See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/358604389

Chapter 4: Statistical procedure in forest inventory

Presentation · February 2022

CITATIONS READS

0 409

1 author:

Mahmood Hossain
Khulna University
134 PUBLICATIONS   1,046 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Allometric equations for the Sundarbans and Sal Zone of Bangladesh View project

Allometric Equations for the Sundarbans and Sal zone of Bangladesh View project

All content following this page was uploaded by Mahmood Hossain on 15 February 2022.

The user has requested enhancement of the downloaded file.

Chapter 4: Statistical
procedure in forest
inventory

Dr. Mahmood Hossain

Professor
Forestry and Wood Technology Discipline
Khulna University
Contents

Lecture 12, 13,

14, 15

o Sampling designs commonly applied in forest inventory

o Random, systematic, cluster, two phase, multiphase, stratified
sampling
o Principles and application of point sampling
o Point centered Quarter Method (PCQM)
o Variable Area Transect (VAT)
 Types of sampling

There are two main type of sampling used in

forest inventories

Random sampling

Unrestricted or simple random

sampling
Stratified random sampling
Multistage sampling
Multiphase sampling
Sampling with varying probability
List sampling
Non-random sampling
Selective sampling
Systematic sampling
Sequential sampling
1
2
Random sampling
 Random sampling:
Random sampling is that method of sampling in
which sampling units are selected in such a
manner that all possible units of the same size
have equal chance of being chosen.

 Unrestricted or simple random sampling: In this

type of sampling, sampling units are selected by
some strict random process for the whole
population or area.
 Example: Unrestricted or simple random
sampling:
o Sample unit is 0.1 ha, -plot size (50 m x 20 m) =1000 m2
o The population area is 1000 ha
o The population is divided into 1000/0.1 = 10,000
sampling units
o Each of the sampling unit is numbered from 0000 to
9999. The sampling intensity is 5%, which give 500
sample plots.
o These 500 plots are selected from the 4 digit random
number from random table
o Example of generating random number
o (Excel random number Generation using function
=RANDBETWEEN(1,9999)
 2 3 4

1 5

9999

Forest area divided into equal sized plots

Simple random sampling
 Stratified random sampling:
In this sampling method, population is first
divided into sub-populations of different strata.
The sampling units are selected from each of
the stratum in proportion to their size.

This modification of unrestricted random

sampling gives good results only if the forest
crop is uniform but generally forest populations
are characterized by considerable
heterogeneity.
 2 3 4

1 5
1 1

Site quality I Site quality Iv

Site quality Iii

321
500 124
Site quality iI
230

Forest area divided into equal sized plots

 Example: Stratified random
sampling:
o Forest area 1000 ha
o Sampling unite is 0.1 ha
o Sampling intensity is 5%
o This forest area is sub-divided into 4 strata on the basis of
site quality
o Site quality I = 300 ha, required 150 sampling unit
o Site quality II = 350 ha, required 175 sampling unit
o Site quality III = 200 ha, required 100 sampling unit
o Site quality IV = 150 ha, required 75 sampling unit
o The selection of sample units from each stratum (site
quality) will be done by following the randomization
methods
Stratified Sampling
 Multistage sampling:

In random sampling, sampling units are

selected from a population at one time or
stage. ----------- It is therefore called single-
stage sampling. In the multistage sampling,
sampling units are selected in two or three
stages.

In this sampling, the sampling units get smaller

at each successive stage of sampling and
selection of sample units from each stage will
be done by following the randomization
methods
 Example 1: Multistage sampling

o Forest area 100 ha, divided into 100 sampling

units of 1 ha (100 m x100 m).
o With 5% sampling intensity, 5 sampling units will
be selected in the primary stage (100 m x 100
m plot)
o Then each of the selected primary sampling
unit further be broken up into (10 m x 10 m)
sampling units-------- Sampling intensity with 5%,
sampling units will be selected at randomized
methods.
1st stage 100 ha = divided into 100 no of plots (100 m x 100 m)

5 Plots (100 m x 100 m ) with red colour well be selected @ 5%sampling intensity
 2nd stage 1 ha = divided into 100 no of plots (10 m x 10 m)

5 Plots (10 m x 10 m ) with red colour well be selected @ 5%sampling intensity

 Multiphase sampling:

In this method, sampling units are used at the different

phases of sampling to collect different information or
same information by different methods.

 Example 1:
Number of bamboo clumps in a forest area will be
estimated
In the 1st phase, the number of clumps/ha will be
determined by large survey in randomized
procedure.
In the 2nd phase, number of stem/clump will be
determined from much smaller area than taken in
the 1st phase in randomized procedure.
 Example 2: Multiphase sampling:

o Volume estimate from the aerial photograph

and field plots

o In the 1st phase, random sampling consists of

estimating the volume on aerial photograph of a
large number relatively inexpensive sampling
units employing photo-interpretation and
measurement technique.
o In the 2nd phase, a sub-sample of these plots
is selected and visited in the field for direct
measurement of their volume. ……… This
sub-sample is much smaller since field-plots
are more expensive to lay than photo-plots.

o ------------------------ A regression is then

obtained between field-plot volume and
photo-plot volumes permitting a correct
volume estimate to be made for the large,
less expensive sample of the 1st phase.
Subset of P2, with additional
Variables: soil, ozone, etc. Phase 3

Field data collection:

tree species, height, diameter, Phase 2
land use, etc

Forest area estimates ~ based

on photo interpretation Phase 1
*Highest sampling intensity
 Cluster sampling:

In cluster sampling, total population is

divided into groups. Groups or clusters are
randomly selected and all elements in
each selected group will be enumerated.
Cluster Sampling
 Cluster Sampling is a type of multistage sampling
that consists of selecting primary sample points
and establishing a number of closely located
secondary sampling units (often 3 or 4) as a group
or cluster at each selected primary point. The
cluster then forms the primary sampling unit.
 Cluster sampling has been used with success in
tropical forests in remote areas involving difficult
access. Under these circumstances, it may take
several days to reach the location of a sample
plot, and with little additional cost or effort, several
plots in a group or cluster can be taken. The
location of the clusters can be determined either
systematically or randomly.
 Area sampling:
 Area sampling is quite close to
cluster sampling

 This sampling is used in

connection with selection of
samples area with help of maps

 The procedure is to divided the

larger areas into several smaller
areas
 At first, the total area is
divided into a number
of smaller non-
overlapping areas,
generally called
geographical cluster.

 Secondly, numbers of
smaller areas are
randomly selected
and all the units in
these smaller areas are
included in the sample.
Sampling with varying probability:

 When the sampling units are not equal in their

size and variables under study is correlated
with size then the simple random sampling may
not be effective (Sampling units should be
equal in size for the simple random sampling).

 When sampling units vary in their size and the

variable under study is directly related with the
size of the unit, the probabilities may be
assigned proportional to the size of the units.
 Point sampling: (sampling with
varying probability)

The theory and technique of

point sampling is known as
angle count sampling, plotless
sampling, Bitterlich’s method
and relascope sampling, and is
widely recognized as a
breakthrough in forest
mensuration, was developed by
the Australian forester Bitterlich
in 1948.
This sampling method is used to
estimate the basal area per
hectare.

Point sampling estimates stand

parameters from sample plots
with imaginary plot boundary.

A critical angle (45o) is

introduced, which defines the
ratio between the diameter of a
tree and its distance from the
sampling point
 For each of the trees surrounding the sample point,
a decision is made whether it is “in” or “out” or
“borderline tree”
 A tree is “in” --------------- It exceeds the critical angle
being used.
 A device with a fixed angle of view is used to sight
the surrounding trees at breast height in a 360o
sweep
 The basal area per unit area is obtained by
multiplying the number of trees being tallied by a
basal area factor
 List sampling:

This is another form of sampling with varying

probability. This method consists of making
a list of sampling units along with their
measure of size in any order.

Example:
15 compartment of sampling units. They
are listed as shown followings
Compartment Area (ha) Commutative Total of area Associated number
(ha)
1 20 20 1-20
2 15 35 21-35
3 30 65 36-65
4 18 83 66-83
5 22 105 84-105
6 30 135 106-135
7 15 150 136-150
8 17 167 151-167
9 22 189 168-189
10 19 208 190-208
11 21 229 209-229
12 25 254 230-254
13 23 277 255-277
14 24 301 278-301
15 11 312 302-312
  If it is decided to select a
sample size of 6 items, then 6
random numbers are drawn
from the range of integers 1 to
312.

Random Compartment  A compartment is chosen as a

sample if the random number
number chosen chosen
falls within the interval indicated
299 14
in the column of associated
123 6 numbers. The following may be
chosen in this case.
247 12
405 15
269 13
064 4
Non-random sampling
 Non-random sampling:
In non-random sampling, samples are selected
according to the subjective judgment of the
observer on the basis of certain rules or
guidelines indicating what sample should be
chosen.

Example:
In case of volume table construction,
The sample tree should be straight, uniform
canopy, unaffected from disease or top
broken etc.

These are the selection criteria of trees for

volume table construction, here selection is
based on the subjective judgment of the
observer
 Selective sampling:

Selection sampling consists of choosing

sample according to the subjective judgment
of the observer.
Selective sampling may give good
approximation of population parameters.
 Systematic sampling:
In systematic sampling, sampling units are selected
according to a predetermined pattern. Most commonly,
the pattern consists of regular spacing of units.

 Example:
Sampling units may be located at certain fixed
distance (e.g. 80 m) in the raw and at certain fixed
distance (e.g. 200 m) in the column.
The selection of first unit may be either random or
according to a fixed rule
When 1st unit is selected at random is referred as
systematic sampling with a random start.
Systematic Sampling
  Sequential sampling:
Sequential sampling is a method
of sampling whose characteristic
feature is that the number of
observations in the sample is not
determined in advance but
sampling units are taken
successively from a population.

• Each sample includes all the sampling units of

the former sample.

• Sampling is stopped when the desired precision

is reached
 Quota Sampling:

This is one type of non-random sample. When

strata are present but stratified sampling is
not possible due to the nature of the
investigated parameter of the population
and the objectives. Here required quota
sample are drawn from each strata.

Example:
If it is important to include a significant
number of respondents (certain number)
from any particular group.
Sampling techniques can be used both for
random and non-random sampling

 Strip or belt transect:

The width and interval of strip or belt depends
on sampling intensity.

Strip or belt transect can be taken

systematically or randomly
 Line transect:

Line transect can be taken randomly or

systematically

The width of line is consider to be one

centimeter
 The loop method:

 This type of sampling technique is suitable for

grassland and herbaceous community
 Loops are made with steel wire with varying
size upto 1 m2
 This loops can be used with the line transect
or can be an individual sampling unit.
 If loops used with the line transect, the size of
loops usually smaller about 1 cm2
 Individual useable loop can be 1 m2 in size.
 Loops used together the line transect:

The loop is placed beside the tape to

the ground at every designated point
along the transect

All vegetation inside the loops are

observed or measured

Usually 100 loops are used in one

transect , the number can also vary
 Loops used alone:

The loops are kept at designated

point (if systematic)

The loops also thrown at any point on

the study area (randomly)

All vegetation inside the loop is

observed
Estimating abundance using
distance methods
 Estimating abundance using
distance methods
 Plant ecologists have developed a variety of
plotless sampling methods that work well on
trees and shrubs. These are all called distance
methods because they utilize distances
measured by means of two general approaches

 Select random trees and measure the distance to

its nearest neighbors

 Select random points and measure the distance

from the point to the nearest organisms
Applicability

 Plotless sampling is usually applied to a single

species and it is popular in ecology

 We can use plotless sampling to estimate the

density of the population.

 On the other hand, we can determine the

spatial pattern of the distribution like random,
aggregated, or uniform.
T-Square sampling procedure
 T-Square sampling procedure

 A distance sampling scheme

called T-square sampling was
first described by Besag and
Gleaves (1973).

 It is simple to implement in the

field.

 Random points are located in

the study area and at each
random point two distances
are measured
The distance (xi) from the random point (O) to the Layout and
nearest tree (P) measurement

The distance (zi) from the tree (P) to its nearest

neighbor tree (Q) with the restriction that the angle

o
Ordered Distance Method
Ordered Distance Method

 The ordered distance method was

first suggested by Morisita (1957) and
further developed by Pollard (1971).

 The method involves measuring the

distance from a random sampling
point to the nth closest individual.
Plot layout

Try to get 30 to 50 random points

 Measurement
 Locate a random point in the study area.
 Determine the third nearest individual to the
random point.
 Measure the distance from the random point to
the third individual (Ri).
 Repeat the entire procedure for the next random
point. Try to obtain n = 30 to 50.
  The general formula for estimating
population density

Ď = population density estimated by the ordered-

distance method

n = number of random points sampled

π = 3.14159

Ri = distance from random point to third nearest tree

Variable Area Transect (VAT)
 Variable Area Transect (VAT)
 This method is a combination of distance and
quadrat methods and was first suggested by Parker
(1979).

 A fixed width strip is established from a random

point until the nth individual is located in the strip.
 A field worker needs to search in only one direction
from the random point, and once the nth individuals
(desired number) found, the length of transect is
measured from the random point to point at which
the nth individual occurred.
 This method can be used for any number of
individuals, and we shall use n = 3 as a convenient
number for field workers.
Layout
 Layout and measurement
 Locate a random point on a transect line in your
study area.
 Move along the transect from the random point until
you have found three individuals of the species
being studied. Project a perpendicular to the line of
travel (marked by a tape measure usually).
 Measure the distance l along the tape from the
random point to the perpendicular line that projects
to the third individual.
  The formula for population density estimated

Ďv= Estimate of population density for the variable area

transect method

n = Number of random points

w = Width of transect (fixed)

€ li is the sum of lengths searched to the selected

individual
Point-Centered Quarter Method
 Point-Centered Quarter Method

 The point-centered quarter method is used to

gather information on the density, frequency, and
coverage of tree species found in a forest.

 This provide information on the followings

 Number of individual trees

 How often a certain tree occurs?
 How common a tree species compared to other
trees?
 Size of the trees
 Layout

Measure distance from the center point to the nearest

tree ≥ 40 cm, and nearest tree > 2.5 cm to < 40 cm

Identify the species

Principles:

 In the point-centered quarter method, a point in the

forest is identified and the area around it is separated
into four quarters.
 In each quarter, the nearest trees were selected based
on DBH value
 DBH ≥ 40 cm is identified – considered as “large tree”
sample.
 DBH > 2.5 cm to < 40 cm – considered as “small tree”
sample.
 Identifying a large tree and small tree in each quadrant
provides the ability to compare the over story to the
understory.
 Layout and measurement

Measurements:
 Distance measured from the center point to the
nearest tree ≥ 40 cm, and nearest tree > 2.5 cm to <
40 cm
 Identify the species
 DBH of the selected trees
Outputs:

 Density
 Relative density
 Dominance
 Relative dominance
 Frequency
 Relative frequency
 Basal Area
 Importance Value of each tree species
 Calculation

 It is necessary to sum the point-to-tree/organism

distances for all species and calculate a mean.

 Calculate the mean point-to-tree distance for the

entire sample of large trees of all species.
 Calculate the mean point-to-tree distance for the
entire sample of small trees of all species.
Calculate the average density (the number of trees/hectare)
for both the large trees and the small trees.
 Never regret a day in your life
Good days give you happiness
&
Bad days give you experience

Thank you
View publication stats