I). The value of 0.02 is
the product of the fraction of C in the FOM(L) (40%), the biological efficiency of C turnover by
the microbes (40%) and the N:C ratio of the microbes (0.125). FOM(L) and FON(L) are then
updated.

FOM(L)=FOM(L)-GRCOM

FON(L)=FON(L)+RNAC-GRNOM

The balance between RNAC and GRNOM determines whether net mineralization or
immobilization occurs. The net N released from all organic sources (NNOM) is:

NNOM=0.8*GRNOM+RHMIN-RNAC.

85
Note that only 80% of GRNOM enters this pool since the remaining 20% was incorporated into
NHUM(L). NNOM can then be used to update the ammonium pool (SNH4(L)).

SNH4(L)=SNH4(L)+NNOM

If net immobilization occurs (NNOM negative) ammonium is first immobilized and if there is
not a sufficient amount to retain this pool with a concentration of 0.5 ppm, withdrawals are made
from the nitrate pool.

Nitrification

Nitrification refers to the process of oxidation of ammonium to nitrate. It is a biological process

and occurs under aerobic conditions. The main factors which limit nitrification are: Substrate
NH4+, oxygen, soil pH, and temperature. The approach used in the CERES models has been to
calculate a potential nitrification rate and a series of zero to unity environmental indices to
reduce this rate. This potential nitrification rate is a Michaelis-Menten kinetic function
dependent only on ammonium concentration and is thus independent of soil type. A further
index, termed a "nitrification capacity" index, is introduced which was designed to introduce a
lag effect on nitrification if conditions in the immediate past (last 2 days) have been unfavorable
for nitrification. Actual nitrification capacity is calculated by reducing the potential rate by the
most limiting of the environmental indices and the capacity index. The capacity index is an
arbitrary term introduced to accommodate an apparent lag in nitrification observed in some data
sets. The functions reported below were found to be appropriate across the range of data sets
tested. The nitrification routine in subroutine NTRANS calculates the nitrification of ammonium
in each layer. First, an ammonium concentration factor (SANC) is calculated.

SANC=1.0-EXP(-0.01363*SNH4(L))

This is a zero to unity index which has approximately zero values when there is less than 1 ppm
of ammonium present and has a value of 0.75 at 100 ppm. The temperature factor calculated
above for mineralization (TF) and a soil water factor for nitrification (WFD) (Fig. 3.4) are used
together with SANC to determine an environmental limit on nitrification capacity (ELNC).

ELNC=AMIN1(TF,WFD,SANC)

To accommodate lags which occur in nitrifier populations ELNC and the previous day's relative
microbial nitrification potential in the layer (CNI(L)) are used to calculate the interim variable
RP2 which represents the relative nitrification potential for the day.

RP2=CNI(L)*EXP(2.302*ELNC)
86
RP2 is constrained between 0.01 and 1.0. Today's value of the nitrification potential (CNI(L)) is
then set equal to RP2. Since EXP(2.302*ELNC) varies from 1.0 to 10.0 when ELNC varies
from 0.0 to 1.0, relative nitrification potential can increase rapidly, up to tenfold per day. An
interim variable A is then determined from these indices and an index for pH effect on
nitrification. This pH index is calculated in subroutine SOILNI and represents the conclusions
drawn by Schmidt (1982) on the pH effect in nitrification.

A=AMIN1(RP2,WFD,TF,PHN(L))

This interim variable A is used together with the ammonium concentration (NH(L)) in a
Michaelis-Menten function described by McLaren (1970) to estimate the rate of nitrification.
The function has been modified to estimate the proportion of the pool of ammonium (SNH4(L))
which is nitrified on a day.

B=(A*40.0*NH4(L)/NH4(L)+90.0))*SNH4(L)

A maximum of 80% at the ammonium pool is allowed to nitrify in one day. A check is made to
ensure some ammonium is retained in the layer and thus the daily rate of nitrification (RNTRF)
is

RNTRF=AMIN1(B,SNH4(L))

Following this calculation, soil nitrate and ammonium pools can be updated.

SNH4(L)=SNH4(L)-RNTRF

SNO3(L)=SNO3(L)+RNTRF

Finally, the soil temperature, moisture and NH4 after nitrification are used to update (CNI(L)),
which is used in the subsequent day's calculations.

SARNC=1.0-EXP(-0.1363*SNH4(L))

XW=AMAX1(WF,WFY(L))

XT=AMAX1(TF,TFY(L))

CNI(L)=CNI(L)*AMIN1(XW,XT,SARNC)

87
SARNC is a zero to unity factor for ammonium availability. WFD and WFY(L) are today's and
yesterday's soil water factors, respectively, and TF and TFY(L) are today's and yesterday's soil
temperature factors, respectively. The least limiting of the current day's and the previous day's
water and temperature factors are used in the calculation of the new value of CNI(L). This
prevents a single day of low soil temperature or water from severely reducing CNI(L). It is
important to note that the relative nitrification potential CNI(L) is calculated twice each day.
Since (EXP(2.302*ELNC)) varies from 1.0 to 10.0, CNI(L) increases prior to the calculation of
the nitrification rate. After the nitrification calculations when the level of ammonium has
declined, CNI(L) is reduced. The relative magnitudes of (EXP(2.302*ELNC) and
AMIN1(XW,XT,SARNC)) determine whether relative nitrification potential increases or
decreases over the short term.

Denitrification

Denitrification is the dissimilatory reduction of nitrate (or nitrite) to gaseous products including
N0, N20, and N2 (Knowles, 1981).

Denitrification is a microbial process which occurs under anaerobic conditions and is influenced
by organic carbon content, soil aeration, temperature and soil pH. The approach adopted in the
CERES models has been to adapt the functions described by Rolston et al. (1980) to fit within
the framework of the model and to match inputs derived from the water balance and
mineralization components of CERES. The basic function used by these authors was also used
by Davidson et al. (1978a) and was the subject of field testing under a variety of conditions in
California. Predicted rates of denitrification compared favorably with direct measures of gaseous
losses in the field experiments. Denitrification calculations are only performed when the soil
water content (SW) exceeds the drained upper limit (DUL). A zero to unity index (FW) (see Fig.
5.1) for soil water in the range from DUL to saturation (SAT) is calculated.

FW = 1.0 - (SAT(L)-SW(L))/(SAT(L)-DUL(L))

Linn and Doran (1983) used percentage of water filled porosity as an index of soil water
availability effects on soil N transformations. In their studies, denitrification commenced with a
water-filled porosity of 60% and increased linearly up to 100% water filled porosity. This
approximates the linear increase in FW as SW increases from DUL to SAT. A factor for soil
temperature (FT) is also calculated.

FT=0.1*EXP(0.046*ST(L))

88
Rolston et al. (1980) using the data of Burford and Bremner (1976) and Reddy et al. (1971) to
estimate the water-extractable C in soil organic matter (CW) as:

CW=24.5 + 0.0031*SOILC

In the CERES model, SOILC is calculated as 58% of the stable humic fraction. To this is added
the carbon contained in the carbohydrate fraction organic matter pool (40% of FPOOL(L,1)).
Appropriate unit conversions are made using FAC(L) and the total water extractable carbon
(CW) estimated.

CW = FAC(L)*(SOILC*0.0031+0.4*FPOOL(L,1))+24.5

Denitrification rate (DNRATE) is then calculated from the nitrate concentration and converted
to a kg N/ha basis for the mass balance calculations.

DNRATE = 6.0*1.0E-05*CW*NO3(L)*FW*FT*DLAYR(L)

Following the calculation of DNRATE the nitrate pool in the layer is updated with appropriate
checks to ensure that a minimum concentration of nitrate is retained in the layer.

SNO3(L)=SNO3(L)-DNRATE

SOIL TEMPERATURE

The soil temperature in each layer is used in the functions describing most of the major soil N
transformations. The soil temperature model used in CERES is based on that used in the EPIC
model (Williams et

al., 1984). This method is based upon some simple empiricisms and requires only two additional
inputs to those soil parameters required by the water balance and N transformation routines.
These inputs are: TAV, the annual average ambient temperature and AMP the annual amplitude
in mean monthly temperature. The method used to calculate the soil temperature at various
depths in the profile requires the determination of a damping depth (the depth at which no
diurnal variation in temperature is experienced). At depths more shallow than this, diurnal
change in temperature occurs with the greatest fluctuation happening near the surface. The
location of this damping depth (DD) is dependent upon parameters which influence the flux of
heat in the soil, notably the bulk density and the moisture content. DD is updated daily to allow
for changes in soil moisture content. Soil surface temperatures are modelled as a function of the

89
ambient temperature, the solar radiation, and the albedo. The 5-day moving average surface
temperature is used to compute the temperatures in each layer as follows:

TMA(1) = (1.0-ALBEDO) * (TEMPM + (TEMPMX - TEMPM) *

SQRT(SOLRAD * 0.03)) + ALBEDO * TMA(1)

where:

TMA(1) = Daily surface temperature

ALBEDO = The albedo of the soil surface and is an input variable for bare soils. As the crop
canopy develops ALBEDO becomes a function of the leaf area. These calculations of albedo are
performed in the water balance routine as they are a fundamental component of the evaporation
model.

SOLRAD = Solar radiation in MJ/square metre.

TEMPMX,TEMPM = Daily maximum and mean temperature C, respectively.

The long-term average daily ambient temperature (TA) for the current day of the year can be
estimated from TAV and AMP.

TA = TAV + AMP * COS(ALX)/2.0

ALX is a variable (in units of radians) to relate the current day of the year (XI) to the time of the
hottest day of the year (HDAY). In the northern hemisphere this is assumed to be day 200 and in
the southern hemisphere day 20.

ALX = (XI - HDAY) * 0.0174

The coefficient 0.0174 is 1/365 days multiplied by 2 radians. Deviations in the actual dates of
the hottest day of the year in lower latitudes are of little importance since the volumes of AMP
will be small and hence TA will approximate TAV. The departure (DT) of the moving average
temperature from TA is used in the calculation of the soil temperature in each layer (ST(L)) as
follows:

ST(L) = TAV + (AMP/2.0 * COS(ALX + ZD) + DT * EXP(ZD)

Where ZD = depth of layer L/current day's damping depth.

90
PLANT CRITICAL N CONCENTRATIONS AND N DEFICIT FACTORS

Plant growth is greatly affected by the supply of N. Typically the supply of N to plants at the
beginning of the season is often relatively high and becomes lower as the plant reaches maturity.
The concentration of N in plant tissues also changes as the plant ages. During early growth, N
concentrations are usually high due to synthesis of large amounts of organic N compounds
required by the biochemical processes constituting photosynthesis and growth. As the plant ages,
less of this new material is required and export from old tissues to new tissues occurs lowering
the whole plant N concentration. At any point in time there exists a critical N concentration in
the plant tissue below which growth will be reduced. These concentrations are determined as a
function of crop ontogenetic age and are used within the model as part of the procedure to
simulate the effects of N deficiency. The model's critical concentration functions are based upon
the often used Zadoks' growth scale (Zadoks et al. 1974). Zadoks' growth scale is a decimal
index of crop development generalized for all cereals. The intervals between growth scale index
values are based on crop morphological observations and are not related to a thermal time
concept. To incorporate the Zadoks' scale, a scheme to provide a conversion between the integer
growth stages recognized by the model (ISTAGE) and a functional form of the Zadoks' scale
had to be devised. XSTAGE is a fractional growth stage which is used to determine an
approximate value for the corresponding Zadoks' stage (ZSTAGE). The conversions were
performed using several functions which are tabulated below (Table 1). The functions are
located in subroutine NFACTO.

Table 1. Functions Used for Converting From Fractional Growth Stage (XSTAGE) to
Zadoks'Growth Stage (ZSTAGE)

Morphological Stage XSTAGE Range Function

Emergence to terminal spikelet 0.0-2.0 ZSTAGE = XSTAGE

Terminal spikelet to booting 2.0-3.0 ZSTAGE = 2.0 + 2.0*(XSTAGE-2.0)

Booting to ear emergence 3.0-4.0 ZSTAGE = 4.0 + 1.7*(XSTAGE-3.0)

Ear emergence to anthesis 4.0-4.4 ZSTAGE = 5.7 + 0.8*(XSTAGE-4.0)

Anthesis to maturity 4.4-6.0 ZSTAGE = 6.02 + 1.86*(XSTAGE-4.4)

91
To develop appropriate relationships for critical N concentrations in wheat, published data from
field experiments that met the following criteria were assembled:

1. Experiments had a series of N rates with sufficient range to define optimal or near-optimal
growth patterns.

2. Experiments were considered to have been conducted under conditions where the potential
effects of other interacting factors (e.g., heat stress, moisture stress, frost, supply of other
nutrients, etc.) were minimized.

3. Plant tops N concentration was reported at several times during the growing season.

4. The growth stage or phenological age of the crop was reported for the times of plant sampling.

In some cases, critical concentrations were defined by the authors and where appropriate were
adopted. In two studies only one N rate was used but was described as being an optimal rate by
the authors. Data were drawn from the following sources (Table 2) representing a diversity of
wheat genotypes and wheat-growing environments.

Table 2. Data Sources Used for Determination of Critical N Concentration Relationships

Author Spring or Winter Wheat Location

Engel and Zubriski (1982) Spring North Dakota

Campbell et al. (1977a) Spring Canada

Wagger et al. (1981) Winter Kansas

Leitch and Vaidanathan (1983) Winter U.K.

Wagger (1983) Winter Kansas

Waldren and Flowerday (1979) Winter Nebraska (?)

Page et al. (1977) Winter U.K.

Alessi et al. (1979) Spring North Dakota

Mugwira and Bishnoi (1980) Winter Alabama

92
Boatwright and Haas (1961) Spring North Dakota

Gasser and Thorburn (1972) Spring U.K.

Bhargava and Motiramani (1967) Spring Australia

Walia et al. (1980) Spring India

McNeal et al. (1968) Spring Montana

Spratt and Gasser (1970) Spring U.K.

From these data, relationships defining critical N concentration as a function of Zadoks' growth
stage were determined. The critical N concentration was defined as the N concentration in the
plant tissues at optimal or near optimal growth (as defined by biomass, yield or leaf area from
the response data). The relationship thus determined is defined as the concentration above which
no further increases in crop growth occur and below which some effect on a growth process will
occur. Winter wheats and spring wheats were found to have different relationships (Fig. 3.7).
The differences between winter and sprin wheats may be an artifact created by the different
growing conditions of the experiments cited above. It has been difficult to characterize critical
concentrations particularly for the period of rapid growth in the spring when phenological age, N
uptake and biomass are all increasing rapidly. These relationships for the tops critical N
percentage TCNP) appear in subroutine NFACTO as a function of Zadoks' growth stage
(ZSTAGE).

For winter wheats:

TCNP = -5.0112 - 6.3507 * ZSTAGE + 14.9578 * SQRT(ZSTAGE) +

0.2238 * (ZSTAGE * ZSTAGE)

For spring wheats:

TCNP = 7.4532 - 1.7908 * ZSTAGE + 0.6093 * SQRT(ZSTAGE) +

0.0934 * ZSTAGE * ZSTAGE

Root critical N concentration (RCNP) relationships were derived from the greenhouse data of
Peterson et al. (1983) and Day et al. (1985).

93
RCNP = 2.10 - 0.14 * SQRT(ZSTAGE)

The minimum concentration of N in plant tissues as a function of plant age is seldom reported.
To formulate an appropriate relationship for use in the model, some of the minimum
concentrations reported in the above studies were used as well as those reported from an
extensive survey of N concentration in wheat crops spanning several years and locations in
South Australia by Schultz and French (1976). In the model the tops minimum concentration
(TMNC) is calculated as a function of model growth stage (XSTAGE):

TMNC = 2.97 - 0.455 * XSTAGE

Root critical minimum N concentration (RMNC) is used during the grain filling calculations (in
subroutine GROSUB) and is assumed to be a constant 75% of the critical concentration.

RMNC = 0.75 * RCNP

The coupling of these functions to the phenology routines thus enables critical concentrations to
be determined for any variety growing in any environment. The critical and minimum
concentrations are used to define a nitrogen factor (NFAC) which ranges from zero to slightly
above unity. NFAC is the primary mechanism used within the model to determine the effect of
N on plant growth. It is an index of deficiency relating the actual concentration (TANC) to these
critical concentrations. NFAC has a value of zero when TANC is at its minimum value of
TMNC and increases to 1.0 as concentration increases toward the critical concentration. NFAC
is calculated as:

NFAC = 1.0 - (TCNP - TANC)/(TCNP - TMNC)

Since all plant growth processes are not equally affected by N stress, a series of indices based on
NFAC are used. For photosynthetic rate (NDEF1) the index is calculated as:

NDEF1 = 0.10 + 2.0 *NFAC (NDEF<1.0)

For leaf expansion growth (NDEF2) a more sensitive factor is used:

NDEF2 = NFAC

For tillering (NDEF3) the index is calculated as:

NDEF3=NFAC*NFAC

94
For the calculation of these indices NFAC has a maximum value of 1.0. This implies that when
TANC exceeds TCNP no extra growth occurs. A fourth factor used to modify the rate of grain N
accumulation (NDEF4) is also calculated from NFAC, and can range from 0.0 to 1.5.

NDEF4 = NFAC * NFAC

These relations are depicted in Fig. 3.8. In the growth subroutine, GROSUB, the law of the
minimum is used extensively to modify rates of plant growth. For each of the major functions
(e.g., photosynthetic rate, leaf expansion rate, tiller number determination) the minimum of
several zero to unit stress indices is used to modify a potential rate for the process.
N UPTAKE

The approach used in the CERES models has been to separately calculate the components of
demand and supply and then use the lesser of these two to determine the actual rate of uptake.
Demand can be considered as having two components. First there is a "deficiency demand." This
is the amount of N required to restore the actual N concentration in the plant (TANC for tops) to
the critical concentration (TCNP for tops). Critical concentrations for shoots and roots are
defined in section 3.7. This deficiency demand can be quantified as the product of the existing
biomass and the concentration difference as below:

TNDEM = TOPWT * (TCNP - TANC)

Similarly for roots the discrepancy in concentration (difference between RCNP and RANC) is
multiplied by the root biomass (RTWT) to calculate the root N demand.

RNDEM = RTWT * (RCNP-RANC)

If luxury consumption of N has occurred such that TANC is greater than TCNP then these
demand components have negative values. If total N demand is negative then no uptake is
performed on that day. The second component of N demand is the demand for N by the new
growth. Here the assumption is made that the plant would attempt to maintain a critical N
concentration in the newly formed tissues. To calculate the new growth demand, a potential
amount of new growth is first estimated in the GROSUB subroutine. New growth is estimated
from potential photosynthesis (PCARB) and is partitioned into a potential root growth
(PGRORT) and a potential tops growth (PDWI). Partitioning between potential shoot and root
growth occurs as a function of phenological age:

PGRORT = PCARB * (60 - XSTAGE * 8)/100

95
PDWI = PCARB - PGRORT

These potential growth increments provide a mechanism for the tops actual N concentration
(TANC) to exceed TCNP. This occurs when some stress prevails and the actual growth
increment is less than the potential. New growth demand for tops (DNG) is calculated as

DNG = PDWI * TCNP

and the new growth demand for roots is calculated as

PGRORT * RCNP.

During the early stages of plant growth the new growth component of N

demand will be a large proportion of the total demand. As the crop biomass increases the
deficiency demand becomes the larger component. During grain filling, the N required by the
grain is removed from the vegetative and root pools to form a grain N pool. The resultant
lowering of concentration in these pools may lead to increased demand. The total plant N
demand (NDEM) is the sum of all of these demand components. Calculations of soil supply of N
are on a per hectare basis which necessitates recalculation of the per plant demand into a per
hectare demand (ANDEM).

ANDEM = NDEM * PLANTS * 10.0

To calculate the potential supply of N to the crop, zero to unity availability factors for each of
nitrate (FNO3) and ammonium (FNH4) are calculated from the soil concentrations of the
respective ions:

FNO3 = 1.0 - EXP(-0.0275 * NO3(L))

FNH4 = 1.0 - EXP(-0.025 * NH4(L))

The coefficients used in these two functions, obtained by trial and error, were found to be
appropriate over a range of data sets. The greater mobility of nitrate ions in soil is reflected by
the larger coefficient (0.0275) in these equations. A zero to unity soil water factor (SMDFR)
which reduces potential uptake is calculated as a function of the relative availability of soil
water:

SMDFR = (SW(L) - LL(L)/ESW(L)

96
To account for increased anaerobiosis and declining root function at moisture contents above the
drained upper limit, SMDFR is reduced as

saturation is approached.

IF(SW(L) . GT . DUL(L))SMDFR = 1.0 - (SW(L) - DUL(L))/(SAT(L) - DUL(L))

The maximum potential N uptake from a layer may be calculated as a function of the maximum
uptake per unit length of root and the total amount of root present in the layer. The first of these
is a temporary variable (RFAC) which integrates the effects of root length density (RLV(L)), the
soil water factor described above, and the depth of the layer:

RFAC = RLV(L) * SMDFR * SMDRF * DLAYR(L) * 100.0

The second of these equations incorporates the ion concentration effect (FNO3) and the
maximum uptake per unit length of root (0.009 kg N/ha cm root) to yield a potential uptake of
nitrate from the layer (RNO3U(L)).

RNO3U(L) = RFAC * FNO3 * 0.009

(RNO3U(L)) is thus the potential uptake of nitrate from layer L in kg N/ha constrained by the
availability of water, the root length density and the concentration of nitrate. Initial estimates for
the maximum uptake per unit length of root coefficient were obtained from the maize root data
of Warncke and Barber (1974). This estimate was the subject of continuing modification during
early model development. The value reported here appears to be appropriate across a broad
range of data sets. The effect of each of these parameters on determining potential uptake can be
seen in Fig. 3.9. A similar function is employed to calculate the potential uptake of ammonium
(RNH4U(L)).

RNH4U(L) = RFAC * FNH4 * 0.009

Potential N uptake from the whole profile (TRNU) is the sum of RNO3U(L) and RNH4U(L)
from all soil layers where roots occur. Thus TRNU represents an integrated value which is
sensitive to (a) rooting density, (b) the concentration of the two ionic species, and (c) their ease
of extraction as a function of the soil water status of the different layers. This method of
determining potential uptake enables the common condition, where N is concentrated in the
upper layers of the profile, where most of the roots are present and where a nutritional drought
due to shortage of water in these upper layers may occur, to be simulated. This can occur when

97
the crops demand for water is satisfied from soil water located deeper in the profile but where
there may be little N present. If the potential N supply from the whole profile (TRNU) is greater
than the crop N demand (ANDEM) an N uptake factor (NUF) is calculated and used to reduce
the N uptake from each layer to the level of demand.

NUF = ANDEM/TRNU

This could occur when plants are young and have a high N supply. If the demand is greater than
the supply then NUF has a value of 1.0. When NUF is less than 1.0, uptake from each layer is
reduced as follows:

UNO3 = RNO3U(L) * NUF

UNH4 = RNH4U(L) * NUF

Following these calculations the soil mineral N pools can be updated for the actual uptake which
has occurred.

SNO3(L) = SNO3(L) - UNO3

SNH4(L) = SNH4(L) - UNH4

Under conditions of luxury N uptake (TANC > TCNP) exudation of organic N compounds can
occur. Rovira (1969) found changes in the shoot environment which cause more rapid growth
can increase exudation. Bowen (1969) reported that N deficiency can cause exudation to
decrease. In the CERES-N model this exuded N is added to the fresh organic N pool (FON(L))
and can be mineralized and subsequently made available to the plant again. The amount of N
which can be lost from the plant in this manner is calculated as 5% of the N contained in the
roots/day. These losses are distributed to the FON(L)) pool according to the differing root length
densities present in each layer as a proportion of the total root length.

IF(TANC . GT . TCNP)RNLOSS = RANC * RTWT * 0.05 * PLANTS * RLV(L)/TRLV

Following uptake, concentrations of N in each of the shoots and roots are updated. To do this
TRNU is converted from kg N/ha to a g N/plant basis.

TRNU = TRNU/(PLANTS * 10.0)

98
The proportion of the total plant demand (NDEM) arising from shoots (TNDEM) and roots
(RNDEM) and the total root N loss (TRNLOS) are used to calculate the changes in N content of
the shoots (DTOPSN) and roots (DROOTN).

DTOPSN = TNDEM/NDEM * TRNU - PTF * TRNLOS/(PLANTS * 10.0)

DROOTN = RNDEM/NDEM * TRNU - (1.0 - PTF) * TRNLOS/(PLANTS * 10.0)

TRNLOS is distributed over shoots and roots according to the plant top fraction (PTF) and must
also be converted from a unit area basis to a per plant basis. Shoot and root N pools (TOPSN and
ROOTN, respectively) can then be updated and new concentrations calculated:

TOPSN = TOPSN + DTOPSN

ROOTN = ROOTN + DROOTN

TANC = TOPSN/TOPWT

RANC = ROOTN/(RTWT - 0.01 * RTWT)

When updating the root concentration allowance is made for the losses in root biomass occurring
due to root exudation.

N REDISTRIBUTION DURING GRAIN GROWTH AND GRAIN N DETERMINATION

In many wheat-growing areas when the crop reaches the grain-filling stage soil supplies of N are
very low. In these cases the nitrogen requirement of the developing grains is largely satisfied by
remobilization of protein from vegetative organs. When nitrogen supply is increased, the
proportion of grain N arising from remobilization declines, and the proportion from uptake
increases (Vos 1981). Many studies (e.g., Benzian et al., 1983, Terman et al., 1969) have found
negative correlations between grain yield and grain protein concentration. Temperature and soil
moisture also affect the grain nitrogen content. When constructing the N grain-filling routines,
procedures were adopted to closely mimick those predicting grain mass (or carbon)
accumulation. In this procedure the rate of grain filling (RGFILL) (mg/day) is determined by
temperature and thermal time (DTT).

To define similar functions for the rate of grain N accumulation (RGNFIL) (in micrograms per
kernel per degree C day), the controlled environment studies of Sofield et al. (1977), Vos (1981)
and Bhullar and Jenner (1985) were used. These studies examined various cultivars over a range
99
of temperature conditions and other treatments. The relationship which best described these
studies and mimicked the grain mass accumulation functions was:

RGNFIL = 4.8297 - 3.2488 * DTT + 0.2503 *(TEMPMX - TEMPMN) +

4.3067 * TEMPM

and when the mean temperature is less than 10

RGNFIL = 0.483 * TEMPM

Where TEMPMX, TEMPMN, TEMPM are the maximum, minimum, and mean temperatures
(C), respectively. A whole plant grain N sink (NSINK) can then be determined in similar manner
to GROGRN.

NSINK = RGNFIL * GPP * 1.E-6 (g N/plant)

Since N stress will affect the rate at which plant tissues can mobilize N and supply it to the
grain, an N stress factor NDEF4 from subroutine NFACTO is also introduced.

NSINK = NSINK * NDEF4

If N is present in the plant vegetative tissues (TANC greater than TCNP) the size of the sink is
increased. If there is no grain N demand (NSINK = 0) on a day then no grain N accumulation
occurs. Two pools of N within the plant are available for translocation, a shoot pool (NPOOL1)
and a root pool (NPOOL2). These pools are determined from the N concentration (VANC or
RANC) relative to the critical concentration (VMNC or RMNC) and the biomass of the pool
(RTWT or TOPWT).

NPOOL1 = TOPWT * (VANC-VMNC)

and

NPOOL2 = RTWT * (RANC-RMNC)

Not all of the N contained within these pools can be immediately mobilized. The fraction of
these pools which is labile will depend on the N status of the plant. this fraction (XNF) is
calculated by considering the N stress index NDEF2 used for vegetative growth and senescence.

XNF = 0.15 + 0.2 * NDEF2

100
The labile fraction will range between 15% and 35% of each of the pools depending on the plant
N status. The labile poos can be calculated as:

For tops:

TNLAB = XNF * NPOOL1

and roots:

RNLAB = XNF * NPOOL2

The total N available for translocation (NPOOL) is the sum of these two labile pools. When
NPOOL is not sufficient to supply the grain N demand (NSINK), NSINK is reduced to NPOOL.
If NSINK is greater than that which can be supplied by the tops (TNLAB), then TNLAB is
removed from TOPSN and the remaining NSINK which must come from the root pool
(RNOUT) is calculated. If (NSINK.GT.TNLAB) Then

TOPSN = TOPSN - TNLAB

RNOUT = NSINK - TNALB

TNLAB = 0

ROOTN = ROOTN - RNOUT

When NSINK is less than TNLAB it can be totally satisfied from the shoot pool and the root
pool need not be modified.

TOPSN = TOPSN – NSINK

Following the removal of N from shoot and root pools the simulated tissue concentrations
(VANC and RANC) are updated. The total amount of N contained in the grain can then be
accumulated.

GRAINN = GRAINN + NSINK

The grain nitrogen concentration will vary daily but is only calculated at the end of the simulatin
run (in subroutine PHENOL) as:

GNP = GRAINN/GRNWT

101
These procedures together with the remainder of the growth routine and the N deficiency indices
can provide several pathways by which N stress during grain filling can affect grain yield and
grain protein content. First, as N is removed from the vegetative tissues NFAC will become
lower. This will in turn lower NDEF4 and lower the sink size for N thus providing for the
capability of reduced grain N concentration. Lowering NFAC will also lower NDEF1 which will
cause the rate of crop photosynthesis to fall thus lowering the assimilate available for grain
filling. A declining NFAC will also speed the rate of senescence which will reduce the leaf area
available for photosynthesis. Different temperature regimes during grain filling will also affect
the final grain N concentration since the function for RGNFIL is more sensitive to temperature
than RGFILL. Soil water stress during grain filling can also increase the grain N concentration
since SWDF1 will reduce photosynthesis, lowering assimilate availability and thus not diluting
grain N as much as would occur in an unstressed crop.

102
 Agro-Hydrological Modelling
B. Bapuji Rao
Principal Scientist (Agromet)
Central Research Institute for Dryland Agriculture
Santoshnagar, HYDEABAD – 500 059

Introduction
Dryland agriculture still remains the backbone of Indian agliculture, as large areas
of cultivated land are rainfed, which contribute about 42 per cent to country's food
basket. Characterization and understanding of the environment is imperative to dryland
agricultural research. To formulate judicious soil and crop management practices for
varied dryland conditions, the crop growth processes under stress conditions are to be
properly understood. The information on moisture and nutrient uptake pattern under
moisture stress condition is presently incomplete. The studies on these lines will
unravel many of the unanswered questions that would lead to a better understanding of
nutrient and moisture interactions. Moreover, information on moisture deficits in
different periods is very helpful in crop planning. The efforts are on for some years to
develop models that would predict the crop water use pattern. An attempt is made here
to present a comprehensive review of the currently available agro-hydrological models
and to illustrate the basics behind the development of such models.

Some of the currently available agrohydrological/soil water balance models are

listed in Table 1. They differ in complexity, operation and purpose. The models also
differ in the use of theoretical or empirical descriptions of the process in the model.
These can be broadly classified as:

(a) Agroclimatic models - These are usually single layer models used for regional
characterization of environments for water availability.
(b) Management models - In these models the soil profile is divided into two or three
layers. Information generated on soil moisture availability is used for soil and crop
management.
(c) Physical process models - The soil profile is divided into many layers for studying
the flow processes more precisely.

103
Table 1. List of some agro-hydrological models
Name of the model Reference
Versatile Soil Moisture Budget (VSMB) Baler et al. (1979)
WATER Burt et al. (1980)
Unnamed Belmans et al.(1983)
Unnamed Brisson et al.(1992)
Unnamed Cordery and Graham et al.(1989)
SMEP Edey (1980)
Unnamed Greacem and Hignett.(1984)
SWATRE Feddes et al(1976)
Unnamed Jagtap and Jones (1989)
SWACRO Feddes et al(1984)
Unnamed Hansen (1975)
Unnamed Holst and Madsen (1984)
Unnamed Jones and Smajstrla (1980)
Unnamed Lascano and van Bavel (1986)
Unnamed Norman and Campbell (1983)
Unnamed Place and Brown (1987)
Unnamed Rama Prasad (1984)
PLANTGRO Retta and Hanks (1980)
Unnamed Robinson and Hubbard (1990)
SPAW Saxton et al(1974)
Unnamed Seliorio and Brown (1979)
Unnamed Stockle and Campbell (1985)
SIMBAL Stuff et al(1975)
EMWATBAL Van Bavel and Lascano (1987)
Unnamed Victor et al(1988)
Unnamed Visser (1974)
Unnamed Vossen (1990)
Unnamed Wright et al(1994)

1.1 Model components

The basic components of all the soil water balance models are presented in Fig. 1.
The components are not independent but they are interrelated. For instance, amount
of runoff depends partially on the rainfall intensity, hydraulic properties of the soil,
104
and surface water content.
Precipitation or irrigation is usually measured and the other components of water
balance are estimated, in the same units.

1.2 Surface runoff and infiltration

Whenever rainfall occurs, some amount of water often runs off and becomes
unavailable to the crop. Therefore, to estimate the recharge of soil profile it is
important to estimate the amount of runoff that occurs with each rainstorm. Various
approaches have been made to estimate this component in water balance models.
Some models do not consider runoff and deep drainage separately. Any amount of
water input into the soil after the soil profile is full to its maximum storage capacity
(field capacity) is considered as water loss (runoff + deep drainage). However, in
some models the runoff or infiltration is estimated to calculate profile recharge.
Kanemasu et al. (1978) calculated effective precipitation (Pe) and runoff as follows:
Pe = R 0.75 when R> 25.4 mm;
Pe = R when R≤ 25.4 mm
Therefore, runoff = R-Pe (1.1)
Baier et al. (1978) used a simplified relationship between moisture content in the
topsoil zone and daily total precipitation to estimate infiltration into the soil. On days
with rainfall 25.4 mm, the total amount of rainfall is considered to infiltrate into the soil.
On days with rainfall 25.4 mm. the amount of infiltration (Infl) into the soil is less than
the daily rainfall, because it is limited by run- off as a function of rainfall and the
moisture already in the top zone of soil, and is computed as
Infl = 0.9177 + 1.811 log RRi – 0.97
[S j(i-1)Cj] log RRI (1.2)
where,
RRi = rainfall on day i
S j(i-1) = soil moisture in the jth zone on day i-1
Cj = available water capacity of the jth zone
j=1
The remainder of the daily rainfall is assumed to be lost as runoff. Dale et al. (1982)
employed the same relation in their SIMBAL model for cropland drainage effects on soil
moisture and evapotranspiration.

1.3 Soil water recharge and drainage

105
 The soil profile in the water balance models is either considered as a single layer or
divided into discrete layers of either uniform or variable thickness. Infiltration and
redistribution of water throughout the layered soil profile is often treated in two rather
in different procedures. In simple models, the infiltrated water is freely transmitted to
lower layers by gravity or out of the profile if it was the lower most layer. The upper
limit of water for each layer is set at field capacity. When the antecedent water content
plus inflow of water exceeds field capacity of that layer then the excess water is
allocated to next lower layer. This process is repeated for all layers and excess water
from the lowest layer is considered as deep drainage. In this method upward movement
or redistribution of water is not allowed unless it is an added feature.
In the Versatile Soil Moisture Budget (VSMB) of Baier et al (1979), the
partitioning of infiltrated water to each zone is simulated by the following function.
Inflij = { 1- [Sj(i-j) / Cj ) b} {Infli - Inflin} (1.3) where, Inflij = new infiltration into each
zone of soil Sj{i-1) == soil moisture in the jth zone on day i-1 Cj = available water
capacity of the jth zone b = percolation coefficient ranging from 0 to 1.
This equation is applied only when the ratio of soil moisture to capacity in any zone
is less than 0.9. The amount of water that can infiltrate into and remain in each zone
can not exceed the deficit (DEF) for that zone (j) and day(i). The DEF is given by
DEFji = Cj – S j(i-l) + AE ji (1.4)

In addition, the VSMB assumes that the amount of water that can be budgeted to
th
the j zone can never exceed what remains from the total water infiltrated after water
has been budgeted to zones 1 to j-l. If, after all zones have been recharged, there is still
infiltration water remaining, then this water is allocated to subsurface drainage. This
means that the infiltration is distributed over the zones as a function of the amount of
infiltration, the relative moisture content in each zone, and the percolation coefficient
(b). The percolation coefficient is the fraction of water infiltrating to the next zone. For
b=O, the water content of each zone must reach field capacity before the remaining water
infiltrates into the next zone. For b= 1, a fraction of the infiltration water percolate to
the next zone before field capacity is reached, depending on the moisture content in the
upper zone. This feature of the infiltration equation was found to be useful in heavy
textured soils.
In WTGROWS - a wheat growth model of Aggarwal et al. (1994), the amount of
water available either by rainfall or irrigation after deducting runoff is allocated to

106
various soil layers starting from the surface layer. In this model, inter layer fluxes of
water are considered only at the time of rainfall or irrigation and at all other times the
fluxes are considered negligible. Depending upon the amount of water available, the
layers are charged to field capacity. Water in excess of field capacity of a layer, if
available, is immediate drained to the next layer. The amount water above field capacity
of the bottom layer is drained out of the profile and is not available for crop use. Similar
procedure was adopted for profile recharge COTTAM model of Jackson et al. (1990).
In tlie other method of profile recharge, it is customary to use Darcy's unsaturated
flow equation, in which each layer assumed to be uniform in moisture content, capillary
pressure, and unsaturated conductivity. Mathematical solutions vary from the simple
finite difference with large time steps to finite element with near analytical results.
There are several models considering soil water movement in response to pressure-
head gradients in accordance to the Darcy and continuity equations, for example,
SWACROP of feddes et al (1978), SPAW of Saxton et al. (1974) and Rama Prasad
(1984). This treatment of water flow can be used to represent nearly all situations
including upward or downward flow between layers, widely varying characteristics
within the profile, time distribution of infiltration and redistribution among layers,
water tables and plant water withdrawal. But this approach requires specifications of
the soil water retention and hydraulic-conductivity curves, upper boundary conditions
of precipitation and potential evapotranspiration and a lower boundary condition
appropriate for the site under consideration. The choice of which soil water movement
calculation to employ depends upon the accuracy required. For readily drained soils
where withdrawal of water by the profile development and casual accuracy is
required, the free flow procedure would suffice.

Potential evapotranspiration

The estimation of potential evapotranspiration (PET) is essential to know the

evaporating power of the environment so crop evapotranspiration (ET) could be
estimated. Soil moisture models use different methods for estimating PET. In
IBSNAT crop models, PET is calculated using an equilibrium evaporation concept
developed from the Priestley and Taylor (1972) model. The equation calculates the
approximate daytime net radiation and equilibrium evaporation, assuming that
stomata are closed at night and no ET occurs then. Equilibrium evaporation, Eeq, is
computed in CERES - Maize model of John’s et al. (1986) as
Eeq = Rs( 4.88* 10-3 – 4.37 * 10 -3* α) (T+29) (1.5)

107
where,
Rs = solar radiation, MJ m-2 day-1
α = albedo of crop soil surface
T = average daily temperature (°C), estimated as
T = 0.6 * Tmax + 0.4 * Tmin (1.6)
Tmax = daily maximum temperature (°C)
Tmin = daily minimum temperature (°C)
The PET is then computed by
PET = 1.1 * Eeq if 5< Tmax ≤ 35°C (1.7)
PET = Eeq [1.1 + 0.05 (Tmax -35} if Tmax> 35°C (1.8)
PET = Eeq *0.01 e [0.18 (Tmax+20)]
If Tmax >5° (1.9)

The PET is calculated as the equilibrium evaporation times 1.1 to account for the
effects of unsaturated air. The multiplier is increased above 1.1 to allow for advection
effect when the maximum temperature is greater than 35°C, and reduced for
temperatures below 0°C to account for the influence of cold temperatures on stomatal
closure.

The versatile soil moisture budget (VSMB) of Baier et al. (1979) uses regression
formulae for estimating PET from various combinations of available meteorological
data. This is because cf the development of these relations, for most of the Canadian
sites where the VSMB was tested, from the historical weather data. For the sites
where these regression equations are not available, the VSMB calculates PET from
the daily maximum (T max) and minimum (T min) temperatures and radiation at the top
of the atmosphere (Qa) as
PET = 0.0034 [Tmax*0.928] + 0.933 (Tmax – Tmin) + 0.0486 Qa – 87.03] (1.10)

108
 The Penman's formula has been used to compute PET in several models like
WTGROWS of Aggarwal et al. (1994) and WATER of Burt et al. (1980). In
determining the PET, primarily the choice of method should be based on the type of
meteorological data available.

1.5 Actual evapotranspiration

The actual evapotranspiration (AET) is generally estimated by two approaches in

water balance models as

(a) using crop coefficients and soil dryness curves

(b) separating: evapotranspiration into evaporation and transpiration.

1.5.1 Crop coeffICients

Crop coefficients are generally empirical ratios of crop ET (ETc) to some

reference ET (PET) that have been derived from experimental data according to the
relationship
Kc = ETc/PET (1.11)
where,
Kc = crop ceofficient for a particular crop for a given growth phase and soil moisture condition –
dimensionless
ETc = daily crop ET (mm)
PET = daily reference ET (mm)
The reference ET characterizes the evaporative demand determined by meteorological
conditions and a standard crop surface and Kc indicates the relative ability of a specific
crop - soil surface to meet that demand. Since PET is affected by many variables, it can
not be simplified for all climate and crop situations. This is because of the effects of
relative leaf area and the morphological and physiological characteristics of the
reference crop canopy on the energy exchange and aerodynamic diffusion processes
within the atmosphere over a field. Different methods have been proposed for
estimating PET for either grass or alfalfa with corresponding crop coefficients
(Doorenbos and Pruitt, 1979: Wright, 1981).

Normally, the crop coefficient includes the effects of evaporation from both plant
and soil surfaces and is dependent upon available soil water within the root zone and
the wetness of the exposed soil surface. Soil water depletion data obtained by
gravimetric or neutron probe methods and lysimetric data can be used to obtain Kc
values.
109
Table 2. Crop coefficients (Kc) values for different crops and weather conditions (after
Doorenbos and Pruitt, 1979)
Crop Crop stage Humidity >70 % <20%
windspeed 0-5 5-8 0-5 5.8
(m/sec)
Wheat 3 1.05 1.10 1.15 1.20
4 0.25 0.25 0.20 0.20
Maize 3 1.05 1.10 1.15 1.20
4 0.55 0.55 0.60 0.60
Cotton 3 1.05 1.15 1.20 1.25
4 0.65 0.65 0.65 0.70
Millet 3 1.00 1.05 1.10 1.15
4 0.30 0.30 0.25 0.25
Groundnut 3 0.95 1.00 1.05 1.10
4 0.55 0.55 0.60 0.60
Sorghum 3 1.00 1.05 1.10 1.15
4 0.50 0.50 0.55 0.55
Soybean 3 1.00 1.05 1.10 1.15
4 0.45 0.45 0.45 0.45
Sunflower 3 1.05 1.10 1.15 1.20
4 0.40 0.40 0.35 0.35
Potato 3 1.05 1.10 1.15 1.20
4 0.70 0.70 0.75 0.75
Onion (dry) 3 0.95 0.95 1.05 1.10
4 0.75 0.75 0.80 0.85
Cabbage & 3 0.95 1.00 1.05 1.10
Cauliflower 4 0.80 0.85 0.90 0.95
Carrot 3 1.00 1.05 1.10 1.15
4 0.70 0.75 0.80 0.85
Sugarbeet 3 1.05 1.10 1.15 1.20
4 0.90 0.95 1.00 1.00
Radish 3 0.80 0.80 0.85 0.90
4 0.75 0.75 0.80 0.85

The crop coefficient values for different crops are presented in Table 2. A generalized
110
crop coefficient curve proposed by Doorenbos and Pruitt (1979) is presented as Fig 2.
The Kc values for growth stages 1 and 2 for the crops in Table 2 can be obtained by
interpolation from Fig 2. A crop coefficient curve represents the seasonal variation of
the empirically derived Kc values. Once this curve has been developed for a given
location, daily crop ET can be estimated using Eq. 1. 11. The variation in time of
season of the crop coefficient in Eq. (1.11) has not proven to be generalizable because it
is often management, site and weather specific as evident from the values furnished in
Table 2. Values of Kc may be management specific as a result of planting date, plant
population and row spacing. The value may be site specific because of large-scale soil
spatial variability and may not be reproducible from one year to the next for a given
location because weather sequence are usually not reproducible from year to year. Crop
coefficients are dependent on weather because air temperature, radiation and frequency
of rainfall effect Es and Ep directly and temperatures influence the rate of crop
development. Hanks (1985) compared crop coefficients measured at Logan and Davis
for the same crop. The crop coefficient curve differed markedly for the two locations,
especially early in the conditions where the soil surface is dry. He attributed this site
specificity to the dependence of Es on the rainfall frequency and amount or irrigation
regime when plant cover is low and suggested that the crop coefficient vary from
year to year for the same reason. Wright and Jensen (1978) recognized this limitation
of the crop coefficient procedure and developed a crop coefficient curve that was
based only on conditions where the soil surface is dry. Their model accounts for
increased evaporation when the soil surface is wet and efficiently reduced the
estimation of ET of snap bean during the leaf area development. Wright (1982)
defined these modified crop coefficients, designed to represent conditions where the
soil surface is dry but water is readily available, as “basal ET crop coefficients
(Kcb)". Estimates of an adjusted crop coefficient in terms of Kcb were accomplished
through use of the following equation : .
Ka = Kcb + (1-Kcb) [1-(t/td)0.5] f(w) (1.12)
where,
Ka = adjusted crop coefficient
t = number of days after major rain or irrigation
td = usual number of days for the soil surface to dry
f(w) = relative proporation of the soil surface originally wetted
Most of the adjustment takes place in the first few days after wetting the soil. The
usefulness of this adjustment procedure may be extended mainly to arid regions where

111
evaporative conditions are relatively uniform during the season and td may indeed be
constant for a given soil. In humid regions, it may be necessary to accommodate the
unpredictable temperature and radiation conditions by considering the constant rate
stage of soil evaporation and its upper limit (Ritchie and Johnson, 1990). However, Eq.
1.12 helps to diminish the year-to-year variation of the crop coefficients caused by
varying frequency of rainfall and irrigation. Ritchie and Johnson (1990) proposed the
following relationship which incorporates the effects of temperature and leaf
appearance and expansion for the influence of varying plant population on ET as :
Ks+p = (Es + Ep)/PET (1.13)
where,
Ks+p = a daily crop coefficient based on separate calculation of Es and Ep.

The calculations of Es and Ep require prediction of daily values of LAl and a

temperature-based relation for leaf area development can accomplish this. Several
authors have developed relations to evolve Kc values as a function of thermal time or
days after emergence. Sammis et al. (1986) applied the following polynomial for
estimating Kc values of winter wheat and spring barley as
Kc= Bo + B1ΣG + B2ΣG2 + B3ΣG3 (1.14)
where,
ΣG = accumulated growing degree days starting planting
B1 = regression coefficients.
The CALEX/COTTON irrigation module program has been evaluated by Plant et
al. (1992) with the following relation for Kc as
Kc= 0.038 + 21.158 x – 56.579 x2-263.86 x3 (1.15)
where,
x = DD / 10000
DD = total elapsed growing degree days 15°C
The Kc values for spanish peanuts were determined from the third-order polynomial
(Elliott et al.. 1988) as a function of the fraction of the peanut growing season as
Kc = -1.644 + 12.05 F – 17.155 F2 + 7.499 F3 (1.16)
where,
F = fraction of the growing season (ratio of days since planting to days between planting and
harvesting)
In this case, a growing season length of 140 days was assumed when
calculations of F were made. Since this polynomial relationship gives a negative result

112
for mall values of F, a minimum value of 0.4 for Kc is suggested.
A second order polynomial equation has been employed by Idike et al. (1982) for
estimating Kc of corn as function of days after emergence as
Kc = 0.152 + 0.0164 D – 0.00012D2 (1.17)
where,
D = days after emergence of the crop
A third degree polynomial has been fitted for spring wheat, barley, canola.
sugarbeet and potatoes by Foroud et al. (1992) with a lower and upper limit of 0.1 and
1.2 for Kc respectively.

1.5.2 Soil dryness curves

Depending on the energy available in the atmosphere, the soil is able to provide all
or part of the water requirement of the plant for transpiration. This depends on soil
characteristics, crop type, crop stage and the magnitude of PET itself. As the soil
dries, its hydraulic conductivity decreases, and water moves more slowly towards the
roots. Only in certain circumstances AET becomes equal to the PET, and normally the
former is a fraction of the later. When the moisture tension becomes too high, AET
becomes less than PET. This break - off tension is higher for lower values of PET and
vice - versa. It is high for some crops like horsegram (drought - resistant) and low for
sugarcane. For a given crop it is higher during the initial and late stages than during
mid-season stage (Rama Prasad, 1984). In the range of plant available water i.e.,
amount of water from field capacity (FC) to permanent wilting point (PWP), the
amount of available soil water gradually decreases to become zero when PWP is
reached. Contradictory view points exist on the availability of soil moisture and the
versatile soil moisture budget of Baier et al. (1979) who have used 8 types of
relationships between the available water for plants and the ratio AET /PET. From
these, an adjustment factor 'z' was proposed for different types of soil dryness curves as :
Z = [AET / PET] / [AW / AWC] (1.18)
where,
AW = available water in the zone concerned on a given day
AWC = available water capacity of that zone

The AET on a given day can thus be computed as

AET = [AW/AWC] * Z * Kc * PET (1.19)
1.5.3 Separating PET into evaporation and transpiration

113
 Several types of models are available for calculating Es and Ep. Because of their
differences in their purposes and organization, they can be broadly categorized according
to Addiscott and Wagenet (1985) into (a) deterministic or stochastic (b) mechanistic or
functional and (c) rate or capacity types.
The deterministic models produce a unique outcome for a given set of events.
However, due to the spatial variability of the mediating processes there will be a certain
degree of uncertainty associated with the results. Stochastic models have been developed
to accommodate this spatial variability and to quantify the degree of uncertainty.
Stochastic models produce an uncertain outcome because they include one or more
parameters that are random variables with an associated probability distribution. But
stochastic models have been applied little in modelling ET (Ritchie and Johnson, 1990).
Most models used for estimating Es and Ep are deterministic and can be further
categorized as mechanistic or functional. The mechanistic models are based on dynamic
rate concepts and incorporate basic mechanisms of processes such as Darcy’s law or
Fourier’s law and the appropriate continuity equations for water and heat fluxes
respectively. Functional models are usually based on capacity factors and treat processes
in a more simplified manner, reducing the amount of input required. Mechanistic models
are useful primarily as research tools for better understanding of an integrated system,
and are usually not used by non-authors due to their complexity. On the other hand, the
functional models have modest input requirements making them useful for management
purposes. Because of their simplicity, functional models are more widely used and
independently validated than mechanistic models.

The most important causes ofunproductive loss of water is direct evaporation from the
soil surface and especially so under arid and semi-arid conditions where deep drainage
can generally be ignored. Thus, the ratio of soil evaporation to transpiration is of decisive
importance for overall water use efficiency. Evapotranspiration can be divided into two
parts as follows
(a) firstly, under fallow conditions evaporation proceeds at rate depending upon soil type,

frequency of wetting and evaporative demand; and

(b) secondly, under a cropped situation evaporation and transpiration proceed and in turn

depend on the soil type, evaporative demand, available water in the root zone and the
type of crop cover at different stages of crop gtowth.
1.5.2. 1 Soil evaporation (Es)
There are several mechanistic models available on Es in which general equation of
water flow is used. Some of these models predict evaporative losses of water from a bare

114
soil (Gardner and Gardner, 1969: van Bavel and Hillel. 1976; Hillel and Talpaz, 1977;
Lascano and van Bavel, 1986). Some models facilitate the separate calculation of Es and
Ep in the presence of a crop (Feddes et aI., 1976; Norman and Campbell, 1983; Huck
and Hillel, 1983; Lascano et al., 1987). Functional models are less evident in the
literature and some models that have been used to calculate E s and E p separately are
the models of Ritchie (1972), Hanks (1974), Kanemasu et al. (1976) and Tanner and
Jury (1976). Because of the complexity in applying mechanistic models, as explained
in section 1.5.2, the review presented here mostly deals with functional models.

The rate of evaporation from the soil can be grouped into several stages. During the
first stage, which may be lost for only one to three days in mid-summer, the rate of
evaporation is controlled by heat energy input and is about 90 percent of PET (Jensen
et aI., 1990). The duration of first stage is influenced by the rate of evaporation, soil
depth and hydraulic properties of the soil (Gardner and Hillel, 1962). By noting the
changes in albedo, the transition from first to second stage of drying can be identified
(Jackson et al., 1976). Immediately after wetting, the evaporation from a wet bare soil
is approximately same as that from a free water surface, the duration of which is again
dependent on soil type and evaporative demand of the atmosphere. The period is
shortened under coarse textured soils. This relates to the amount of water retained in
the top 10 cm of soil layer (Reddy, 1993).
During the second or falling stage, the surface has begun to dry and evaporation is
occurring below the soil surface. Water vapor reaches the surface by molecular
diffusion and mass flow caused due to the fluctuations in soil air pressure. The dry
surface soil greatly influences the effective internal resistance. After the mulch has
been formed then the rate of evaporation is less than PET and the rate is controlled by
soil characteristics like hydraulic conductivity but not by the meteorological
conditions. During the second stage the cumulative evaporation tends to increase with
the square root of time for a given soil and evaporation potential as
Es dt = K (t-t1) 0.5 if t > t1
where,
K = empirical constant for a given soil that depends on the soil characteristics and
water content (Black et al. 1969)
t1 = time at which the falling stage begins.

The value of K can be determined experimentally from cumulative evaporation data

for a single drying cycle of a given soil. Several direct measurements of the
coefficient K in a diversity of soils have consistently resulted in values of about 3.5

115
mm day-0.5. Mason and Smith (1981) used a value of 5 in their model. Thus, the
cumulative loss of water after an irrigation or rain can be approximated by
Ec = 0.9 PET dt for t t1 and Ec = 0.9 PET dt + K (t-t1)0.5 for t t1
where,
Ec = cumulative evaporation
t1 = time since evaporation began

Ritchie (1972) summarized Ec at tl and K value for several soils and the values are
reproduced in Table 3.
Table 3 . Typical coefficients for second stage evporation
Soil Ec (mm) K (mm day -0.5)
Adelanto clay loam 12 5.08
Yolo loam 9 4.04
Houston black clay 6 3.50
Plain field sand 6 3.34

PLANTGRO model of .Retta and Hanks (1977) computes Es on the assumption

that E s occurs only from the top layer, usually of 20 to 30 em in thickness, as
E = PET t-0.5 (1.23)
where,
t = time in days since last irrigation or rain.

It also allows the top layer to loose moisture below wilting point to the airdrying
moisture content. The amount of water that may be lost by evaporation after the soil
has reached its permanent wilting water content is taken as equal to the amount of
water lost when the first 10 cm of the top layer is dried to air-drying. Thus during the
constant rate stage, the evaporation occurs at the potential rate until the upper limit of
stage 1 evaporation (U) is reached (Ritchie and Johnson, 1990). This U is reached
more rapidly under condition of high PET than under low PET. Ritchie and Johnson
(1990) reported value of U to be about 5 mm in sands and heavy shrinking clays to
about 14 mm in clay loams. Mason and Smith (1981) assumed a value of 7 mm.

However, contradictory view point exists on the second-stage evaporation rates.

Arora et al. (1987) used the following empirical relation for sandy loam soils
E = PET t-0.30 (1.24)

It was also assumed that the top 30 cm soil layer contributed towards soil
evaporation. Hill et a1. (1979) used the following relation to estimate Es as

116
Es = Ep / 2 (2t-1) (1.25) and Ep = Ks * PET (1.26)
where,
Ep = potential soil evaporation
Ks = soil evaporation factor which depends on the value of Kc.
t = time in days since the last soil surface wetting.
It is subject to the constraint that the surface soil can not be drier than air-dry. The
top 10 cm soil is assumed to be dried by evaporation and transpiration to the wilting
point and then by evaporation only to air dry.

However, Ritchie's (1972) model is the most frequently used model for E s. This has
been used by Cull et al (1981), Mason and Smith (1981), Sharpley arid Williams
(1990) and Jain and Murthy (1985).
The calculation of E s in the Ritchie's (1972) model requires prediction of the net
radiation through the canopy to the soil surface. Ritchie and Burnett (1971) quantified
the influence of partial cover on ET and found that LAI of sorghum and cotton were
more generally related to plant evaporation (ET) as fractions of PET than ground
cover or plant dry weight. The empiricism used to estimate ET from crops with an
adequate supply of water in the root zone usually make ET a function of PET. LAI or
plant cover.
In the presence of canopy, the fraction of energy (Rso) supplied to the soil
surface depends on crop cover or leaf area index (LAI) and is given by
Rso = Rns / Rn = exp (-0.4 LAI) for LAI <2.7 (1.27)
where,
Rns & Rn = 24 hour net radiation at soil surface and above the crop canopy, respectively.

Thus, the soil evaporation during the constant rate stage (stage 1) is calculated
using a Priestley - Taylor type equation (Ritchie, 1974);

(1.28)

where,
qn = 0.92 + 0.4 [Rns / Rn] for LAI < 2.7 (1.29)
qn = 1 for LAI >2.7 (1.30)
Some of the functional relations reported in the literature are presented here.
CERES-Maize model of Jones et al. (1986) computes potential soil evaporation as
Esp = PET [1.0 -.0.43 (LAI)] for
LAI <1 (1.31)
-0.4 (LAI)
and Esp = (PET /1.1) e for

117
 LAI <1 (1.32)

In this case, a modified Preistley-Taylor equation is used to calculate daily values of

PET although other equations could be used. The potential rates for PET and Esp are
equivalent in this example when applied to bare soil conditions (i.e.: LAI = 0).

For wheat crop of North-Western part of India, Jain and Murthy (1985) computed
Esp as '

Sammis et al.(1986) used the following relation for winter wheat and spring barley to
compute daily soil evaporation as
Es = PET * e -0.623 LAI
In a water balance model that was used to calculate dry matter yield of wheat, Hanks and
Puckridge (1980) computed potential soil evaporation as a function of LAI and dry
matter production (DM) in g m-2 as
Esp = [PET - Tp] [1 – DM/2000] (1.35)
where,

Tp = 0.9 PET if LAI>3 (1.36)

and Tp = 0.9 PET (LAI/3) if LAI <3 (1.37)

In the Soil Water Leaf. Extension of Winter Wheat and Wheat Growth [SWLEWW-
WTGRO) model of Farshi et al. (1987) for the values of ground cover (Gc) smaller than
0.45, it was assumed that PET f winter wheat was equal to ETo, the PET of a full grown
grass cover. The PET was separated into potential transpiration and evaporation on the
basis of predicted values of leaf area index as
Gc = 1 - exp (-0.6 LAl) (1.38)
Ep = PET if Gc = 0 (1.39)
Ep = 0.9 (1 - Gc) PET if Gc > 0 (l.40)

The equation (1.32) was used for LAI up to 3.5. For LAI >3.5, it was assumed that
the crop covers the ground surface completely, consequently Gc was put equal to 1.

In an attempt to simulate water content under barley using simple empirical

approach, AI Khafaf et al. (1989) estimated bare soil evaporation using the expression
Es = a (t)b - a (t-l)b (1.41)
118
 where,

a,b = empirical coefficients depending on soil type

t = time after irrigation [days).

The potential soil evaporation was calculated as

Esp = PET if Fs < 0.08 (1.42)

that is, if PET is not higher than the value of 'a' in Eq.(1.41), otherwise Esp = 7.24 mm
day-1. Further, they used
Esp = (1.21 - 2.343 Fs) PET
if 0.08 < Fs < 0.48 [1.43)
,Esp = 0.08 PET if 0.48 < Fs (1.44)

where, Fs is fraction of degree of shading which is the ratio of LAl at any given time to
the maximum value of LAl during the season under non-limiting water conditions.

1.5.2.2 Plant transpiration (Et)

As the case with modeling Es, there are also mechanistic and functional models for
estimating plant transpiration. Two classical examples of mechanistic models are
Penman-Monteith model (Monteith, 1973) and the EMWATBAL model (Lascano et
al, 1987). Though they differ in their, input requirement and detail but were stated to
have good theoretical bases, the former being simpler than the later and intended to
calculate E t or AET. Both these models operate on 1-hour time step and Penman-
Monteith has been used extensively to estimate AET in recent years. This equation
requires hourly input or estimates from empirical functions for humidity,
temperature, net radiation, soil heat flux, heat storage rate in the canopy air column,
transport resistance from the leaf surface to instrument height (ra), and the resistance
of the crop (rc). However, use of this relation to predict AET in applications such as
irrigation scheduling is a problem because the required meteorological inputs are
difficult to obtain. The value of the surface resistance is a complex function of many
climatological and biological factors (Monteith, 1985). All the meteorological
variables that must be known are dependent to some extent on the properties of the
vegetation. Thus, they must be estimated on the basis of previous experience or
calculated from models of the exchange processes (McNaughton and Jarvis, 1984).
The other mechanistic model, EMWATBAL calculates the water and
energy balance for both the soil surface and crop canopy. This model calculates water
evaporation and transpiration fluxes. At each time step, the model calculates and
updates values of water content and temperature for each soil layer. From these
119
values the inter-layer fluxes of water and heat below the soil surface are calculated.
At the soil surface the net radiation, latent, sensible, and soil heat fluxes are
calculated from an energy balance equation. Instead of assuming that evaporation at
the soil surface at the equilibrium rate corresponding to the net radiation flux above
that surface (Ritchie, 1972), the EMWATBAL calculates evaporation at the soil
surface by finding the surface temperature that satisfies the energy and water balance
at the soil surface by the method of van Bave1 and Hillel (1976). The model requires
information on soil water retention curve and the unsaturated hydraulic conductivity
for each soil horizon, and the number and thickness of the soil layers that determine
the geometry of the soil system. Plant inputs are the relation between leaf
conductance and leaf water potential, the root distribution as a function of soil depth
and time, and the LAI as a function of time. Weather inputs are daily total solar
radiation, daily maximum and minimum air and dew point temperatures, daily wind
speed, and the quantity of rain or irrigation as a function of time of day. In addition,
initial values of the water and temperature profiles must be specified at the start of the
simulation period. The daily weather data will be disaggregated into hourly values
using empirical relations to produce 24 estimated values from one or two measured
values.

Though EMWATBAL is a mechanistic model, it uses empirical relations also. For

example, several polynomials were used to compute short wave absorptance of the
crop (ABSC) , of the soil (ABSS) , and the view factor from soil to sky (FTSR), all as
a function of LAI, as

ABSC == 0.5809 LAI - 0.2231 LAI 2 + 0.04640 LAI3 - 0.004759

LAI4 + 0.0001875 LAI 5 (1.45)

ABSS= 0.825-0.6447LAI +0.2646LAI2 - 0.05695 LAI 3 + 0.005937

LAI4 - 0.0002355 LAI5 (I.46)
FTSR == 1.0 - 0.6780 LAI + 0.2052
LAI2 - 0.02799 LAI3 + 0.001383 LAI 4 (1.47)
Further, the LAI is estimated with third-order polynomials as a function of calendar
day number (CD). For the irrigated cotton crop, this relationship was given as

LAI = 105.6713 - 1.61088 CD + 0.008035084 CD2 -

0.0000130257 CD3 (1.48)

and for the dryland cotton crop, the relation was given as

LAI=70.93205-1.13014CD+0.005848976CD2-0.0000097607CD3 (1.49)

120
 Ritchie and Johnson (1990) have identified several problems that may reclude the
use of a mechanistic model for rediction purposes and for supporting farm decisions
such as irrigation scheduling. Mechanistic models may provide the information
needed to derive some of the empiricism upon which functional models are based.

The second type of models for estimating Et are functional models. These are quite
widely used in many crop growth models because they require few inputs, most of
which are readily obtainable.

In an earlier attempt to estimate the evaporation rates from developing cotton and
grain sorghum canopies under water non-limiting conditions, Ritchie and Burnett
(1971) estimated the potential transpiration (T p) as

T p = PET ( - 0.21 + 0.70 LAI 0.5),

0.1≤LAI≤ 2.7 (1.50)

The non-linearity of the relation between T p and LAl is stated to be the result of
two interacting factors

(a) less competition for radiation per unit of leaf area during initial stages of plant
growth and
(b) the partitioning of a large fraction of net radiation at the dry soil surface between
plant rows to sensible heat flux causing increased canopy temperature and
consequently increased T p (Ritchie and Burnett, 1971).

Upper limit of 2.7 of LAl represents minimum requirement of LAI necessary for
full cover of canopy. For crop canopies with LAl> 2.7, Tp = PET. When LAl < 0.1,
T p is considered negligible.

The T p is computed from actual (Ea) and potential soil evaporation (Esp) and
PET by Brisson et al. (1992)using the following relation

T p = (PET - E p) [β + (1 - β) Es/Espl (1.51) where, the value of β was 1.1, which

simulates a 10% increase in Tp/PET between wet soil conditions (Es=Ep) and dry soil
conditions (Es=O).

The potential transpiration for a wheat crop was estimated by Hanks and
Puckridge (1980) using the relations already explained in section 1.5.2.1 as Eq.
i (1.53) to (1.54), which are reproduced here, as T p = 0.9 PET if LAI is 3
T p = 0.9 PET (LAl/3) if LAI is 3.

Assuming that T p is largely controlled by evaporative demand and degree of

121
shading under non-limiting water conditions, Al-Khafaf et al. (1989)
estimated T p using the following functional relations
T p = 0 ifFs< 0.08 (1.52)
T p= [l - (1.21 - 2.343 Fs)J PET if 0.08
<Fs<0.48 (1.53)
T p= 0.92 PET if 0.48 < Fs (1.54)
where,
Fs = fraction of degree of shading .

In most of the soil water balance models like that of Farshi et al. (1987), Murthy
et. al (1992) and Sammis et al. (1986), the T p is calculated as
Tp = PET - E s (1.55)
where Es is calculated by one or the other method discussed in section 1.5.2.1
In CERES-Maize,the calculation of Et is through a functional model on the lines
of Ritchie (1972). The functional model for estimation of E s has been described
earlier in section 1.5.2.1. Where the soil water is non-limiting the functional model
calculates Et using the relationships
Et = PET (1.0 – e -LAI) if LAI ≤ 3 (1.56)
Et = PET if LAI > 3
and if Es + Et > PET, then
Et = PET - Es (1.58)

The conditional Eq (1.58) was felt necessary because values for Es and E t are
calculated independentIy and their sum can exceed the potential rate on a given day
because Eq (1.56) and (1.57) are for estimating Et when the soil surface is dry.
Once the potential transpiration (Tp) is computed, most of the models estimate actual
transpiration (E t) on the basis of soil water availability. As the soil dries, the
conductivity of soil to water flow decreases, thereby decreasing the uptake of water
by the root system. Actual transpiration (E t) by the crop falls below the potential
transpiration demand (T p). There are essentially two approaches to estimate Et. In the
first approach T p is decreased in proportion to the water deficit in the rooting zone.
The transpiration from sorghum or corn as observed by Ritchie (1973) is not affected
by soil water deficit until the available water (θA) in the root zone is less than 0.3 of
the maximum available moisture content (θ max). Thus, when the available water
content in the root zone is in between 1 and 0.3 of the maximum, Et is considered
equal to T p. When available water content is less than 0.3 of the maximum then
Et = Tp θA/0.3 θmax (1.59)

122
 The concept was used by Cull. et al. (1981) for cotton crop. However, Hanks
(1974), Sammis et al. (1986) and Abdul Jabbar et al. (1983) assumed a value of 0.5 of
θmax. Singh and Wolkewitz (1988) adopted the critical value of 0.65 to 0.84 for
different growth stages of wheat. In the second approach, potential water supply (Pw)
by the root system is considered in relation to the potential demand by the crop (T p).
If water supply is greater than demand, then demand is the actual transpiration. If
water supply falls below the demand due to water deficits in the soil, then supply is
the actual transpiration.

123
 Overview of crop-weather models with emphasis on empirical crop-weather relationships
B. Bapuji Rao, Principal Scientist, (Agromet)

Crop-weather models integrate the crop development growth and production as a

function of weather and has many uses in agriculture. They serve as tools in land use planning,
crop adaptation, crop monitoring and forecasting, crop management, pest and disease control
and finally in prioritizing the research needs.
The main purpose of developing the crop-weather models are:
1. To understand crop weather interactions, processes involved and their limitations.
2. To assess the affect of environment, crop genotype and management of input resources
on crop yields, and to quantify the yield gaps with existing knowledge.
3. To undertake strategic and policy decisions to increase the productivity of resource based
efficient cropping systems.

During the last four decades, crop weather models have been used to solve practical problems
such as
1. Yield assessment of cereals, pulses an oilseed crops based on varying crop management
decisions during the growing season as well as over different rainfall years for risk
analysis using CERES models for cereal, pulses and oilseed crops (Boote et al., 1998).
2. Potential productivity of crops for regional agricultural planning. Yield gaps and decision
support systems (Naab et al., 2004).
3. Genetic improvement of cereals, pulses and oilseeds for yield, pest resistance food value
and input requirements (CERES models, IBSNAT programme, 1988).
4. Quantifications of impact of global climate change on agricultural productivity
(Pickering et al., 1995).
5. Management decisions on evaluating sowing date, row spacing, plant populations,
scheduling irrigation, evaluation of yield variations in different rainfall years, impact of
moisture and temperature stresses on yield.
6. To simulate growth, development and yield levels.
7. To define optimum management strategies regarding drainage, irrigation, soil, water,
weather, fertilizer, pest control, planting dates, tillage, crop residue management (For
example EPIC Model-Sharpley and William, 1990).
8. Evaluation of new crops for introducing at al location (Jones, 1990).

The models can be broadly categorized into Empirical statistical models, crop weather analysis
models and crop growth simulation models.

124
Empirical-statistical models
In this type of models, one or more variables representing weather/climate, soil water
availably, crop’s biological character etc., are related to crop responses such as dry matter yield
or seed yields. The independent variables are climatic or derived agrometeorological variables
such as moisture adequacy index (MAI) or soil water availability parameter or crop biological
characters such as LAI or GDD or plant characters. After removing the technological trend, the
significant agrometeorological variables are related to crop yield through standard statistical
procedures such as multivariable regression analysis. The weighing coefficients of these
expressions are obtained. Variance analysis, regression analysis, correlation and multivariable
regression analysis are some other common procedures. Ulanova (1975) forecasted winter wheat
using agrometeorological variables such as soil moisture reserves in growing season, average
number of head per m2, average height of plant, average number of kernels per head at heading
stage explaining 86 per cent of variability of wheat crop yields expressed as
Y = -19.92 + 0.29 X1 – 0.0013 X12 + 0.045 X2 – 3 x 10-5 X22 + 0.23 X3 – 14 X 10-5 X32 – 0.805
X4 + 0.057 X42 ...(Eq. 1)
where, Y is winter wheat yield in q/ha, X1 is soil moisture reserve in mm at heading stage
in 100 cm of soil depth, X2 is average number of heads per m2, X3 is average plant height in cm,
X4 is average number of kernels per head at heading stage.
Crop yields need predicted on operational mode utilizing agrometeorological
parameter/variables like rainfall, reference croo evapotranspiration temperature, soil moisture,
leaf area index etc. Empirical statistical relationships thus developed helps in
1. Assessing regional crop yields at regional level,
2. Evaluating impact of technology on crop yield production,
3. Assessing suitability of area for growing crops and yield potential and zoning of crop
productivity,
4. Assessing impact of climatic variability on agricultural production.

(A) Crop weather analysis models

These models are based on the product of two or more factors each representing the
functional relationship between a particular plant response i.e., crop yield and the variations in
selected weather variables at different crop development stages.
Input requirement of the models are only two to three effective weather variables
influencing crop growth and development at different phenophases, but the output is dependent
on the interactions of input factors with grain yield at different phenophases. However
conventional statistical procedures are used to determine the coefficients relating to crop
125
responses to agrometeorological data. There are two important examples in this category (i)
Baier’s crop weather analysis model (1973), and (ii) Robertson’s factorial yield weather model
(1974).
(i) Baier’s crop weather analysis model:

Baier (1973) studied the responses of daily contributions of upto three selected
agrometeorological variables at different phenophases in wheat. Fig. 1 illustrates the crop
response to each of the three input variables is either linear or quadratic, and this response
gradually changed during the crop life cycle as a function of biometeorological time (Robertson,
1968). A fourth power polynomial, with biometeorological time as independent term, was
adequate for fitting daily weighing factors associated with the daily contribution of each variable
to the final yield. Baier (1973) selected solar energy, temperature and soil moisture-the three
variables for predicting Canadian wheat yields. These three variables modify each other on any
particular day during the life cycle of a crop and produce a positive or negative effect on the
final yield expressed as

...(Eq. 2)
m

å variable grain yield,

Where, Y is dependent is the summation
t=1
of daily V values from

biometeorological time t = 0 (sowing) to m (physiological maturity) with intermediate values

expressed in decimals of the time units.
On the above relation (Eq. 2), V1, V2 and V3 are functions of the selected independent
agrometeorological variable. Each V function is of the form as
Vj = (u1t + u2t2 + u3t3 + u4t4) + (u5 t + u6t2 + u7t3 + u8t4) Xj + (u9t + u10t2 + u11t3 + u12t4) Xj2
...(Eq. 3)
Where, u1, u2, … u12 are coefficients which are evaluated for each Vj in an iterative
regression analysis in which the intercept has been suppressed and Xj in Vj represents a specific
selected variable for the analysis. Computer analysis is done to evaluate and testing of 12
coefficients in each of V1, V2 and V3 after every iteration. Model output provides the daily
contribution to the final yield in response to the variations in each of the input variables. The
critical threshold values for each of three variables are also provided as shown in figure 1 with
dashed lines of lower and upper thresholds. Baier used daily minimum temperature in °C(V1),
daily maximum temperature in °C(V2), and daily soil moisture parameter as the ratio of
estimated soil moisture (AET) to maximum available soil moisture (PET) within the crop

126
rooting zone varying between 0 to 1. The combination of minimum temperature, maximum
temperature and AET/PET ratio was found to give the closest estimate with CD value of 0.77 for
wheat yield.
(ii) Robertson’s factorial yield weather model (FYWM)
Robertson (1974) proposed a factorial yield weather model, which involved the
summation of the products of several quadratic functions for different weather variables. The
weather variables adopted are precipitation, maximum and minimum temperatures, global
radiation and pan evaporation. Time was used as an indicator of advancing technology and one
function was involved for the antecedent crop condition.
The model is of form readily adaptable for assessing, at anytime during crop
development period, the influence of past and current weather on future expected yield as
Ŷt = V1 (ŷ t-1, Pt) V2(T1)t X V3(T2)t X V4(Q)t ...(Eq. 4)
Where, Ŷt is the expected estimated yield at anytime t at a given crop stage, ŷ t-1 is the
estimated yield at the end of the previous stage, Pt is the rainfall between stages, T1 is average
daily maximum temperature during the period between stages, T2 is average minimum
temperature between stages, Q is average daily solar radiation during the period between stages.
V1, v2, V3 and V4 functions are of the form.
V1 (ŷ t-1, Pt) = a0 + a1 ŷ t-1 + a2Pt ...(Eq. 5)
V2(T1)t = b0 + b1 T1 + b2 T12 ...(Eq. 6)
V3(Y2)t = c0 + c1 T1 + c3 T22 ...(Eq. 7)
V4(Q)t = d0 + d1 Q + b2 Q2 ...(Eq. 8)

Where, a, b, c and d with subscripts 0, 1 and 2 are regression coefficients for each crop
period evaluated through statistical procedures.

The crop weather analysis models can also be used to study

1. The impact of climatic variability on crop yield in order to study the sensitivity analysis
and relative importance of various input weather elements in crop yield,
2. The analysis of crop weather data to illustrate their relative contributions to crop yields
as a function of biometeorological time, and
3. Evaluation of crop responses to weather elements at different critical phenophases in
crops life cycle.
(B) Crop growth simulation models

Crop growth simulation models are dynamic in nature considering physical, biological and
chemical processes in the system. They are intended to mimic the crop growth and several

127
models of varying degree of accuracy are available (Table 1). The processes considered in these
models are as follows:

1. PAR interception and biomass growth.

2. Carbon dioxide fixation.
3. Dry matter accumulation and its partitioning.
4. Tissue expansion and leaf area development.
5. Morphological development.
6. Phenological development
7. Soil water balance and soil water movement.
8. Grain yield prediction relations.
9. Soil environmental and crop management stresses.
10. Model testing, validation and sensitivity analysis.
11. Crop monitoring, yield forecasting and potential productivity of ecosystem.
12. Crop breeding.
13. Physiological insight into the crop and cropping systems.
14. Insect and disease management and linkage with pest and disease models.
15. Expert system for crop risk insurance management.
16. Soil erosion and long term soil productivity, soil conservation relations with time.
17. Crop adaptation and introduction of new crops.
18. Agro-ecological characterization and crop zonations.
19. Feasibility of inter-cropping, crop rotations and multiple cropping.
20. Operational programmes for agroadvisory services for assessing crop conditions with
prevailing weather conditions.
21. Operational on farm crop management decisions for crop production for determining
appropriate sowing dates, row spacing, plant population, fertilizer requirement, irrigation
scheduling, cultural operations, monitoring of soil and water resources.
22. Future trends in use of models for identifying research gaps and priorities, matching
technology with resources, yield forecasting and global food management, tactical and
strategic decisions (Bishnol, 2007).

The principles underlying in some of the processes considered in majority of crop growth
models are presented here under.

(A) Light interception and dry matter production:

The amount of biomass w(gm-2) accumulated by a vegetative crop stand can be expressed as
(Monteith, 1977).

128
w = ∫ Rs . e. f dt ...(Eq. 9)

where, Rs is the incident solar radiation (MJ d -1), t is time in days, e is coefficient or
constant for conversion of radiant energy into biomass dry matter (g MJ-1), f is the fraction of
incident radiation intercepted by the foliage (1-Rg/Rs) and Rg is transmitted radiation at ground
surface after passing through the foliage. The radiation/light use efficiency varies directly with
crop attenuation coefficient, crop genotypes, sowing date, plant population and environmental
variables particularly temperature and vapour pressure deficit (Rosenthal and Gerik, 1991b).

These models of biomass growth depend on leaf area index to accurately determine PAR
interception or absorption. LAI needs to be accurately depicted with the advancement of thermal
time. Several crop models use empirically derived relationship describing the leaf area as a
function of thermal time (EPIC, COTTAM, AUSCANE) in the form as (Sharpley and Williams,
1990)

LAIi = LAIi-1 + ΔLAI ...(Eq. 10)

ΔLAI = ΔTT . LAIMax [1 - exp{5(LAIi-1 – LAIMax)}] X √REGi
...(Eq. 11)
...(Eq. 12)

...(Eq. 13)

Where, TTIi is thermal time index for day i and PTT is the potential thermal time
required for maturity of crop, a is a parameter that regulates LAI decline rate of the crop and
TTId is the value of thermal time index factor when LAI starts declining attaining LAId value.

Monteith (1977) approach for assimilate production rate during vegetative cycle of crop
also works well for potential increase in daily biomass (EPIC, RESCAP models) as

ΔDM = LUE X PARi X (1 + ΔDL) ...(Eq. 14)

Where, ΔDL is the change in day length in hours per day. PARi is daily intercepted or
absorbed PAR, ΔDM is daily increase in biomass productivity.

In carbon driven models leaf area growth depends on the assimilate supply and leaf specific
weight. In RESCAP model, daily increase in leaf area was augmented as product of increase in
dry weight with leaf area ratio (LAR) (Monteith et al., 1989) expressed as

ΔLAI = ΔDM X LAR

Where, LAR is leaf area ratio expressed as the ratio of leaf area to dry weight of the plant
leaves.
129
 ...(Eq. 15)

Where, LAIt and LAIt-1 are leaf area at time t and t-1, and Wt and Wt-1 are dry weight during
the same period, respectively.

(B) Partitioning of assimilates

Partitioning of dry matter or carbon to various plant organs in the plant is facilitated with the
use of appropriate partitioning factors varying with plant development in the growing season
(Wilkerson et al., 1981; Van Heemest, 1986).

(C) Water use

Monteith et al., (1989) assumed that the amount of dry matter produced per unit of water
transpired (q) is inversely proportional to mean saturation deficit (SD) expressed as

q. SD = Constant

The quantity (q) is conservative for most of crops and will have a value around 9 gm kg-1
KPa (Monteith, 1989).

The demand for water to transpire (Tp) in producing daily dry matter ΔDM is

...(Eq. 16)

It Tp is less than the amount of water which the roots can supply then growth is assumed
to be light limited and dry matter or biomass accumulation is computed as

...(Eq. 17)
...(Eq. 18)

EPIC model uses Ritchie (1972) model for potential water use (PET) as a fraction of
potential evaporation (PE0) by using leaf area index (LAI) relationship on any ith day and
expressed as

130
 ...(Eq. 19)

Where PET is predicted plant water evaporation rate (mm/day). Potential soil water
evaporation is simulated by considering soil cover according to

PEs = min[PE0 – PET]

Actual soil water evaporation rate is estimated on the basis of root depth as

...(Eq. 20)

Where ET is total water used (mm) to depth z(m) on any day, Rz is the root zone depth in
m which is simulated as a function of thermal time and potential root zone depth (RDMX)
expressed as

...(Eq. 21)

The constant 2.5 allows root depth to reach its maximum when TTI reaches 0.4. The parameter
l is a water use distribution parameter.

(D) Yield estimations

In EPIC model harvest index increased non-linearly from zero at planting to maximum
value in the form of an expression as (William et al., 1989)

...(Eq. 22)

Where, HIi is harvest index on day i and HIc is harvest index of crop and TTFH is
thermal time factor that affects harvest index.

The influence of stress parameters on harvest index is reflected though growth

constraints. Crop yields are reduced through reduction in harvest index caused by water, nutrient
and crop management stresses. Most grain crops are very sensitive to water stress at flowering
and anthesis when major yield components are determined. Optimum conditions for growth may
reduce harvest index slightly where economic yield is limited by sink size. Harvest index is
affected by water stress expressed (William et al., 1989) as

131
 ...(Eq. 23)

Where AHI is adjusted harvest index, WSYFc is a crop parameter representing the
sensitivity of harvest index to soil moisture stress for the crop, CGS is a function of crop growth
stage and SMS is soil moisture stress factor for day i.

Therefore soil water stress influence harvest index between 0.3 to 0.9 of maturity with
maximum effect at 0.6. The water stress factor limiting biomass production is in proportion to
transpiration reduction. (Hanks, 1983).

Fischer model (1979) determines kernel number as the final outcome of vegetative
matter in pre anthesis period. A critical period of 25 days before anthesis and anthesis duration
during which the radiation and temperature values influence the kernel number in wheat.
Accumulated dry matter at anthesis (DMa) is strongly related to kernel number (KNO) in wheat
(Fischer and Kohn, 1966) expressed as

KNO(Cv. Yecora 70) = 4000 + 13 DMa R2 = 0.35 ...(Eq. 24)

KNO (Heron) = 2360 + DMa R2 = 0.83 ...(Eq. 25)

(E) Influence of stress factors on yield

(i) Water stress factor: It is ratio of actual daily water used by the crop to the potential water use
on the same day. Hanks (1983) proposed moisture stress limits in biomass production in
proportion to transpiration reduction are useful and can be linked with the crop model-water use
functions.

(ii)Temperature stress: William et al., (1984) proposed plant temperature stress factor (TSF)
expressed as

...(Eq. 26)

(iiI) Nutrient stress factor: The N and P stress factors arebased on the ratio of simulated plant N
and P contents to the optimal values. Jones (1983) expressed nutrient stress factors as a non-
linear term varying from 1.0 at optimal N and P to zero when N or P is half the optimum level.
The scaling factor expression (SF) for N stress was expressed as

...(Eq. 27)

(iv) Aeration stress factor: When soil water content approaches saturation, plants may suffer

132
from aeration stress. Water contents in top 100 cm soil depth is considered for assessing degree
of aeration stress expressed as (William et al., 1984)

, and ...(Eq. 28)

...(Eq. 29)

Conclusion:

A variety of crop-weather models are developed varying from simple empirical to

complex process oriented models. The accuracy with which they can mimic he crop growth
depends largely on the input data requirement and spatial attributes. A good data base essential
for the development of models that can simulate the system more accurately.

References

Angus, J.F. and Zandstra, H.G. 1980. Climatic factors and the modeling of rice growth and
yield. In: Proceedings of Syms on Agrometeorology of Rice Crop. IRRI, Los Banos, p.
189-99.

Baker, D.M., Lambert, J.R. and McKinion, J.M. 1983. GOSSYM: A simulation of cotton crop
growth and yield. South Carolina Agril. Expt. Station, Tech. Bull. 1089, Clemson.

Boote, K.J., Jones, J.W., G. Hoogenboom, G.G. Wilkerson, and S.S. Jagtap 1989. PNUTGRO.
Peanut crop growth simulation model. User’s guide departments of agronomy and
agricultural engineering. Univ. of Florida, Gainesville, Florida, p. 1-76.

Curry, R.B., Baker, C.H. and Streeter, J.G. 1975. SOYMODI: A dynamic simulation of soybean
growth and development. TSAE – 18, p. 963-68.

de Wit, C.T. 1982. Simulation of living systems. In: F.W.T. Penning de Vries and H.H. Van
Laar (Eds.). Simulation of plant growth and crop production. Simulation monograph
series. PUDOC, Wageningen, p. 3-8.

Duncan, W.G. 1975. SIMAIZ: A model simulation growth and yield in corn. In: D.N. Baker,
P.G. Creech and F.G. Maxwell (Eds.). An application of system methods to crop
production. Miss. Agric. and for. Expt. Stn., Miss. State Univ., Miss., p. 32-48

Hodges, T., Johnson, B.S. and Manrique, L.A. 1989. SUBSTOR. A model of potato growth and
development. In: Agronomy abstracts. Amer. Soc. Of Agron., Madison, p.16.

Hoogenboom, G., J.W. White, J.W. Jones and K.J. Boote, 1990. Dry bean crop growth
simulation model. Florida Expt. Station., Jour. No. N – 00379 Univ. of Florida,
Gainesville, Florida, p. 1-120.

133
Jackson, B.S., G.F., Arkin and A.B. Hearn, 1988. The cotton simulation model COTTAM:
Fruiting model calibration and testing. Trans. ASAE, 31, p. 846-54.

Jones, C.A., Wegener, M.K., Russell, J.S., McLeod, I.M., and Williams J.R. 1989a AUSCANE.
Simulation of Australian sugarcane with EPIC. CSIRO, Division of Tropical crops and
pastures, Tech. Paper no. 29.

Kiniry, J.R., J.R. Williams, P.W. Gassman, and P. Debaeke 1992. A General process oriented
model for two completing plant species. ALMNC 2. Texas A & M – BRC Report, p. 1-
34.

Monteith, J.L., A.K.S. Huda and D. Midya 1989. Modelling sorghum and pearl millet.
ICRISAT, Bulletin no. 12, p. 30-39

Ng, E. and Loomis, R.S. 1984. Simulation of growth and yield of potato crop. Simulation
monograph series. PUDOC, wageningen.

Retta, A. and Hanks, R.J. 1980. Manual for using PLANTGRO. Utah Agr. Exp. Stn., Res. Rep.
No. 46, Logan, Utah, USA, p. 1-14.

Ritchie, J.T. and G. Algaswamy 1989. Physiology of sorghum and pearl millet. In: Modelling of
the growth and development of sorghum and pearl millet. Res. Bull. No. 12, ICRISAT,
India, P. 24-29.

Ritchie, J.T., Alocilja, E.C., Singh, U., and Uchara, G. 1986. CERES – Rice Model. In: Proc.in
workshop on Impact of Weather Parameters on Growth and Yield of Rice IRRI, Los
Banos.

Ritchie, J.T., Godwin, D.C., and Otter – Nacke, S. 1985. CERES – Wheat. A simulation model
of wheat growth and development. Texas A & M Univ. Press College Station.

Shawcroft, R.W., Lemon, E.R., Allen, L.H. Stewart, D.W., and S.E. Jensen 1974. The soil-plant
atmosphere model and some of its predictions. Agril. Meteorology, 14(1/2), p. 287-307.

Stapper, M. 1984. SIMTAG. A simulation model of wheat genotypes. Univ. of New England,
Dept. of Agron & Soil Science, Armidale.

Stapper, M. and G.F. Arkin 1980. CORNF: A dynamic growth and development model for
maize (Zea mays L.) Texas A & M, BRC, documentation no. 80-82, p. 1-83

Van Keulen and Seligman, N.G. 1987. Simulation of water use, nitrogen nutritio, and growth of
a spring wheat crop. PUDOC. Wageningen.

Wilkerson, G.G., Jones, J.w., Boote, K.J., and Mishoe, J.W. 1985. SOYGRO. Soybean crop
growth and yield model. Tech. Documentation, Univ. of Florida, Gainesville, p. 1-253.
134
 Williams, J.R., C.A. Jones and P.T. Dyke 1984. A modeling approach to determining the
relationship between erosion and soil productivity. Trans. ASAE. 27(1), p. 129-144.

Table 1: Crop growth simulation models for different crop/processes

Model Name Crop Processes involved Reference

CERES- Sorghum Growth and development, grain Ritchie and Alagarswamy

Sorghum yield (1988)
RESCAP Sorghum Dry matter, water use, grain yield, Monteith et al., (1989)
radiation interception
RESCAP Pearl Millet Dry matter accumulation, water Monteith et al., (1989)
use, grain yield
SIMAIZ Maize Growth, grain yield Duncan (1975)

CERES-Maize Maize Growth and development, grain Stapper and Arkin (1980)
yield
COTTAM Cotton Growth, development, soil water Jackson et al. (1988)
budget, morphology
GOSSYM Cotton Growth, yield Baker et al. (1983)

SUBSTOR Potato Growth, development, yield Hodges et al. (1989)

IRRIMOD Rice Growth, development, yield Angus and Zandstra (1980)

CERES-Rice Rice Growth, yield, phenology Ritchie et al. (1986)

AUSCANE Sugarcane Growth, development, cane yield Jones et al. (1989)

SOYGRO Soybean Vegetative, reproductive, grain Wilkerson et al. (1985)

yield, phenology

SOYMOD Soybean Growth, development, yield Curry et al. (1975)

BEANGRO Dry bean Growth, development, phenology, Hoogenboom et al. (1990)

grain yield
PNUTGRO Peanut Growth, development, phenology, Boote et al. (1989)
yield
POTATO Potato Growth, yield Loomis (1984)

SIMTAG Wheat Genotypes growth and Stapper (1984)

development, grain yield

135
CERES- Wheat Growth, development, grain yield Ritchie et al. (1985)
Wheat
Wheat Water use, nitrogen nutrition, Van Keulen and Seligman,
growth, grain yield (1987)
Any living Plant growth and crop production de Wit (1982)
systems
EPIC Any crop and Soil productivity, erosion, plant William et al. (1984)
cropping growth processes, yield
SPAW Any crop Plant environment interaction, Shawcroft et al. (1974)
microclimate, dry matter, grain
ALMANAC Crop-weed Crop-weed competition Kiniry et al. (1992)
competitions
PLANTGRO Any crop Evapotranspiration and grain yield Retta and Hanks (1980)

136
 Climate Change Research in India : Past, Present and Future

Dr.B.Venkateswarlu
Director
Central Research Institute for Dryland Agriculture (CRIDA),
Hyderabad

Introduction

Climate change impacts on agriculture are being witnessed all over the world, but countries like
India are more vulnerable in view of the high population depending on agriculture and excessive
pressure on natural resources. The warming trend in India over the past 100 years (1901 to
2007) was observed to be 0.510 C with accelerated warming of 0.21oC per every10 years since
1970 (Krishna Kumar 2009). The projected impacts are likely to further aggravate yield
fluctuations of many crops with impact on food security and prices. Climate change impacts are
likely to vary in different parts of the country. Parts of western Rajasthan, Southern Gujarat,
Madhya Pradesh, Maharashtra, Northern Karnataka, Northern Andhra Pradesh, and Southern
Bihar are likely to be more vulnerable in terms of extreme events (Mall et al. 2006).

Rainfall and Temperature Trends

Rainfall is the key variable influencing crop productivity in rainfed farming. Intermittent
and prolonged droughts are a major cause of yield reduction in most crops. Long term data for
India indicates that rainfed areas witness 3-4 drought years in every 10-year period. However,
no definite trend is seen on the frequency of droughts as a result of climate change so far. For
any R&D and policy initiatives, it is important to know the spatial distribution of drought events
in the country. A long term analysis of rainfall trends in India (1901 to 2004) using Mann
Kendall test of significance by CRIDA indicate significant increase in rainfall trends in West
Bengal, Central India, coastal regions, south western Andhra Pradesh and central Tamil Nadu.
Significant decreasing trend was observed in central part of Jammu Kashmir, Northern MP,
Central and western part of UP, northern and central part of Chattisgarh. Analysis of number of
rainy days based on the IMD grid data from 1957 to 2007 showed declining trends in
Chattisgarh, Madhya Pradesh, and Jammu Kashmir. In Chattisgarh and eastern Madhya Pradesh,
both rainfall and number of rainy days are declining which is a cause of concern as this is a
rainfed rice production system supporting large tribal population who have poor coping
capabilities.
137
Due to increase in CO2 levels the earth is warming up. In the last 15-20 years, there has been a
sharp rise in the global temperature. While there are varying projections of temporal variations
in temperature, there is a near unanimity in its direction and trend (Figure 1). In India too, the
overall mean temperature is showing an increasing trend in most parts of the country. The
central and western parts showed decreasing trend in maximum temperature while increasing
trend of minimum temperature was observed in east, north & southern parts. Until last year,
2009 was the warmest year on record since 1901 (+0.913o C above the normal of 24.64oC) and
now 2010 has surpassed it (+0.93oC). The other warmer years on record in order are
2002(0.708), 2006(0.6), 2003(0.560), 2007(0.553), 2004(0.515), 1998(0.514), 1941(0.448),
1999 (0.445), 1958(0.435), 2001(0.429), 1987(0.413) and 2005(0.410).

Fig.1 Changes in global temperature trends over the last century

As far as Indian agriculture is concerned, temperature rise during rabi is of more significance.
For minimum temperatures, most of the locations in India are showing an increasing trend.
This is a cause of concern for agriculture as increased night temperatures accelerate respiration,
hasten crop maturity and reduce yields. The increasing trend is more evident in central and
eastern zones where rainfall is also showing a declining trend which makes this area more
vulnerable and requiring high attention for adaptation research.
Impact on agriculture, livestock and fisheries
The impact of climate change on agriculture may accentuate at regional level creating more
vulnerability in food security rather than global level as a whole. The potential impact will be
shifts in sowing time and length of growing seasons, which may necessitate adjustment in
sowing and harvesting dates, change in genetic traits of cultivars and sometimes total adjustment
of cropping system itself. With warmer environment associated with erratic rainfall distribution

138
the rate of evapotranspiration will increase and quick depletion of soil nutrient reservoir would
call for much greater efficiency in use of water and nutrients to sustain crop productivity. Apart
from these, tackling with frequent and more intense extreme events like heat and cold waves,
droughts and floods may become norm of the day for common farming community (IPCC,
2001). Such phenomena will impact agriculture considerably through their direct and indirect
effects on crops, livestock, and incidences of pest-disease-weeds, increasing deterioration of soil
health in totality and thereby threatening the food security like never before.
Crop production
The Indian Council of Agricultural Research (ICAR) initiated an all India Network project in
2004 to study the possible impacts of climate change on major crops, livestock, fisheries, soils
and other biotic factors as well as to understand natural adaptation capabilities of both flora and
fauna. The possible interventions to increase the adaptability of crop-livestock systems and
mitigation measures to minimize the adverse impacts were studied across different agro-
ecosystems of India. The output of the studies (Aggarwal, 2009) so far indicated that a marginal
1 0C increase in atmospheric temperature along with increase in CO2 concentration would cause
very minimal reduction in wheat production in India if simple adaptation strategies like
adjustment of planting date and varieties are adopted uniformly. But in absence of any adaptive
mechanism, the yield loss in wheat may go up to 6 million tones. A further rise by 5 0C may
cause loss of wheat production up to 27.5 million tones. Similarly, rice yields may decline by
6% for every one degree increase in temperature (Saseendran et al. 2000). In addition to direct
effects on crops, climate change is likely to impact natural resources like soil and water.
Increased rainfall intensity in some regions would cause more soil erosion leading to land
degradation. The study on wheat and rice suggested that high temperature around flowering
reduced fertility of pollen grains as well as pollen germination on stigma. These effects are more
pronounced in Basmati rice as well as Durum wheat cultivars. A positive finding of the study
was that the Aestivum wheat cultivars are more or less tolerant to such adverse affects. But
differential impact of increasing temperature is observed with respect to grain quality of wheat
where it is found that Aestivum wheat cultivars are more prone to reduced grain quality due to
increasing temperature during the fruit setting stage than Durum cultivars.
Field experiments using advanced ‘Temperature gradient tunnels’ with different dates of sowing
to study impact of rising temperature on growth and development of different crops revealed that
an increase of temperature from 1 to 4 0C reduced the grain yield of rice (0-49%), potato (5-
40%), green gram (13-30%) and soybean (11-36%). However, one of the important pulse,
chickpea, registered 7-25% increase in grain yield by an increase in temperature up to 3 0C, but
was reduced by 13% with further 1 0C rise in temperature.

139
Horticulture
A significant decrease in average productivity of apples in Kullu and Simla districts of Himachal
Pradesh have been reported which is attributed mainly to inadequate chilling required for fruit
setting and development. Reduction in cumulative chill units of coldest months might have
caused shift of apple belt to higher elevations of Lahaul-Spitti and upper reaches of Kinnaur
districts of Himachal Pradesh. However results from simulation models suggest that climate
change could benefit coconut crop. Coconut yields are likely to increase by 4, 10, and 20% by
2020, 2050 and 2080, respectively, in the western coastal areas of Kerala, Maharastra, Tamil
Nadu and Karnataka. But the impact may be negative in east coast areas as they are already
facing a much warmer atmospheric thermal regime than western coast.
Insect and pest dynamics
The impact of rising temperature and CO2 are also likely to change insect pest dynamics.
Dilution of critical nutrients in crop foliage may result in increased herbivory of insects. For
example, Tobacco caterpillar (Spodoptera litura) consumed 39% more castor foliage under
elevated CO2 conditions than controlled treatments (Srinivasa Rao et al. 2009). The
advancement of breeding season of major Indian carps as early as March has been reported from
West Bengal which is extended from 110 to 120 days due to increase in environmental
temperature, which stimulates the endocrine glands of fish and helps in the maturation of the
gonads. This brings about a possibility to breed these fishes twice a year at an interval of 30 to
60 days. Increased heat stress associated with rising temperature may, however, cause distress to
dairy animals and possibly impact milk production. A rise of 2 to 6 0C in temperature due to
climate change is expected to negatively impact growth, puberty and maturation of crossbred
cattle and buffaloes. As of now, India losses 1.8 million tones of milk production annually due to
climatic stresses in different parts of the country. The low producing indigenous cattle are found
to have high level of tolerance to these adverse impacts than high yielding crossbred cattle.
Soil and water resources
Besides, the nutrient loss from soil through high rate of mineralization and CO2 emissions from
soil could be accelerated as a result of increase in temperature. Low carbon soils of mainly
dryland areas of India are likely to emit more CO2 compared to high or medium carbon
temperate region soils. Simulation of water balance using Global and Regional Climate Models
revealed likely increase in annual as well as seasonal stream-flows of many Indian river basins
pointing to the need for adoption of more effective runoff and soil loss control measures to
sustain crop production across the country. At the farm level increased temperatures will also
increase crop water requirement. A study carried out by CRIDA (unpublished) on the major
crop growing districts in the country for four crops, viz., groundnut, mustard, wheat and maize

140
indicated a 3% increase in crop water requirement by 2020 and 7% by 2050 across all the
crops/locations. The increase in water requirement for major crops like maize, cotton and
groundnut in different agroclimatological zones of AP by 2020 is given Table 1. The crop
duration is also likely to be reduced by 1-2 weeks.

Table 1. Crop water requirements to rise: crop duration to decrease (eg. AP in India)

Station Crop Increase in water Reduction in crop

requirement in mm duration (weeks)
(2020-2025)
Anankapalli Maize 51.7 1
Groundnut 61.3 1
Anantapur Groundnut 70.1 1
Redgram 174.3 1
Jagityal Cotton 60.5 2
Maize 49.0 1
Rajendranagar Red gram 114.5 2
Groundnut 73.0 1
Tirupathi Groundnut 73.0 1

(Source : NPCC, CRIDA, 2007)

Managing Weather Risks in Agriculture
A comprehensive strategy of utilization of existing knowledge, strengthening R&D in
key areas and evolving a policy frame work that builds on risk management and providing
incentives to sustainable use of natural resources will be required for successful adaptation by
farm sector to climate. The goal of this strategy is to minimize as risks associated with farming
and enable farms to cope with these risks (Singh et al., 2009).

Technology options
Small changes in climatic parameters can often be managed reasonably well, by altering
dates of planting, spacing, input management, new cultivars adapted to drier conditions, salt
water resistant varieties of crops in the areas where drainage is poor development of irrigated
agriculture and farming systems like mixed cropping, crop-livestock and that are more adapted
to changed environment can further ease the pressure. In addition to these, improving technology
to increase production in climate favourable sites in order to offset uncertain production in

141
marginal areas, better adaptation of agricultural calendar, crop diversification to spread risks and
setting up processing and storage facilities.
World over, crop diversification is regarded as the most common and effective risk
management strategy that is employed by farm households. Multiple cropping system is another
strategy that even if a particular crop does not do well, the loss will be compensated by gains in
another crop. Optimum use of fertilizers and ecologically clean agro technologies would be
another risk management strategy. There are some limitations of this strategies however. First,
diversification is clearly a feasible strategy to the extent that crop risks are independent,
however, if returns are strongly correlated across crops, the risks facing farmers are similar to
systemic risks and crop diversification will not be effective in reducing producer risk. Second,
crop diversification calls for spreading resources across crops even when a particular crop offers
higher average net returns than other crops. Therefore, the price of diversification is the income
foregone, on average, by not growing the remunerative crop. Third, if there are fixed costs in the
cultivation of a particular crop, then there is a minimum efficient scale and that may conflict
with the requirements of crop diversification. Farmers with smallholdings are likely to run into
this constraint. The major impact of climate change in arid and semi-arid regions is likely to be
an acute shortage of water resources associated with significant increases in surface air
temperature. Some of the management strategies in semi-arid and arid region are as follows:
Semi-arid regions:
1. Shift to drought tolerant cultivars
2. Enhancement and maintenance of soil fertility and protection of soils from degradation
3. Development of complementary irrigation
4. Development of early warning system on drought and other climate induced natural
disasters
5. Implementing crop livestock integration
6. Implementing agroforestry systems
Arid regions:
1. Shifting from agriculture to other less climate sensitive activities (Livestock, Agroforestry)
2. Use of short duration varieties
3. Optimize planting dates

Policy Options

Apart from the use of technological advances to combat climate change, there has to be
sound and supportive policy framework. The frame work should address the issues of
redesigning social sector with focus on vulnerable areas/ populations, introduction of new credit

142
instruments with deferred repayment liabilities during extreme weather events, weather
insurance as a major vehicle to risk transfer. Governmental initiatives should be undertaken to
identify and prioritize adaptation options in key sectors (storm warning systems, water storage
and diversion, health planning and infrastructure needs). Focus on integrating national
development policies into a sustainable development framework that complements adaptation
should accompany technological adaptation methods.

In addition, the role of local institutions in strengthening capacities e.g., SHGs, banks
and agricultural credit societies should be promoted. Role of community institutions and private
sector in relation to agriculture should be a matter of policy concern. There should be political
will to implement economic diversification in terms of risk spreading, diverse livelihood
strategies, migrations and financial mechanisms. Policy initiatives in relation to access to
banking, micro-credit/insurance services before, during and after a disaster event, access to
communication and information services is imperative in the envisaged climate change scenario.
Some of the key policy initiatives that are to be considered are:
Mainstreaming adaptations by considering impacts in all major development initiatives
Facilitate greater adoption of scientific and economic pricing policies, especially for water, land,
energy and other natural resources. Consider financial incentives and package for improved land
management and explore CDM benefits for mitigation strategies.
Establish a “Green Research Fund” for strengthening research on adaption, mitigation and
impact assessment. (Venkateswarlu and Shanker 2009).
Globally, weather insurance plays an important role in mitigating climatic risks. In
several developed countries this strategy has worked successfully as these countries have
excellent long term weather data, farmers have large holding and have a business approach for
farming. In India, the small holders are generally more prone to risks but they are averse to buy
insurance policies. The crop insurance scheme has made some progress but it is a long way to
go. Considering the climate trends being witnessed in recent years all over the country, weather
based insurance appears to be a better alternative for mitigating risks in agriculture for Indian
farmers. The research institutes and insurance companies jointly should develop crop wise data
on the weather sensitivity so that appropriate policies can be designed which are friendly to
farmers, at the same time keep the insurance companies viable. The Government also should
share the premium burden. Instead of spending huge amounts of money on rehabilitation after
the disaster, it is prudent to spend on premium subsidy.

143
 Finally, there is a need to make climate change adaptation and mitigation measures as an
integral part of overall planning and development strategy of the country on long term.
(Venkateswarlu and Shanker, 2009).

References
Agarwal, P.K. 2009. Global Climate change and Indian agriculture; Case studies from ICAR
network project. Indian Council of Agricultural Research. 148p.

IPCC. 2001. Climate Change 2001: The Scientific Basis. Contribution of Working Group – I to
the Third Assessment Report of the IPCC. [Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M.,
van der Linden, P.J., Dai, X., Maskell, K. and Johnson, C.A. (Eds)]. Cambridge University
Press, United Kingdom and New York, USA, 94p.
Krishna Kumar. 2009. Impact of climate change on India’s monsoon climate and development
of high resolution climate change scenarios for India. Presented at MoEF, New Delhi on
October 14, 2009 (http: moef.nic.in)
Mall, R. K., Gupta, A., Singh, R., Singh, R.S. and Rathore, L. S. 2006a. Water resources and
climate change: An Indian perspective. Current Science. 90 (12): 1610-1626.
NPCC, 2007. Consolidated Annuall Report 2004-2007. Network Project on Climate Change,
Central Research Institute for Dryland Agriculture, Hyderabad, 317p.
Saseendran, A.S.K., Singh, K.K., Rathore, L.S., Singh, S.V. and Sinha, S.K. 2000. Effects of
climate change on rice production in the tropical humid climate of Kerala, India. Climate
Change. 44: 495-514
Singh, A. K., Aggarwal, P. K., Gogoi, A. K., Rao, G. G. S. N. and Ramakrishna, Y. S. (2009).
Global Climate Change and Indian Agriculture: Future priorities. In: Global Climate Change and
Indian Agriculture: Case studies from the ICAR Network Project (Ed. Aggarwal, P. K.), ICAR,
New Delhi. 146-148
Srinivasa Rao, Ch., Ravindra Chary, G., Venkateswarlu, B., Vittal, K.P.R., Prasad, J.V.N.S.,
Sumanta Kundu, Singh, S.R., Gajanan, G.N., Sharma, R.A., Deshpande, A.N., Patel, J.J. and
Venkateswarlu, B. and Arun K. Shanker. 2009. Climate change and agriculture: Adaptation
andmitigation strategies. Indian Journal of Agronomy. 54(2): 226-230
World Bank. 2008. Climate change impacts in drought and flood affected areas: Case studies in
India. Report No.43946-IN, South East Asia Regional Office, New Delhi.

144
 Agrometeorological instruments and data collection
AVR Kesava Rao
Scientist (Agroclimatology), Resilient Dryland Systems, ICRISAT, Patancheru

Introduction

Agroclimatic analysis and characterization of watersheds is carried out based on the long-period
weather data. To assess the impacts of interventions made during the development phase,
weather needs to be monitored by establishing a manual agromet station or by installing an
automatic weather station (AWS). Manual agromet station is to be established by following the
standard procedures prescribed by the India Meteorological Department (IMD) and the observer
or the volunteer has to be properly trained in recording the data and maintaining the instruments.

Layout of agrometeorological observatory

The site for the agromet observatory at the watershed is to be selected is such a way that it
represents the general climatic conditions of the area. The site should not be on the top of a hill
or at the bottom of a valley. No tall tress or buildings should be near the observatory which will
affect the wind flow and exposure to sunshine. As per the IMD, a site with dimensions of 55 m
in the N-S direction and 36 m in the E-W is required for a class-A type observatory. A site
having 25 x 15 m area is sufficient for a class-B observatory. Class-A observatory will have all
145
the manual / eye-reading and automatic instruments, while class-B will have only manual
reading instruments.

The IMD has prescribed the specifications for all the agrometeorological instruments and these
can be procured from various supplying firms in the country. Once procured, these instruments
are to be installed as per the standard lay-out. The observatory should be protected with a barbed
wire fence and suitable gate with locking arrangements is to be provided.

Basic weather parameters useful for agricultural research purpose are maximum temperature,
minimum temperature, soil temperatures at different depths, relative humidity (morning),
relative humidity (afternoon), wind speed, wind direction, rainfall, soil moisture at different
depths, evaporation, sunshine and dew at different heights. Air temperatures are measured by
four thermometers installed in a single Stevenson screen. These are dry bulb, wet bulb,
maximum and minimum thermometers. Relative humidity is computed from the dry bulb and
wet bulb temperatures using hygrometric tables. Soil temperatures are generally measured at soil
depths 5, 10 and 20 cm. Rainfall is manually measured from an ordinary raingauge (also called
FRP raingauge). Wind speed is measured by a cup anemometer and wind direction by wind
vane. Evaporation is measured by an open pan evaporimeter and hours of bright sunshine by a
sunshine recorder. Duvdevani dew gauges are used to measure dew at different heights. Soil
moisture is measured using gravimetric method, neutron-probe equipment, tensiometers or time
domain reflectometers.

Times of observation
Air temperature, soil temperature, relative humidity, wind speed and wind direction are recorded
everyday at 07:00 h and 14:00 h Local Mean Time (LMT). These are the normal times of
minimum and maximum temperature conditions. Rainfall and evaporation are observed at 08:30
h Indian Standard Time (IST) and 14:00 h LMT. Depending on the longitude of the watershed
location the time in IST corresponding to the time in the LMT can be computed. Soil moisture is
observed at specific intervals based on experimental requirement. Dew is recorded just before
sunrise and sunshine card is changed for exposure everyday morning before sunrise or after
sunset for estimating bright sunshine hours in a day.

Proper protection against theft and damage is to be ensured for the instruments. Weather data
monitored at the watersheds need to be quality checked and datasets developed.

Quality checking and database development

Quality of agromet data is essential for proper understanding the weather of the watershed and
for later computing the derived parameters for interpretation. Instruments in the observatory are
to be inspected by authorized personnel once in every year by comparing with standard
146
equipment and identifying the calibration errors and calibration drifts if any. Meteorological
observer needs to be trained in following the times of observations, maintaining of instruments,
recording the measurements with sincerity and keeping the data forms and books. Agromet data
once collected needs to be entered in to a computer in the form of an MS-Excel file or a MS-
Word file. Agromet Databases can be developed using the data of one watershed or multiple
watersheds using software like MS-Access with the help of professional software developers.
While developing databases, crop and soil data may also be considered for inclusion. Suitable
data retrieval programmes and software for computing derived weather parameters like reference
crop evapotranspiration, water surplus, water deficit and various agroclimatic indices are to be
developed along with the databases.

Automatic Weather Station

Automatic Weather Station (AWS) is a system to record the changes in the weather
continuously without any human intervention. The AWS consists of a datalogger, set of
sensors, power supply, solar panel, mounting stand and other accessories. The AWS should
be located in such a place that it represents the general agroclimatic conditions of the
watershed area. Datalogger program should be optimized for power and memory usage and
checked thoroughly for any programming bugs. Depending on the manufacturer and model,
the cost of the AWS can vary between 1.0-4.5 lakh Rupees. While choosing the model of the
AWS, budget will obviously be one important factor, but there are other key considerations
like sensor quality, communication and data transfer facility, data storage capacity and
battery back-up time. A balance between cost and these is required. Proper protection against
theft and damage is to be ensured.

Benefits of AWS:

· AW Stations can run for weeks and months without attention. Weather can be easily read
direct from the console display or monitored from faraway places using wired or wireless
modem, mobile phone or satellite communication.

· AW Stations automatically record maximum and minimum values for all weather
elements through each day and also keep track of total daily, monthly and yearly rainfall.
Routine daily maintenance like setting of maximum and minimum thermometers,
emptying the raingauge, change of sunshine card etc., are not required.

· Much greater within-the-day details are available, like complete pattern of wind speed &
direction.

147
· Derived parameters like degree days, reference crop evapotranspiration can be
automatically computed using specific software.

148
 Agromet Database Management
A.V.R. Kesava Rao and Suhas P Wani
Resilient Dryland Systems, ICRISAT, Patancheru
Reliable and long-term agroclimatic data are needed for undertaking climatic analyses,
particularly those aiming to assess climate variability and change and their impact on
agricultural production. Data on crops, varieties, and production at district-level for several years
is needed to understand the variations in agricultural productivity and changes in the cropping
patterns at the region. Data on crop morphology, phenology and yield characters obtained from
the various field experiments conducted by the State Agricultural Universities and ICAR
Institutes in the country, when made available at one location and provided easy access, will be
of great use to quantify crop -weather relationships and validating the crop-growth simulation
models. Use of meteorological instruments and dedicated meteorological observers and
organizations in country have paved the way for the availability of a very long-period weather
data in India. With the availability of electronic computer systems, database management has
become a reality.
“Database is a collection of non-redundant data, sharable between different application
programs”
Conventional method:
The conventional method of handling data is to store it in a file. User requires application
programs to manipulate the data stored in files.

Example: Agromet Department in a State Agricultural University contains data on

different weather parameters as recorded at the various research stations under the SAU.

Database management following conventional method requires application programs or software

to:
· Add new stations and related data
· Calculate means, sums etc. on weekly and monthly basis
· Compute derived parameters and indices
· Generate various reports

These application programs are developed according to the needs of the user. New programs are
to be added to the system as need arises. New files with different record format may have to be
added after some time. New programs have to be developed or existing programs updated to
manipulate the data in the new files. Thus, as the time goes by, more files and more application
programs are created. In file management system, data declarations and executable statements
are all part of the application program, while the actual data is in a file. If any changes are made

149
in data file structure, all the application programs that use this particular data file need
modifications. Any program that does not reflect this change will suffer “Data Inconsistency”.

Some other disadvantages of this scheme are:

· Files may have more redundant data
· Data stored in such files may not be secure

Therefore, we can conclude that this method of handling data is not suitable and we need a
system specifically for managing a database i.e., Database Management System.

Data Base Management System

A database management system (DBMS) is a set of application programs that acts as a

layer between the physical database and its users. All the requests from users for access to
the database are handled by the DBMS. One general function provided by the DBMS is
shielding database design details to users.

Features of DBMS:

1. Data Independence:

A database management system with its catalog facility helps to achieve application programs
“Data Independence”. It also provides for a centralized management and control of data
avoiding the “Data Inconsistency” that is faced in conventional file system. This also allows
sharing of data, thus avoiding data redundancy.

2. Data Integrity:

DBMS overcomes the problem of data inconsistency by providing integrity constraints with
data definition.

3. Data Representation:
DBMS provides conceptual representation of data, which frees users from the details of how
data is stored.

4. Data Security:
DBMS ensures security of data by providing different security and access levels to different
types of users. Therefore rights of users on database can be controlled effectively.

150
5. Data Concurrency:
DBMS takes care of multi-user issues by providing powerful locking mechanisms. It places
automatic locks on database and records when any operation that affects the data takes place.
These prevent updating of record or field by more than one user at a time in a multi-user
environment.

6. Data Sub language:

DBMS provides DDL (Data Definition Language), DCL (Data Control Language) and DML
(Data Manipulation Language), which allow defining data structure, data control and easy
retrieval / updating of data.

Data Models Supported by DBMS

Three most popular data models are:
· Hierarchical Model
· Network Model
· Relational Model

Hierarchical Model:

IBM has developed the Hierarchical Model database management system in 1968, also known
as Information Management System (IMS). A Hierarchical Model is a simple parent-child
structure or tree structure; each child can have only one parent. The data is represented as a
collection of trees. Data items are grouped into logical records.

Network Model:

The Network Model was designed as an improvement over hierarchical model. Here multiple
parent-child relationships are allowed. This reduces data redundancy and provides easy access to
information. It consists of a database of records where each record has a pointer to the record
preceding or following record.

Relational Model:

The Relational Model eliminates explicit parent-child relationships. There are no pointers
maintained and records are logically connected by key values. Hierarchical and network models
deal with one record at a time while relational model reads and writes data in units of a set of
records. In this model, data is organized in the form of tables comprising rows and columns. Any
row is identified by a column or set of columns that form a primary key. Dr. E.F. Codd of IBM
has proposed the relational model in 1985. He presented 12 rules that a database must obey if it
is to be considered as truly relational.
151
Codd’s Twelve Rules:
1. Information Representation at the logical level.
2. Guaranteed Access
3. Systematic treatment of Null values
4. Dynamic catalog based on relational model
5. Comprehensive data sub-language
6. View updating
7. High-level update, insert, delete
8. Physical Data Independence
9. Logical Data Independence
10. Integrity Independence
11. Distribution Independence
12. Non-Subversion Rule

No currently available relational DBMS fully satisfies all twelve of Codd’s rules. But it has
become a common practice to compile ‘score-card’ for commercial relational DBMS products to
show how well they satisfy each of the rules.

Concepts of Relational Model:

The relational DBMS takes its concepts from Relational Algebra.

In a relational literature, tables are considered as relations, rows are termed as tuples and
columns as attributes. The equivalent terms used by different people are

Relational Model Programmer User

Relation File Table

Tuple Record Row

Attribute Field Column

A domain is a pool of values from which the actual values appearing in given column are drawn.
The relational model provides a relational language, called SQL (Structured Query Language).

To summarize, Agromet database management system includes:

o Data acquisition, entry, storage and archiving

o Data quality control

152
 o Designing an appropriate Agroclimate database management system with a scope
for scalability

o Computer hardware and software

o Data access and application software development

o Data administration and monitoring

o Policy on data sharing

Applications of the Agromet database management include:

o Climate change assessment and impact studies

o Crop weather modelling

o Developing strategies for sustainable agricultural production

o Urban and tourism development

o Coastal zone management

o Resource characterization and research prioritization

General guidelines regarding agrometeorological observations:

Observations should be taken in as little time as possible (to avoid vitiation due to presence of
the observer). Punctuality is a matter of prime importance in recording the observations.
Faithful recording is important and every observation should be recorded as faithfully as read.
Each observation must be written down in the meteorological register immediately after it is
taken. Each observation must be checked after it is noted down in the meteorological register to
make sure that no mistake has been made. The observatory surroundings need to be maintained
in such a way that there are no tall buildings or trees nearby which may affect the weather
measurements. The positions of the instruments must never be changed.

Important agromet data entry forms of the IMD are:

1. CWS 1 MET-1 (AGRI)-1 - Agromet observations
2. CWS 2 MET-1 (AGRI)-2 - Micrometeorological observations
3. MET-T-149 - Hours of bright sunshine observations
4. MET-1 (AGRI)-56 - Pocket Register for agromet observations
5. MET-1 (AGRI)-65, CWS-27 (a) - Dew observations
6. MET-1 (AGRI)-66, CWS-27 (b) - Dew observations

153
Benefits of automated measurements are:

· AW Stations can run for weeks and months without attention. Weather can be monitored
from indoors; data can also be easily read direct from the console display. Detailed
weather conditions may be viewed at any distance from the station itself, for example
over the Internet.

· Routine daily maintenance like setting of maximum and minimum thermometers,

emptying the raingauge, change of sunshine card etc., are not required.

· AW Stations automatically record maximum and minimum values for all weather
elements through each day and keep track of total daily, monthly and yearly rainfall.

· Much greater within-the-day details are available, like complete pattern of wind speed &
direction.

· Derived parameters like degree days, reference crop evapotranspiration can be computed
using specific software

Choosing and setting up of AWS

While choosing, budget will obviously be one important factor, but there are other key
considerations like:

· Which sensors? Recording of the basic meteorological parameters is the minimum

requirement. However, very often, value of the meteorological observations is
increasingly appreciated after the station is installed. More and more parameters are
required to be monitored and hence, planning initially (by choosing a suitable datalogger)
for having the option of adding more sensors later will help. Measurement of UV
intensity, leaf wetness and soil moisture may become very relevant in the near future.

· Sensor sensitivity is an important factor, however a balance between accuracy, cost and
data application is required.

· Communication and data transfer mechanisms, special data handling requirements like a
live weather reporting website on Internet.

· Memory requirements of the datalogger in case of long-period unattended operation to be

worked out. Other important factors are the availability of matching software for data
download, export to various other software applications like spreadsheets or databases,

154
 automated graphical display of parameters, computation of derived parameters like PET
and indices for pest and disease forecasting.

· The AWS should be located in such a place that it represents the general agroclimatic
conditions of the area. Height of the sensors and other exposure criteria are similar to
that of a manual observatory, such that the data generated from the AWS is comparable
and reliable to that generated from a manual observatory. The AWS is to be installed
initially near a manual observatory and the data compared for a few days and then only is
shifted to the proposed location.

· Datalogger program should be optimised for power and memory usage and checked
thoroughly for any bugs.

· Proper protection against theft and damage is to be ensured.

155
 Crop simulation models in agricultural research and management: steps
S. Naresh Kumar
Division of Environmental Sciences, Indian Agricultural Research Institute,
New Delhi-110012
[email protected]

Introduction

A model is a simplified representation of a complex system. Modelling of a crop has been done
using approaches such as descriptive modelling, which is simple, or by explanatory modelling,
which quantitatively describes the mechanisms and processes that cause the behaviour of a
system. Crop growth simulation models, falling in the latter category, are based on
quantitative understanding of the underlying processes, and integrate the effect of soil,
weather, crop, and pest and management factor on growth and yield. The process could be
crop physiological, meteorological, and soil physical, chemical or biological. Depending
upon the objective, knowledge base of various agricultural disciplines can be integrated in
a crop model. For instance simulating the crop-weather interaction forms the production
level 1; while simulating growth rates determined by the availability of water apart from
weather of a location gives production level 2. Inclusion of availability of factors such as
nitrogen, other nutrients for crop growth provides production level 3…n. Addition of pests,
diseases, weeds, etc. in simulating the crop growth and yield will further provide
production levels more nearer to reality.

Fig 1: Relational diagram of a simulation model at production level 1

156
For simulating, the models need input data that mimic ‘genetics’ of a crop/variety. Further,
the response of variety to water, nutrient, pest limited or actual productivity, knowledge
base of several additional disciplines are tapped and integrated into the model. Once the
integration, calibration and validation is successful, crop simulation models can help us in
analyzing the effect of various climatic factors on crop growth and yield considering the
interaction with edaphic, biotic and agronomic factors. Such an analysis is normally not
possible with conventional experimental methods. There have been over 120 crop models
or compendium of models available across the world which can simulate 151 crops which
include filed crops, horticultural crops, plantations, grasses, etc.

Fig 2: Simplified relational diagram of InfoCrop- a generic crop simulation model

157
In recent years, agricultural system models have shifted from being mainly research oriented to
tools for guiding resource management and policy-making. The linkage of these models to
geographic information systems (GIS) and decision support systems has added dimensions to
model applications. Agricultural system model have gone through more than 40 years of
development and evolution. Prior to the mid-1980’s most of the modeling work focused on
individual processes of agricultural systems, such as soil hydraulic properties,
evapotranspiration, photosynthesis, plant growth and soil nutrients. The earlier models
have served as a foundation for the development of agriculture system models in the last 20
years. Earlier examples of systems models have focused, for example, PAPRAN for
pasture systems, CREAMS for soil, chemical and nutrient run off from cropping system,
EPIC for soil erosion and soil productivity, CERES for crop growth, GLEAMS for ground
water pollution, AquaCrop and CRPWAT for crop water requirement analysis and
CENTURY for plant production, nutrient cycling and soil organic matter dynamics.
Physiological growth and production models have shown to be very useful for guiding
improvements in cropping systems of various annual crops. There have been several crop
models and decision support systems available. Examples include DSSAT, InfoCrop, EPIC,
APSIM, CROPSYST, etc.

Calibration and validation of simulation models

Although the simulation models are flexible enough to perform under a variety
of environments and farming conditions, calibration of model is necessary
before running the model for study area. For this results from the detailed
experiments on varietal performance can be made use of. The calibrated model
can be used to simulate the crop growth performance in other set of
experiments consisting of various treatments for validating the model
performance under a range of conditions. This validated model can used for
simulating impacts in that region. Model performance can be assessed through
various statistical parameters viz., model bias error (MBE), root mean square error
(RMSE), index of agreement (IA) and model efficiency (ME) among others.

158
Fig 3 : Examples of calibration and validation of model- essential steps before the application of
any model

(source: Srivastava et al., 2010; Byzesh et al., 2010)

SImulated and measured nut yield/palm in different Coconut simulation model validation in different agro-climatic zones
expriments
18000
180
16000
Simulated nut yield/ha/year

y = 0.8858x + 13.649
160 2
R = 0.8609 14000 R2 = 0.9374
140
12000
Simulated nut yield/ plam

120
Kasaragod (Kerala) 10000
100 Arisikeri (Karnataka)
Aliarnagar (TN) 8000
80 Ratnagiri (MS)
6000
60
4000
40
2000
20
0
0 0 2000 4000 6000 8000 10000 12000 14000 16000 18000
0 50 100 150 200
Measured nut yield/palm
Measured nut yield/ha/year

(source: Naresh Kumar et al., 2008)

Application of Crop Simulation models

Crop models are increasingly being used for environmental characterization and agro-ecological
zoning, defining research priorities, technology transfer, estimating potential production,
strategic and anticipatory decision making. In past 20 years crop simulation models are
increasingly used for projecting the effects of climate change and climate variability. Recently,
they have been used for quantifying the adaptation gains in climate change scenarios for
prioritizing technology dissemination and also for identification of vulnerable areas.
159
Table 2 : Indicative use of models for several purposes and end uses
Purpose Best used for
Decision support Guiding current management options
Climate change impacts and / or adaptation Climate change adaptation and
mitigation guidance
Productivity / yield prediction or forecasting Yield or productivity forecasting
Research for crop management improvement Inter disciplinary research
Research for crop genetic improvement Better understanding of processes,
Improved crop breeding
Education / training Improving training and education
Operations optimization Policy development
Yield gap analysis Improving agricultural management/
R&D policy development

In a recently conducted global survey on use of crop simulation models (Rivington and Koo,
2010), it was found that the major purpose of the models are seen to be for decision support,
analysis of climate change impacts and/ or adaptation, prediction or forecasting of productivity /
yield and research for crop management improvement.
Crop simulation models are effective tools for the assessment of growth and yield of crops as
well to suggest optimal resource management options (Kalra and Aggarwal, 1994; desired
cultivar characteristics (Aggarwal et al., 1997) performance evaluation of weather forecasters
(Kalra and Aggarwal, 1996; Singh et al., 1997). Apart from these, crop simulation models are
now being seriously investigated as creditable tool for regional yield prediction (Nain et al.,
2002) and integration of crop simulation model with remote sensing data for farm level wheat
yield prediction (Nain et al., 2001).
Use of crop models in climate change studies
Analysis of impact of climate change on crop growth and yield can be carried out for individual
and interaction effects of elevated temperature, rainfall, CO2, etc. But these studies indicate the
individual and interactional influence of various parameters irrespective of temporal scale.
However, by using the climate scenarios, either derived from Global Climate Models (GCM) or
from Regional Climate Models (RCMs), as inputs into the crop models, quantification of
impacts on economic yields can be carried out for future climates. The adaptation analysis can
be done by quantifying the response of different varieties, sowing time, nutrient management,
water management, introduction of new crops, shift in cropping sequences, altered resource
management and introduction of new technologies, etc. in various climate change scenarios

160
so as to derive the best suitable technology package for reducing impacts of climate change at
regional level then up-scaling to state and national level. These are called adaptation gains. The
net different between impacts and adaptation gains is called net vulnerability of crop/system to
climate change. Using the above approach, several studies have been conducted for quantifying
the potential yields impacts, adaptation and vulnerability of coconut (Naresh Kumar et al., 2008;
Naresh Kumar and Aggarwal, 2009), maize (Byjesh et al., 2010), sorghum (Srivastava et al.,
2010) and also sensitivity of fragile ecosystems (Naresh Kumar et al., 2011).
Fig 4: Sensitivity analysis on integration of two important factors influencing crop growth–point
based simulations

(source:
Srivastava
et al., 2010)

Fig 5: Scenario based projections on impacts of climate change on crop production–spatial

integration
(source: Naresh Kumar et al., 2011)

Researchable issues

· Fine tune models based on updated thresholds of factors influencing major processes

161
 · Fine tune/modify the models to best represent multiple stress impacts on crops in a
season
References
Aggarwal, P.K., Kropff, M.J., Cassman, K.G. and Ten Berge, H.F.M. (1997). Simulating the
genotypes strategies for increasing rice yield potential in irrigated tropical environments.
Field Crop Res., 51: 5-17.
Byjesh, K., S. Naresh Kumar and P. K. Aggarwal (2010). Simulating impacts, potential
adaptation and vulnerability of maize to climate change in India. Mitigation and
Adaptation Strategies for Global Change. DOI 10.1007/s11027-010-9224-3; 15:413-431.
Kalra, N. and Aggarwal, P.K. (1994). Evaluating water production functions for yield
assessment in wheat using crop simulation models. In: ten Berge, H.F.M., Wopereis,
M.C.S. and Shin, J.C. (Eds.), Nitrogen Economy of Irrigated Rice: Field of Simulation
Studies, SARP Research Proceedings, AB-DLO, Wageningen, pp. 254- 266.
Kalra, N. and Aggarwal, P.K. (1996). Evaluating the growth response of wheat under varying
inputs and changing climate options using wheat growth simulator –WTGROWS. In:
Abrol, Y.P., Gadgil, S. and Pant, G.B. (Eds.), Climate variability and agriculture. Narosa
Publishing House, New Delhi, pp. 320-338.
Nain, A.S., Dadhwal, V.K. and Singh, T.P. (2002). Real time wheat yield assessment using
technology trend and crop simulation model with minimal data set. Curr. Sci., 82(10):
1255-1258.
Nain, A.S., Vyas, S.P., Dadhwal V.K., and Singh T.P. (2001). Farm level wheat yields
prediction using remote sensing data. In: Proceedings of National Symposium on
Advances in Remote Sensing with Emphasize on High Resolution Imageries, 11-13
December, 2001, ISRS, SAC, Ahmedabad, India.

Naresh Kumar, S., Kasturi Bai, K. V. Rajagopal V. and Aggarwal, P.K. 2008. Simulating
coconut growth, development and yield using InfoCrop-coconut model. Tree Physiology,
28:1049–1058.

Naresh Kumar, S. and P. K. Aggarwal, 2009. Impact of climate change on coconut plantations.
In Global Climate Change and Indian Agriculture-case studies from ICAR Network
Project (PK Aggarwal ed.), ICAR, New Delhi Pub., pp.24-27.
Naresh Kumar, P. K. Aggarwal, Swaroopa Rani, Surabhi Jain, Rani Saxena and Nitin Chauhan
(2011). Impact of climate change on crop productivity in Western Ghats, coastal and
northeastern regions of India. Current Sci. 101 (3):33-42.
Rivington, M. and Koo, J. 2010. Report on the Meta-Analysis of Crop Modelling for Climate
Change and Food Security Survey. CCAFS study. P 70.
162
Singh, K.K., Kalra, N., Mohanty, U.C. and Rathore, L.S. (1997). Performance evaluation of
medium range weather forecast using crop growth simulator. J. Environ. Syst., 25(4):
397-408.
Srivastava, A., S. Naresh Kumar and P. K. Aggarwal (2010). Assessment on vulnerability of
sorghum to climate change in India. Agric. Ecosyst. Environ.
Doi:10.1016/j.agee.2010.04.012; . 138:160-169.

163
 Statistical tools for crop weather modeling

BMK Raju
Senior Scientist
Section of Design & Analysis

Linear Regression

Concept

• Regression’ is ‘stepping back or returning to the average value’.

• Regression: measure of the average relationship between two or more variables in terms
of the original units of the data.
• Regression analysis: estimates are made for a variable from knowledge of the values of
one or more other variables.
• To study the functional relationship between the variables and thereby provide a
mechanism for prediction or forecasting’.
• Regression analysis is an attempt to establish the ‘nature of the relationship between
variables

Dependent vs Independent variables

• In regression analysis, there are two types of variables. The variable, which is used to
predict the variable of interest, is called the independent variable and the variable we are
trying to predict is called the dependent variable.
• In regression analysis independent variable is also known as regresser or predictor or
explanatory while the dependent variable is also known as regressed or explained
variable or criterion.
• The independent variable is denoted by X and dependent variable by Y. Changes in Y
are assumed to be caused by changes in X. Relationship between X and Y is described by
a linear function

Simple Vs Multiple Linear Regression

• Simple regression: Study of only two variables at a time

164
 • But quite often the values of a particular phenomenon may be affected by multiplicity of
factors.
• Multiple Regression: Studying more than two variables at a time

Correlation Vs Regression

• Correlation is only concerned with strength of the linear relationship. No functional

relationship.
• No causal effect is implied with correlation.
• Non-directional.
• No units for Correlation co-efficient.
• Range is -1 to +1

Regression Equation

Y = b + b X +e
0 1
• where Y is dependent variable
• X is independent variable

b0 is intercept

b1 is regression coefficient

e is random error

Interpretation of Regression Coefficient

• Regression coefficient explains the impact of changes in an independent variable on the

dependent variable.

• Say estimated value of b0 is b0 . Then b0 is the estimated average value of Y

when the value of X is zero (if X = 0 is in the range of observed X values).

• Say estimated value of b1 is b1 . Then b1 measures the estimated change in the

average value of Y as a result of a one-unit change in X.

165
 å ( xi - x)( yi - y )
b1 =
å ( xi - x)
2

Yi = β 0 + β1X1i + β 2 X 2i + K + β k X ki + ε
b0 = y - b1 x

Where x and y are averages of X and Y variables, respectively.

Multiple Linear Regression Model

Multiple Regression Model with k Independent Variables will be

Yi = β 0 + β1X1i + β 2 X 2i + K + β k X ki + ε

Two variable model

Y
Ŷ = b0 + b1X1 + b 2 X2

X2

X1

F-Test for Overall Significance

H 0: β 1 = β 2 = 0
H1: β1 and β2 not both zero

MSR
Test statistic F=
MSE

166
Are Individual Variables Significant?

• Use t-test for individual variable slopes

• It tests if there is a linear relationship between the variable Xi and Y

Hypotheses:
H0: βi = 0 (no linear relationship)
H1: βi ≠ 0 (linear relationship does exist between Xi and Y)

bi - 0
t=
Test Statistic:
Sb i

(df = n – k – 1)
Step-wise Regression
Backward Regression:
Begins with all variables and drops variables one by one based on insignificance of regression
coefficients
(Eg: Removal p>0.1)
Forward Regression:
Begins with most significant variable and adds one by one based on significance of regression
coefficients (Eg: Entry p<0.05)
Cluster analysis
• Given a set of p variables X1, X2,…, Xp, and a set of N objects, the task is to group the
objects into classes so that objects within classes are more similar to one another than to
members of other classes.
• Objects that are similar to one another should be in the same group, whereas objects that
are dissimilar should be in different groups.
• All cluster analyses begin with measures of similarity/dissimilarity among objects
(distance matrices)
Distance matrix
Objects that are closer together based on pairwise multivariate distances are assigned to the
same cluster, whereas those farther apart are assigned to different clusters.

Some clustering distances

167
Distance metric Description Data type

Pearson 1- r for each pair of objects quantitative

R2 1 – r2 for each pair of objects quantitative

Euclidean Normalized Euclidean distance quantitative

c2 c 2 measure of independence of counts

rows and columns on 2 X N

frequency tables

MW Increment in SS within if object quantitative

moved into a particular cluster

Scale considerations

• In general, correlation measures are not influenced by differences in scale, but distance
measures (e.g. Euclidean distance) are affected.
• So, use distance measures when variables are measured on common scales, or compute
distance measures based on standardized values when variables are not on the same
scale.
• Remove outliers in the data

Hierarchical clustering of objects

• Begins with calculation of distances among all pairs of objects…with groups being
formed by agglomeration (lumping of objects)
• The end result is a dendogram (tree) which shows the distances between pairs of objects.

168
Hierarchical joining algorithms

• Single (nearest-neighbour): distance between two clusters = distance between two

closest members of the two clusters.
• Complete (furthest neighbour): distance between two clusters = distance between two
most distant cluster members.
• Centroid : distance between two clusters = distance between multivariate means of each
cluster.
• Average: distance between two clusters = average distance between all members of the
two clusters.
• Median: distance between two clusters = median distance between all members of the
two clusters.
• Ward: distance between two clusters = average distance between all members of the two
clusters with adjustment for covariances.

K – means clustering

A method of partitioned clustering whereby a set of k clusters is produced by minimizing the

SSwithin based on Euclidean distances.

In K – means clustering, the objective is to partition a set of N objects into a number k

predetermined clusters by maximizing the distance between cluster centers while minimizing the
within-cluster variation.

• Choose k “seed” cases which are spread apart from center of all objects as much as
possible.
• Assign all remaining objects to nearest seed.
• Reassign objects so that within-group sum of squares is reduced…
• …and continue to do so until SS within is minimized.

Principal Component Analysis

Most of the times, the variables under study are highly correlated and as such they are
effectively “saying the same thing”.

169
Purpose of PCA

• Dimensionality reduction
– Small number of uncorrelated hidden or underlying variables
– Principal components are linear combinations of original set of variables
– Decreasing order of importance
• To tackle with multi-colliniarity

Method

Let x1, x2, x3, . . . , xp be original variables, then first principal component may be defined as

z1 = a11 x1 + a12 x2 + ... + a1 p x p

such that variance of z1 is as large as possible subject to the condition that

a112 + a122 + ... + a12p = 1

This constraint is introduced because if this is not done, then Var(z1) can be increased simply
by multiplying any a1j’s by a constant factor.

The second principal component is defined as

z2 = a21 x1 + a22 x2 + ... + a2 p x p

such that Var(z2) is as large as possible next to Var(z1) subject to the constraint that

a212 + a222 + ... + a22 p = 1

and cov(z1, z2) = 0 and so on.

• It is quite likely that first few principal components account for most of the variability in
the original data. If so, these few principal components can then replace the initial p
variables

170
• An analysis of principal components often reveals relationships that were not previously
suspected

171
 An Introduction to APSIM Model

P.Vijaya kumar
Senior Scientist (Ag.Met),
CRIDA,Hyderabad

1. Introduction

The Agricultural Production Systems Simulator (APSIM) has been used in a broad range

of applications, including support for on-farm decision making, farming systems design for

production or resource management objectives, assessment of the value of seasonal climate

forecasting, analysis of supply chain issues in agribusiness activities, development of waste

management guidelines, risk assessment for government policy making and as a guide to

research and education activity.

APSIM has been developed by the Agricultural Production Systems Research Unit

(APSRU), a collaborative group made up from CSIRO and Queensland State Government

agencies. Development started with the formation of APSRU in 1991 and improvement or

updation of the model has been continuing for the last two decades.

APSIM was designed at the outset as a farming systems simulator that sought to combine

accurate yield estimation in response to management with prediction of the long-term

consequences of farming practice on the soil resource (e.g. soil organic matter dynamics,

erosion, acidification etc.).

The central concept of APSIM is “the soil provides a central focus, crop season and

managers come and go, finding the soil in one state and leaving it in another”

2. Overview of the APSIM system and its components

The APSIM modelling framework is made up of;

a) a set of biophysical modules that simulate biological and physical processes in farming

systems,

172
b) a set of management modules that allow the user to specify the intended management rules

that characterise the scenario being simulated and that control the conduct of the simulation

c) various modules to facilitate data input and output to and from the simulation,

d) a simulation engine that drives the simulation process and controls all messages passing

between the independent modules.

3. Details of APSIM components

3.1. Crop Modules

APSIM contains an array of modules for simulating growth, development and yield of crops,

pastures and forests and their interactions with the soil. Currently crop modules are available for

barley, canola, chickpea, cotton, cowpea, hemp, fababean, lupin, maize, millet, mucuna,

mungbean, navybean, peanut, pigeonpea, sorghum, soybean, sunflower, wheat and sugarcane.

Modules for forest trees, rice, forage grasses and lablab under development.

The plant modules simulate key underpinning physiological processes and operate on a

daily time step in response to input daily weather data, soil characteristics and crop management

actions. The crop modules have evolved from early versions for focus crops such as maize

173
(Carberry and Abrecht, 1991), peanut (Hammer et al., 1995), sorghum (Hammer and Muchow,

1991) and sunflower (Chapman et al., 1993).

Currently in APSIM, all plant species use the same physiological principles to capture resources

and use these resources to grow.

Processes captured:

The seven important processes captured in this model are:

(1) Phenology

(2) Tillering and leaf area production

(3) Biomass accumulation and partitioning

(4) Root growth

(5) Crop water relations

(6) Crop nitrogen relations Senescence and plant death

Phenology Development:

Determinants of developmental changes are different. Germination is thermal time and

soil water dependent while emergence is dependent on thermal time and depth. End of juvenile

stage is influenced by thermal time and water and nitrogen stresses. Floral initiation, however,

depends on thermal time, photo period and stresses due to nitrogen and water limitations. Stages

from the start of grain fill to harvest stage are dependent on thermal time only.

Biomass accumulation:

In APSIM, biomass accumulation (DM) is simulated each day under both water limited

and radiation limited conditions using the equations

DM = Soil water supply x transpiration efficiency (Under water limited conditions)

DM = Radiation use efficiency x radiation intercepts (Under radiation limited conditions)

Minimum value of the above estimates is taken as biomass accumulation of that day.

Biomass Partitioning:

Partitioning of biomass to various plant organs is based on following criteria

Partitioning based on stage-specific ratios/fractions:

174
 · Root, leaf, stem, reproductive, grain

· Roots grown daily in stage-specific proportion to shoot

· Emergence to flowering: biomass partitioned to leaf & stem

· Start grain fill to maturity: grain +/- pod

Leaf area development:

It is calculated based on the formula

Leaf area/m2 (LAI) = plant density x axis no. x leaf no. per axis x area per leaf

· Leaf appearance driven by a variable thermal time rate

· Tiller appearance driven by leaf appearance or separate TT rate

· Lear area per leaf/node = f(Density, genotype, axis/tiller)

· Tiller senescence = f(age after FI, water depth)

· Leaf senescence rate = f(age, light competition, water stress, frost)

The routines in the library are structured in separate blocks corresponding to the crop model

components of phenology, biomass, canopy, root system, senescence pools, water, nitrogen and

phosphorus.

In APSIM there are modules for the two major modelling approaches that are commonly

used for the soil water balance, namely cascading layer and Richard’s equation methods.

SOILWAT (Probert et al., 1998c) is a cascading layer model that owes much to its precursors in

CERES (Ritchie, 1972; Jones and Kiniry, 1986)

and PERFECT (Littleboy et al., 1989, 1992). It operates on a daily time step. The water

characteristics of the soil are specified in terms of the lower limit (LL15), drained upper limit

(DUL) and saturated (SAT) volumetric water contents of a sequence of soil layers.

* runoff which is calculated using a modified USDA curve number approach, that include

effects of soil water content, soil cover both from crop and crop residue, and roughness due to

tillage.

175
* evaporation which is based on potential evaporation (Priestly_/Taylor or Penman - Monteith)

and modified according to the cover provided by surface residues or growing plant

* saturated flow which occurs when any layer ‘fills’ above DUL; a specified proportion (swcon)

of the water in excess of DUL drains to the next layer

* unsaturated flow at water contents below DUL where gradients in soil water content occur

between layers (e.g. in response to rainfall events or evaporation)

* movement of solutes associated with saturated and unsaturated flow of water are calculated

using a ‘mixing’ algorithm whereby existing and incoming solutes and water are fully mixed to

determine the concentration of solute in the water leaving any layer.

4. MANAGER

The early recognition that all the possible management configurations required of the simulator

could not be explicitly identified and addressed a priori, led to the development of the

MANAGER module in APSIM. This module enables users to apply simple concepts of states,

events, actions and conditional logic to build complex management systems whose scope goes

well beyond anything envisaged by the early developers. The MANAGER must be present in all

APSIM configurations and it provides control over individual components and the overall

simulation. This module ‘manages’ by issuing messages to other modules in the system, many of

which are conditional upon states or events within the modules during simulation. It also allows

the user to create their own variables and define these as a function of other variables within

APSIM. The MANAGER script files are prepared by users defining the intended simulation and

are compiled at runtime.

The APSIM MANAGER module can be used to invoke any action available by any module.

Possible actions include:

- Resetting individual module values.

- Reinitialising all data in modules to a given state.

- Sowing, harvesting or killing crops.

- Applications of fertiliser, irrigation or tillage to soil.

176
- Calculation of additional variables to track system state.

- Reporting of system state in response to events and/or conditional logic.

Reference

Carberry, P.S., Abrecht, D.G., 1991. Tailoring crop models to the semi-arid tropics. In:

Muchow, R.C., Bellamy, J.A. (Eds.), Climatic Risk in Crop Production: Models and

Management for the Semiarid Tropics and Subtropics. CAB International, Wallingford, UK, pp.

157-/182.

Chapman, S.C., Hammer, G.L., Meinke, H., 1993. A sunflower simulation model: I. Model

development. Agronomy J. 85, 725 - 735.

Hammer, G.L., Muchow, R.C., 1991. Quantifying climatic risk to sorghum in Australia’s

semiarid tropics and subtropics: model development and simulation. In: Muchow, R.C.,

Bellamy, J.A. (Eds.), Climatic Risk in Crop Production: Models and Management for the

Semiarid Tropics and Subtropics. CAB International, Wallingford, UK, pp. 205 - 232.

Hammer, G.L., Sinclair, T.R., Boote, K.J., Wright, G.C.,Meinke, H., Bell, M.J., 1995. A peanut

simulation model: 1. Model development and testing. Agronomy J. 87, 1085-1093.

Jones, C.A., Kiniry, J.R. (Eds.), CERES-Maize: a simulation model of maize growth and

development. Texas A&M University Press, College Station 1986, p. 194.

Littleboy, M., Silburn, D.M., Freebairn, D.M., Woodruff, D.R., Hammer, G.L., 1989.

PERFECT*/A computer simulation model of Productivity Erosion Runoff Functions to Evaluate

Conservation Techniques. Queensland Department of Primary Industries Bulletin, QB89005.

Probert, M.E., Robertson, M.J., Poulton, P.L., Carberry, P.S., Weston, E.J., Lehane, K.J., 1998a.

Modelling lucerne growth using APSIM. Proceedings of the Ninth Australian Agronomy

Conference, Wagga Wagga, pp. 247 - 250.

Probert, M.E., Carberry, P.S., McCown, R.L., Turpin, J.E., 1998b. Simulation of legume-cereal

systems using APSIM. Aust. J. Agric. Res. 49, 317 - 328.

177
Probert, M.E., Dimes, J.P., Keating, B.A., Dalal, R.C., Strong, W.M., 1998c. APSIM’s water

and nitrogen modules and simulation of the dynamics of water and nitrogen in fallow systems.

Agric. Syst. 56, 1 - 28.

Ritchie, J.T., 1972. A model for predicting evaporation from a row crop with incomplete cover.

Water Resour. Res. 8, 1204- 1213.

178
 Fundamentals of DSSAT Model
AVM Subba Rao
Scientist (Agmet)

Crop modeling enables researchers to integrate knowledge from different disciplines in a

quantitative way. That, in turn, helps researchers to understand the underlying processes that
determine the behavior of complex agricultural systems. Mathematical models are caricatures of
systems made from mathematical equations. Integrating and solving the equations enables a
numerical description of the system to be produced. During the first phase of a modeling
exercise, the modeler seeks to give names, magnitudes and units to the component parts of the
problem. In the second phase of modeling a problem, the processes are described as
mathematical functions. In the final phase, 'what if' questions can be asked about the functioning
of a system and numerical answers provided. Mathematical models that contain no clear logical
link with the basic processes governing the relationship between the system inputs and outputs
are unlikely to contribute much of significance to any debate concerning strategic decisions in
relation to research management.

Models at different levels of detail are developed to meet different objectives, ranging from a
thorough understanding of an existing system to the prediction of crop production in untested
conditions. Four types of crop production systems can be distinguished.

1. Potential production, where production is determined by solar radiation,

temperature, and crop and varietal characteristics.
2. Water-limited production.
3. Water- and N-limited production.
4. Water-, N- and other nutrient-limited production.

Going from type one to type four, production generally decreases and the variables that
determine system behavior increase. At all levels, growth-reducing factors such as insects,
pathogens, and weeds can be introduced. Models for all production levels can be developed.
Models at the first level are further developed than models at the others.

Well-developed models that simulate the growth of a crop in relation to its dynamic environment
can be used to help prioritize research. Crop modeling combined with geographic information
systems (GIS) analysis enables researchers to distinguish agro ecological zones and to
quantitatively rank the technical constraints to agricultural production within them. These
models allow the impact of new technology on agricultural production to be assessed before the
technology is introduced. The GIS database can link the models directly with socioeconomic
aspects.

Crop simulation models have many uses. Models can be used as a research tool and to support
problem solving, risk assessment, and decision making. They can guide researchers in
prioritizing their research and in integrating quantitative knowledge from different disciplines.
Also, models can be used as a framework for training. Further, models can be used to extrapolate
research findings over broad regions and extended time, since the models account for crop-
environment interactions. Using long-term weather data, yield probabilities can be simulated.

An aspect that is beginning to gain more importance is the use of models to set breeding goals.
The physiological attributes that contribute significantly to crop production in a given
environment lend themselves to definition by crop modeling. Through modeling, the optimum

179
timing of seeding or transplanting, irrigation and fertilization can be determined for a given
environment.

Modeling is especially useful in yield gap analysis, a method for identifying constraints to
agricultural production in different agroclimatic zones. From yield gap analysis, constraints that
can be reduced can be identified. Researchers then concentrate on ameliorating those factors that
contribute to the gap between farm yield, potential farm yield, and potential experiment station
yield

DSSAT model

The decision support system for Agrotechnology transfer (DSSAT) was originally
developed by an international network of scientists, cooperating in the International
Benchmark Sites Network for Agrotechnology Transfer project to facilitate the application
of crop models in a systems approach to agronomic research. Its initial development was
motivated by a need to integrate knowledge about soil, climate, crops, and management for
making better decisions about transferring production technology from one location to
others where soils and climate differed. The systems approach provided a framework in
which research is conducted to understand how the system and its components function. This
understanding is then integrated into models that allow one to predict the behavior of the system
for given conditions. After one is confident that the models simulate the real world
adequately, computer experiments can be per- formed hundreds or even thousands of
times for given environments to determine how to best manage or control the system.
DSSAT was developed to operationalized this approach and make it available for global
applications. The DSSAT helps decision-makers by reducing the time and human resources
required for analyzing complex alternative decisions. It also provides a framework for scientific
cooperation through research to integrate new knowledge and apply it to research questions.

The DSSAT is a collection of independent programs that operate together; crop simulation
models are at its center (Fig. 1). Databases describe weather, soil, experiment
conditions and measurements, and genotype information for applying the models to different
situations. Software helps users prepare these databases and compare simulated results with
observations to give them confidence in the models or to determine if modifications are needed
to improve accuracy.

180
 Fig.1: Basic Structure of DSSAT model

Different types of applications are accomplished in DSSAT modal by using different modes to
call the land unit module on a daily basis; the mode is specified as a command line argument
when the model is run. The basic mode provides for inter- active sensitivity analysis and
comparison of simulated vs. observed field data. A second mode of operation simulates crops
over a number of years of weather using the same soil initial condi- tions. This mode allows
one to evaluate the effects of uncertain future weather conditions on deci- sions made
when all soil initial conditions are known. A third mode operates the cropping system
modules to simulate crop rotations over a number of years, and soil conditions are initialized
only at the very start of the simulation. A fourth mode operates the CSM to simulate one or
more crops over space (i.e. for precision agriculture, land use management or other spatial-
based applications). One can also completely replace the main driver for other applications,
thereby providing a highly flexible approach for development of additional applications and
user interfaces without having to modify code for any other module (fig.2).

181
 Fig.2: Overview of components and modules in DSSAT

The primary and sub modules currently used in the CSM and summarizes their functions
are given in table 1.
Table.1: Summary description of modules in the DSSAT — CSM
Modules Sub modules Behavior

182
Main Controls time loops,, determines which
program modules to call based on user input switches,
(DSSAT — controls print timing for all modules
CSM)
Land unit Provides a single interface between cropping
system behavior and
applications that control the use of the
cropping system. It serves as a collection
point for all components that interact on a
homogenous area of land
Weather Reads or generates daily weather parameters
used by the model.
Adjusts daily values if required, and computes
hourly values
Soil Soil dynamics Computes soil structure characteristics by
layer. This module currently reads values
from a file, but future versions can modify
soil properties in response to tillage, etc
Soil temperature module Computes soil temperature by layer
Soil water module Computes soil water processes including
snow accumulation and melt, runoff,
infiltration, saturated flow and water table
depth. Volumetric soil water content is
updated daily for all soil layers. Tipping
bucket approach is used
Soil nitrogen and carbon module Computes soil nitrogen and carbon
processes, including organic and inorganic
fertilizer and residue placement,
decomposition rates, nutrient fluxes between
various pools and soil layers. Soil nitrate and
ammonium concentrations are updated on a
daily basis for each layer
SPAM Resolves competition for resources in soil
— plant — atmosphere system. Current
version computes partitioning of energy
and resolves energy balance processes for
soil evaporation, tran- spiration, and root
water extraction
CROPGRO module
Crop Individual plant growth modules
Template
183
 Computes crop growth
processes including
phenology, photo-
synthesis, plant nitrogen
CERES- and carbon demand,
Maize; growth partition- ing, and
CERES- pest and disease damage for
Wheat; crops modeled using the
CERES- Rice; CROPGRO model Crop
SubStor- Template (soybean, peanut,
Potato; Other dry bean, chickpea,
plant models cowpea, faba bean, tomato,
Macuna, Brachiaria,
Bahiagrass)
Modules that simulate
growth and yield for
individual species. Each is a
separate module that
simulates phenology, daily
growth and partitioning,
plant nitrogen and carbon
demands, senescence of
plant material, etc
Management Planting Determines planting date based on read-in
opera- tions value or simulated using an input planting
module window and soil, weather conditions
Harvesting Determines harvest date, based on maturity,
read-in value or on a harvesting window
along with soil, weather conditions
Irrigation Determines daily irrigation, based on read-in
values or automatic applications based on soil water depletion
Fertilizer Determines fertilizer additions, based on
read-in values or automatic conditions
Residue Application of residues and other organic mMaterial (plant,
animal) as read-in values

184
MINIMUM DATA SETS

The DSSAT models require the minimum data set for model operation. The contents of such a
dataset have been defined based on efforts by workers in IBSNAT and ICASA. They
encompass data on the site where the model is to be operated, on the daily weather during the
growing cycle, on the characteristics of the soil at the start of the growing cycle or crop
sequence, and on the management of the crop (e.g. seeding rate, fertilizer applications,
irrigations) (Table.2)

Contents of minimum data sets for operation and evaluation of the DSSAT — CSM
(a) For operation of model
Site Latitude and longitude, elevation; average annual temperature; average annual
amplitude in temperature
Slope and aspect; major obstruction to the sun (e.g. nearby mountain); drainage
(type, spacing and depth); surface stones (coverage and size)
Weather Daily global solar radiation, maximum and minimum air temperatures,
precipitation
Soil Classification using the local system and (to family level) the USDA-NRCS
taxonomic system
Basic profile characteristics by soil layer: in situ water release curve
characteristics (saturated drained upper limit, lower limit); bulk density, organic
carbon; pH; root growth factor; drainage coefficient
Initial Previous crop, root, and nodule amounts; numbers and effectiveness of rhizobia
condi- (nodulating crop)
tions
Water, ammonium and nitrate by soil layer

185
 Management Cultivar name and type
Planting date, depth and method; row spacing and
direction; plant population Irrigation and water
management, dates, methods and amounts or depths
Fertilizer (inorganic) and inoculant applications
Residue (organic fertilizer) applications (material, depth of incorporation,
amount and nutrient concentrations) Tillage
Environment (aerial) adjustments
Harvest schedule
(b) For evaluation of models
Date of emergence
Date of flowering or pollination (where appropriate)
Date of onset of bulking in vegetative storage organ
(where appropriate) Date of physiological maturity
LAI and canopy dry weight at three stages during the life cycle
Canopy height and breadth at maturity
Yield of appropriate economic unit (e.g. kernels) in dry weight terms
Canopy (above ground) dry weight to harvest index (plus
shelling percentage for legumes) Harvest product individual
dry weight (e.g. weight per grain, weight per tuber)
Harvest product number per unit at maturity (e.g. seeds per
spike, seeds per pod) Harvest product number per unit at
maturity (e.g. seeds per spike, seeds per pod) Soil water
measurements vs. time at selected depth intervals
Soil nitrogen measurements vs. time
Soil C measurements vs. time, for long-term experiments
Damage level of pest (disease, weeds, etc.) infestation (recorded when
infestation first noted, and at maximum) Number of leaves produced on the
main stem
N percentage of economic unit
N percentage of non-economic parts
In addition to research applications, the DSSAT and its crop models have been used in
teaching, both in continuing education courses and in formal university courses at
graduate and under- graduate levels (Tsuji et al., 1998). There also have been attempts to use
these models in advising farmers (through extension services and the pri- vate sector). In
one application, described by Welch et al. (2002), an agricultural company has implemented
versions of three of the DSSAT v3.5 models in a comprehensive farmer support soft- ware
package that is being used by private consultants. This software package, called PCYield,
includes CROPGRO-Soybean, CERES- Maize and CERES-Wheat models. PCYield is available
to clients of the company via the Internet along with daily weather data for specific farm
locations. It has a very simple user interface to allow private crop consultants to operate them
for any of their farmer clients (http://www.mPower3. com).

186
MODEL USES AND LIMITATIONS

Models are developed by agricultural scientists but the user-group includes the latter as
well as breeders, agronomists, extension workers, policy-makers and farmers. As different users
possess varying degrees of expertise in the modelling field, misuse of models may occur. Since
crop models are not universal, the user has to choose the most appropriate model according to
his objectives. Even when a judicious choice is made, it is important that aspects of model
limitations be borne in mind such that modelling studies are put in the proper perspective and
successful applications are achieved.

Agricultural systems are characterized by high levels of interaction between the components that
are not completely understood. Models are, therefore, crude representations of reality. Wherever
knowledge is lacking, the modeler usually adopts a simplified equation to describe an extensive
subsystem. Simplifications are adopted according to the model purpose and / or the developer’s
views, and therefore constitute some degree of subjectivity.

Models that do not result from strong interdisciplinary collaboration are often good in the area of
the developer’s expertise but are weak in other areas. Model quality is related to the quality of
scientific data used in model development, calibration and validation

When a model is applied in a new situation the calibration and validation steps are crucial for
correct simulations. The need for model verification arises because all processes are not fully
understood and even the best mechanistic model still contains some empirism making parameter
adjustments vital in a new situation.

Model performance is limited to the quality of input data. It is common in cropping systems to
have large volumes of data relating to the above-ground crop growth and development, but data
relating to root growth and soil characteristics are generally not as extensive. Using
approximations may lead to erroneous results. Large variations in wheat yields (4.5 to 8.0 t ha-1)
attributable to within-field soil heterogeneity were reported by Russell and Van Gardingen
(1996). Hence, the use of average values of soil characteristics as model inputs could lead to
some errors in simulated output.

Most simulation models require that meteorological data be reliable and complete.
Meteorological sites may not fully represent the weather at a chosen location. In some cases,
data may be available for only one (usually rainfall) or a few (rainfall and temperature)
parameters but data for solar radiation, which is important in the estimation of photosynthesis
and biomass accumulation, may not be available. In such cases, the user would rely on generated
data. At times, records may be incomplete and gaps have to be filled. Using approximations
would have an impact on model performance. Nonhebel (1994) has reported that simulated
wheat yield was overestimated under potential conditions and underestimated under water-
limiting conditions when generated meteorological data were used with SUCROS87 (Spitters et
al. 1989).

187
At times, model developers may raise the expectations of model users beyond model
capabilities. Users, therefore, need to judiciously assess model capabilities and limitations before
it is adopted for application and decision-making purposes.

Generally, crop models are developed by crop scientists and if interdisciplinary collaboration is
not strong, the coding may not be well-structured and model documentation may be poor. This
makes alteration and adaptation to simulate new situations difficult, specially for users with
limited expertise. Finally, using a model for an objective for which it had not been designed or
using a model in a situation that is drastically different from that for which it had been
developed would lead to model failure

Further details and complete information on DSSAT functionality read the paper

“ J.W. Jones et al. 2003.The DSSAT cropping system model, Europ. J. Agronomy 18 (2003)
35 — 265”

188
189
190
191
192
193
194
195
196
197
198
 MODELING INSECT POPULATIONS

Y.G. Prasad
Principal Scientist (Entomology), CRIDA, Hyderabad
[email protected]

1.0 INTRODUCTION

Modeling insect populations is important for understanding and predicting their

population dynamics. Insect pests are influenced by both macro and micro-weather parameters:
temperature, rainfall, humidity, sunshine hours, wind speed and direction. The rates at which
insects complete their life cycles depend mainly on temperature, so that the times of activity of a
given pest insect can vary greatly both from region to region and from year to year. In addition
to influence of weather, insect pest appearance and regulation of numbers are governed by other
interactions with the availability of susceptible plant hosts and their natural enemies such as
parasitoids and predators. Parameters along with the type of influence they exert on insect
abundance are given in Table 1.

Table 1. Parameters influencing insect abundance

Category Parameter Type of influence on insects

Weather Temperature (maximum and minimum) Insect developmental rate, diapause
Humidity (morning and evening) Oviposition, egg hatch
Rainfall (amount and distribution) Adult emergence, oviposition, wash-out
Bright sunshine hours Oviposition
Wind speed and direction Dispersal and migration
Microclimatic parameters (canopy Development and rate of spread
temperature, humidity, leaf wetness, soil
temperature)
Crop Genotype Rate of development, foci of infestation
attributes
Phenology Pest appearance and intensity
Crop stand Dispersal
Nutrition Fecundity
Time of sowing Pest onset
Cropping system/alternate hosts/previous Off-season survival, carry over
crop
Pest Life stages and their duration Generation number and duration
attributes
Conditions for stage transfer Diapause
Timing of reproduction Pest onset
Fecundity Pest abundance
Mortality factors Pest regulation

199
2.0 INSECT POPULATION MODELING – APPROACHES

Important points for consideration in any model development are: the level of
detail at which a given model is to be developed as the level of detail is linked to the
objective and data availability to develop and run the model. Models can range from
strictly empirical to most complex and sophisticated descriptive models. A model
may be discrete or continuous, static or dynamic, and deterministic or stochastic.

2.1 EMPIRICAL APPROACHES

Empirical approaches involve estimating pest and disease incidence and intensity
through experimentation and surveys on crops not subjected to control interventions and
establishing relationships with concurrent, prevailing weather and/or past weather factors. The
studies could be conducted at single stations in which the emphasis is on delineation of
differences in meteorological conditions in epidemic and non-epidemic years or multi-station
studies in which the emphasis is on delineation of meteorological conditions leading to changes
in periods and intensity of infestations. A multi-station study is preferred as it facilitates
corroboration of the general surmises and leads to maximization of data in a short period if
observations are recorded on crop stands sown at periodic intervals at a number of stations
(Venkataraman and Krishnan, 1992). It should be noted that findings from empirical field
studies can straight away be applied in climatologically analogous areas but can give misleading
results when applied to other areas.
Development of an empirical forecast model is not an end in itself. Even the simplest
model must be tested to be proven, but validation over a wide range of conditions will be most
important for models based on empirical rather than biological and physical processes, or where
there is insufficient understanding and quantification of how interactions change under varying
environmental conditions. Any type of forecast model needs to be fully described for running
the model, correct interpretation of the output and its effective dissemination and operational
use. Synthesis of model elements into a computer program would be an ideal logical step to
make available a product for operational use in agro-advisories.
Many empirical models use various types of pest/disease incidence data (trap
catches, population counts and crop damage assessments). Many research articles
published on pest-weather relationships used pest monitoring data from light traps
(for example yellow stem borer in rice), pheromone traps (for American bollworm)
and sticky traps (for whitefly) apart from population counts and damage
assessment data. Long-term data is preferable as it better captures the patterns in
relationships. These models also require access to weather and climate data, in
addition to pest and plant data. Models usually require as inputs, measurements of
temperature, rainfall and humidity, although other variables may be required
either as direct inputs or in computing values for variables not measured. Weather
variables need to be measured at the field level, at regional stations, or on a
broader scale depending on the need. For many farm management actions, data
representative of the field conditions are expected and hence data is taken from
automatic weather stations or the nearest observatory.

200
2.2 INSECT PHENOLOGY MODELS

Insect phenology modeling is based on insect life cycles. These models are
developed using temperature data to forecast timing of insect activity. Insects have
different development stages, the developmental durations of which are based on
temperature above a lower developmental threshold known as base temperature.
Different insect stages can have different lower thresholds (Table 2). A day-degree
is defined as one degree of average temperature above a base temperature over a 24
h period. The duration of the insect stage is expressed in day-degrees. Once the
required day-degrees are completed, the insect moves into the next stage in its
development. Hence, expression of insect development in day-degrees gives it a
mathematical expression and is better than using calendar days. Relatively crude
methods of computing day-degrees are sufficient for many applications. Most
forecasting models on diurnal variation rely on approximations using sine waves.
Degree days are accumulated from a start date known as ‘bio-fix’. These start dates
are generally based on the first adult trap catch using pheromone traps or first
notice of eggs in field or planting date etc. The degree day approach allows the
prediction of the biological events in the insect development using temperature
data. The stage-specific thermal constants are arrived at by studying insect
development at several constant temperatures. Often the lower threshold is an
approximation by extrapolating the linear portion of the development rate curve.
Day-degree forecasts, in general, cannot readily predict insect populations that
have overlapping generations in a year. In this case day-degree approach may be
restricted to the first or second generation of the pest. In temperate countries with
clear start and end of cropping seasons, insect modeling is majorly through insect
phenology models, However, in the tropics where cropping is year-round and winter
is not very severe (except in North India), phenology models can be accurately be
applied mostly to insect pests with one or few distinct generations in a year.

Table 2: Threshold temperatures and degree days for completion of different life
stages of Groundnut leaf miner life cycle

Life Temperature Degree Development Development duration

stage threshold Days duration (days) (days) at variable
(0C) at 300C constant temperatures
temperature (300C MaxT and 240C
MinT)
Egg 12.4 60 3.4 4.8
Larva 11.3 327 17.5 23.9
Pupae 14.7 72 4.7 7.0
Adult 3.0 202 7.5 9.2

2.3 DATA MINING APPROACHES

201
 Many times when statistical correlations and regressions are attempted through step-wise
regression or multiple regression models and applied to pest data and corresponding weather
data from several years, it is observed that in different years different weather parameters show
significant influence. The criteria for best fit model selection is based on the Co-efficient of
determination (R2) which explains the extent of variability in the insect population explained
with the independent factors (weather parameters) chosen. The R2 values in many models are
low and hence poor explain the variation and instill low confidence in the model for prediction
purposes. Data mining is a useful technique to bring out patterns in when long-term data has
been collected on pests, crops, their sowing times, cultivars, cropping pattern, insecticide use
along with weather factors. Similar to regression models, data mining technique also gives two
measures which help in fine tuning the assumptions or association rules made. One is the
measure of support and another is the measure of confidence. For example, we applied data
mining technique to study yellow stem borer population relationship with weather parameters.
An association rule was developed as “When rainfall is less than 8 mm and sunshine hours
greater than 8 hrs, possibility of moth catch greater than 100 has got a support of 40.2 percent
and confidence of 75.4 per cent”. In other words, nearly forty per cent of the cases in the data set
supported total weekly moth catches of more than 100, when the prevailing total rainfall in that
particular week was less than 8 mm and mean sunshine hours were more than 8 h. The high
confidence value (75.4%) showed that YSB adult emergence was strongly influenced by rainfall
and sunshine hours. Neural network model was developed by training the network with weekly
moth catch data and corresponding lag weather data from 1975 to 1996. Validation of neural
network model was carried out with the data of subsequent years i.e. 1997, 1998, 1999 and 2000
for kharif (Fig.1) and 1998, 1999 and 2000 for rabi. The time of occurrence of peaks and trend
in pest dynamics was well predicted in all the four years of study though the intensity of peaks
was not estimated accurately. This information could be very useful to forewarn peak moth
emergence activity of yellow stem borer in rice for the specific location.

Observed
Predicted
1800 1999
1600
1997
1400
YSB population

1200
1000 2000
800
600 1998
400
200
0
32 36 40 44 48 31 35 39 43 48 34 38 42 33 37 41 45
Standard week

202
Fig.1. Validation of neural network model on rice YSB in kharif season

2.4 MODELING SOIL MOSITURE FOR ESTIMATING SOWING TIMES AND

INFLUENCE ON INSECT POPULATIONS

A soil-moisture model was adopted for estimation of leaf miner severity in

Karnataka (Gadgil et al., 1999). In the model it is assumed that leaf miner
populations are always present at a low level. Whenever favorable weather
conditions occur in the appropriate growth stage of the crop, population builds up
rapidly. However, if a drenching shower (> 2 cm/day) occurs in the first 14 leaf
miner days (starting 35 days after sowing), it is assumed that the leaf miner is
eliminated. A leaf miner day has been defined as a non-rainy day with dry soil (<
half of the available soil moisture). In this model the loss in crop yield due to leaf
miner incidence is taken to depend upon the number of leaf miner days. The soil
moisture model estimates the sowing dates in a given region based on rainfall and
soil data and then estimates the leafminer days 35 days after sowing based on
model assumptions.

2.5 PROCESS BASED MODELING OFINSECT POPULATIONS – DYMEX

SOFTWARE

A review of research articles that were published in pest forecasting related aspects since
1980 within and outside India in rice and cotton, two crops of global importance, was
undertaken. A significant percentage of research effort has been directed towards studies on
monitoring and seasonal occurrence followed by studies that establish insect pest relationships
with weather in India in both the crops (Prasad, 2005). Despite the availability of a variety of
information that can become input to building reliable process forecast models, the trend reflects
a lack of concerted and directed research to develop process based models and decision support
systems in India vis-à-vis the trend abroad particularly in crops like cotton and rice where
pesticide use is still the highest in the country. The main limitation is the availability of
appropriate indigenous software packages that can consider dynamics of animal and plant
populations that are influenced by many factors, and understanding the response of a population
to a multitude of external factors can be very difficult.
Simulation models are a powerful means of representing such systems and allowing
users to interact with them. These models help to summarize our understanding of a species’
population dynamics, identify gaps in knowledge and enable rapid evaluation of management
options. Building population models, however, can be expensive in time, and may require
specialist programming skills. The DYMEX package is designed to overcome the bottleneck
caused by inadequate computer programming resources and modeling expertise. DYMEX
enables the user to build a class of ecological models referred to as mechanistic or process-based
models, without the need to know a computer programming language. It is a modular modelling
software package that consists of two parts: a Builder and a Simulator. The Builder is used to
create and modify the model, while the Simulator is used to run a completed model, and display
the results of simulations.

203
DYMEX is a computer software package that enables you to interactively build and then run
models of fluctuating populations of organisms in changing environments. Models are structured
around lifecycles, which in turn consist of the stages that individuals pass through during their
life. A DYMEX lifecycle describes cohorts of individuals and the processes that affect the size,
age and number of individuals in the cohort (individuals of same stage and age). Models created
within DYMEX consist of a series of modules, with each module responsible for a particular
task. Modules use information from other modules as input, and supply information to other
modules. DYMEX comes with a library of modules that can be incorporated into any model
constructed with the Builder. Each module performs a specific function (for example, MetBase
is used to read a standard set of meteorological variables from a file). Models created in the
Builder can be opened in the Simulator within which simulations can be run. The results of these
simulations can be displayed in tables, graphs and maps as well as exported to other programs.
Models will normally be developed around one or more Lifecycle modules. Other modules
provide data to the lifecycle modules, or manipulate lifecycle output in some way. Many
modules have multiple uses (e.g. Function module) and may be used in several places in a
model, while others are more specialised (e.g. the Soil Moisture module). Most modules receive
input fromanother module or from an outside source.

3.0 DECSION SUPPORT SYSTEMS AND AGROMETEOROLGOICAL NETWORKS

FOR PEST FORECAST AND ADVISORY SERVICES

East Germany began developing forecasting procedures for the occurrence of important
crop plant diseases and insect pests by designing simulation models (system PROGEB)
(Gutsche, 2001) in the 1980’s. Later after the German re-unification the project was run under
the name PASO which resulted in introduction of a number of forecasting models in practice
throughout Germany. To ensure stable operation of the forecast and decision support system by
the state crop protection service an keep the system open for innovations, the federal states set
up a ‘Central Service for Decision Support Systems and Programmes in Crop Protection’ which
is successfully being operated since 1998.
The Slovenian plant protection forecasting service runs an agrometeorological network
of 94 uniform weather stations in 7 centres to collect meteorological information (temperature,
humidity, precipitation and leaf wetness) automatically by radio. The data is analysed at the
central data collection facilities to determine pest risk and other farm operations. In addition to
weather station measurements, information from field monitoring, insect and spore traps,
observations of crop phenological phases was used in forecasting models for pests and diseases
processed by software (AgroExpert, ProPlant). The warnings were disseminated to growers by a
variety of means (Knapic et al, 2001)
In Norway, a web-based warning system called VIPS has been developed which
calculates warnings for than 70 weather stations for several pests and diseases in selected fruits,
vegetables and cereals. Warnings are site specific with validated meteorological data from an
authorized station and validated biological data necessary to run the models supplied by the
extension service (Folkedal and Brevig, 2001).
Yonow et al., (2004) developed a cohort-based life cycle model for the population
dynamics of the Queensland fruit fly using DYMEX model (Maywald et al., 1999), a process-
based, modular modelling software package that contains a library of modules. The model is
primarily driven by weather variables, and so can be used at any location where appropriate
204
meteorological data are available. DYMEX model helped to improve the understanding of
fruitfly population dynamics and relative abundance, and in so doing, identify critical gaps in
knowledge.
Using climatic modeling, risk of establishment of invasive species has been successfully
defined a priori in Europe (Sutherst et al., 1991). Samways et al., (1999) used the CLIMEX
model (Sutherst et al., 1995, 1999) and it’s associated ‘Match Climates’, climate-matching
algorithm to make their predictions of species geographical ranges. The CLIMEX model is a
simulation model of moderate complexity for inferring the responses of a species to climate
from its geographical distribution. Once response functions have been fitted, the model can be
run with meteorological data from other parts of the world to estimate the species response to
new climatic environments. The potential range, as determined by climate, can then be
estimated. The model parameter values constitute the hypotheses on the climatic factors that
determine the species population growth, and survival during adverse seasonal conditions, and
so limit the geographical distribution. Alternatively, the meteorological data base can be
manipulated to create scenarios of climate change.
Decision support systems are widely accepted in the Australian cotton industry for
assisting with integrated pest management, crop nutrition and other aspects of information
transfer. Uses of EntomoLOGIC, part of the CottonLOGIC software suite, select sample areas in
their cotton fields and collect information on the types of beneficial and pest insects present,
their stage of development and quantity. The hand held electronic device facilitates data entry
process, running models of pest development, generates in-field reports of pest status, access to
historical data on insects and crops. The software is then used to predict future pest numbers,
using weather data, and indicates when pest numbers are over defined economic thresholds for
crop managers to decide on appropriate pest management interventions (Bange et al., 2004).

CONCLUSIONS

Modeling insect populations requires adequate understanding of their life cycles and the
multitude factors that affect their population dynamics. Empirical models developed from long
term data capture the fluctuations in populations over seasons and years better than short period
data. These models are highly location specific and generally cannot be applied to other
locations except those which are climatologically and ecologically analogous. In contract models
based on insect life cycles such as phenology models are simple to construct and once
established can be run with temperature data. However, their application is limited to
monocyclic pests and regions which have distinct cropping seasons such as those in temperate
regions. Simulation models that are again based on life cycles and also take into account several
factors along with weather data are generally costly to develop. However, once developed they
can be used across locations for modeling the timing and intensity of pest attack.

REFERENCES
Bange M. P., Deutscher S.A., Larsen.D,.Linsley. D, Whiteside.S. 2004. A handheld decision
support system to facilitate improved insect pest management in Australian cotton
systems. Computers and Electronics in Agriculture 43: 131-147
Folkedal, A. and Brevig, C. 2001. VIPS – a web-based decision support system for crop
protection in Norway. In Proceedings of European Federation for Information
205
 Technology in Agriculture, Food and the Environment (EFITA) Congress 2001,
Montpellier, France (http://www.effita.net)
Gutsche, V. 2001. From mathematical models to decision support systems – the development of
the German plant protection forecasting sysem PASO. In Proceedings of European
Federation for Information Technology in Agriculture, Food and the Environment
(EFITA) Congress 2001, Montpellier, France (http://www.effita.net)
Knapic, V., Zmrzlak, M., Simoncic, A., Miklavc, J., Zezlina, I. and V. Skerlavaj. 2001.
Agrometeorological network and computer aided forecasting of plant protection in
Slovenia. In Proceedings of European Federation for Information Technology in
Agriculture, Food and the Environment (EFITA) Congress 2001, Montpellier, France
(http://www.effita.net)
Maywald, G.F., Sutherst, R.W., Zalucki, M.P. 1999. DYMEX Professional: Modelling Natural
Systems Version 1.0. CSIRO Publishing, Melbourne
Prasad, Y.G. 2005. Pest forecasting: Trends and development in India. In: Gleanings in
Entomology (V.V. Ramamurthy, V.S. Singh, G.P. Gupta & A.V.N. Paul eds). Division of
Entomology, IARI, New Delhi. pp. 145-168.
Samways, M. J., Osborn, R., Hastings, H. and Hattingh, V. 1999. Global climate change and
accuracy of prediction of species geographical ranges: establishment success of
introduced ladybirds (Coccinellidae, Chilocorus spp.) worldwide. Journal of
Biogeography 26: 795–812
Sutherst, R.W., Maywald, G.F. & Skarratt, D.B. 1995. Predicting insect distributions in a
changed climate. Insects in a changing environment (eds R. Harrington and N.E. Stork),
Academic Press, London pp. 59–91
Sutherst, R.W., Maywald, G.F., Yonow, T., Stevens, P.M. 1999. CLIMEX: Predicting the
effects of Climate on Plants and Animals. CD ROM and User Guide. CSIRO Publishing,
Melbourne pp. 90
Venkataraman, S. and Krishnan, A. 1992. Weather in the incidence and control of pests and
diseases. In: Crops and Weather (Ed. Venkataraman, S. and Krishnan, A.) Publications
and Information Division, Indian Council of Agricultural Research. pp: 259-302
Yonow, T., Zalucki, M.P., Sutherst, R.W., Dominiak, B.C., Maywald, G.F., Maelzer, D.A. and
Kriticos, D.J. 2004. Modelling the population dynamics of the Queensland fruit fly,
Bactrocera (Dacus) tryoni: a cohort-based approach incorporating the effects of weather.
Ecological Modelling 173: 9–30

206
207
208
209
210
211
212
213
214
215
216
 Concept of Genetic coefficients and their estimation
P. Vijaya Kumar, Senior Scientist (Ag. Met)
CRIDA, Hyderabad.
Introduction:
Genetic coefficients are integral components of all the crop growth simulation models.
Lack of knowledge on genetic coefficients and their non-availability in the crops of interest in
India is posing a problem for use of crop simulation models for evolving better agricultural
management practices.
Definition Genetic coefficients:
Genetic coefficients are mathematical constructs that are designed to mimic the
phenotypic outcome of genes under different environments. Each simulation model is driven by
cultivar, ecotype and species coefficients all known as Genetic coefficients.
Cultivar coefficients: They define traits that differ among cultivars
Ecotype coefficients: They define traits for groups of cultivars
Species coefficients: They define traits specific to any crop or crop species.
Genetic coefficients for each variety are affected by processes or factors viz., life cycle,
photosynthesis, sensitivity to day light (photo period), leaf area, partitioning, remobilization,
seed growth, seed composition, seed fill duration, Vernalization, growing degree days
accumulation etc.
Cultivar coefficients:
Cultivar coefficients govern life cycle and reproduction growth rate of the crop cultivars.
The definition of cultivar coefficients is located on cultivar files of all crop simulation models.

217
Genetic coefficients of some crop:

Genetic coefficients of Sorghum and Millet

Typical
Variety Definition Units
Range
Critical photoperiod or the longest day length
P2O (hours) at which development occurs at a Hr 12.5-16.0
maximum rate. At values higher than P2O, the
rate of development is reduced
Extent to which phasic development leads to
P2R panicle initiation (expressed in degree days) is GDD 1-300
delayed for each hour increase in photoperiod
above P2O

218
 G1 Scaler for relative leaf size (used in place of SLA) 0-20

G2 (Sorghum) Scaler for partitioning assimilates to 1-8

the panicle
G4 (Millet) Scaler for partitioning assimilates to the 0-1.0
panicle

Ecotype coefficients:
Most CERES models now have ecotype coefficients and library of ecotype coefficients
are also defined in the models. All the ecotype coefficient files will have extention of ‘ECO’.
They are traits that are common across many cultivars. Temperature responses of vegetative and
reproductive periods, radiation use efficiency and growing degree days to emergence are some
of the ecotype coefficients to name.
Species coefficients:
All CERES models now have species traits, although these traits are not consistent
among the crops. Species traits are characteristics that define difference between plant species.
These species traits include (i) temperature response (ii) CO2 response (iii) water stress response
and (iv) Tissue composition etc. Temperature response in case of development is defined in
terms of base temperature, Temperature optimum 1 and 2 and maximum temperature.
Temperature response on Crop Growth Rate (CGR) and photosynthesis are also some of the
species coefficients. Response of Radiation Use Efficiency (RUE) to CO2 in both C3 and C4 type
of crops also serve as species coefficients.
Genetic coefficients in cropping system model:
In cropping system model (CSM-CERES) maize, wheat/Barley and rice contain 6, 7 and
8 cultivar coefficients, respectively. In CSM-CROPGRO models, cotton, tomato, cabbage and
green bean crops have 18 cultivar coefficients and potato contain 6 cultivar coefficients.

219
Cultivar coefficients for CROPGRO:

The cultivar coefficients in CROPGRO models and their definition are illustrated in
following tables:
Cultivar coefficients for CROPGRO – Phase Modifiers
Name Value Definition
CSDL 13.40 Critical Short Day Length below which repr. Development
progresses with no day length effects (hours)
PPSEN 0.285 Slope of the relative response of development to
photoperiod with time (1/hour)

Cultivar coefficients for CROPGRO – Life Cycle Duration

Name Value Definition
EM-FL 19.0 Time between plant emergence and flower appearance
(R1) (PD)
FL-SH 6.0 Time between first flower and first pod (R3) (PD)
FL-SD 14.0 Time between first flower and first seed (R5) (PD)
SD-PM 33.2 Time between first seed (R5) and beginning maturity
(R7) (PD)

Cultivar coefficients for CROPGRO – Vegetative

Name Value Definition
FL-LF 26.0 Time between first flower (R1) and end of leaf
expansion (PD)
LFMAX 1.030 Maximum leaf photosynthesis rate at 30 C, 350 vpm
CO2, & high light (mg CO2/m2-s)
SLAVR 375 Specific leaf area of cultivar under standard
conditions (cm2/g)
SIZLF 180 Maximum size of full trifoliate (3 leaflets) (cm2)

Cultivar coefficients for CROPGRO – Reproductive

Name Value Definition
XFRT 1.00 Maximum fraction of daily growth that is
partitioned to seed + shell
WTPSD 0.19 Maximum potential weight per seed (g)
SFDUR 23.0 Seed filling duration for pod cohort at standard
growth conditions (PD)
SDPDV 2.20 Average seed per pod under standard growth
conditions (#/pod)

220
 PODUR 10.0 Time required to reach final pod load under
optimal conditions (PD)
How to obtain and determine genetic coefficients:
Genetic coefficients can be obtained by querying DSSAT or ICASA experts, calibrating
to the measured data and gene based estimation.
Measuring specific traits:
It can be achieved in environment control chamber experiments but they are expensive
and also the conditions inside the chambers will differ from those in experimental or farmers
fields.
The most commonly used method for estimating cultivar coefficients is through the use
of field data. It is also referred as ‘Inverse Modeling’ or ‘Fitting Coefficients’ or ‘Calibrating’ in
some cases.
If you have model and do not have cultivar coefficients you can estimate them with the
help of field data having carefully measured traits.
The objective criteria for determining the coefficients values is by fitting the observed
and simulated data in to statistical regression.
Fitting coefficients to field data:
Data needed (minimum data set):
· Frequent observations of timing of vegetative and reproductive events
· Growth analysis data
· Final yield, yield components
· Weather data (essential)

Multiple environments (locations, years, dates)

· Yield trials (Plant Breeders)
· Multiple planting dates are very useful
· Extreme (non-commercial) dates or locations
· Especially helpful to predict phenology

Estimating Coefficients: Trial and Error

Initial values from ‘best guesses’

Phenology coefficients
221
 · Data are observations of flowering, maturity and other stages
· Simulate available experiments, varying genetic coefficients (e.g., P1, P2 and P5 in
maize)
· Criteria for selection
§ Visual closeness of simulated to observed data
§ Statistical measures (OLS, RMSE, Likelihood function, correlation,
model efficiency, Wilmott (1981) D-index, …) Wallach et al., 2006

Growth, partitioning and yield coefficients

· Yield, yield component data (also growth analysis data?)
· Simulate available experiments many times, varying genetic coefficients (e.g., G2 and
G5 in maize)
· Criteria for selection
§ Visual closeness to simulated to observed data
§ Statistical measures (OLS, RMSE, Likelihood function, correlation,
model efficiency, Wilmott (1981) D-index, …) Wallach et al., 2006

‘Fitting’ Coefficients can be done through software (Gencalc2) in DSSAT v4.5 ‘Rules’
file with information on target traits, controlling coefficients, number of simulations and step
size for each coefficient. Other software such as gradient search or Monte Carlo methods.
GLUE: General Likihood Universal Estimation available in DSSAT v4.5.

222