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

net/publication/283733118

Estimation of soil erosion in northern Kirkuk

Governorate, Iraq using RUSLE, remote
sensing, and GIS

Article · March 2016

CITATION READS

1 219

3 authors, including:

Hussein B Ghalib Wasan AL-Qurnawi

University of Basrah University of Basrah
9 PUBLICATIONS 4 CITATIONS 1 PUBLICATION 1 CITATION

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Hussein B Ghalib on 23 January 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Carpathian Journal of Earth and Environmental Sciences, February 2016, Vol. 11, No. 1, p. 153 - 166

ESTIMATION OF SOIL EROSION IN NORTHERN KIRKUK

GOVERNORATE, IRAQ USING RUSLE, REMOTE SENSING AND GIS

Alaa M. Atiaa AL-ABADI, Hussein B. GHALIB & Wasan S. AL-QURNAWI

Department of Geology, College of Sciences, University of Basra, Basra, Iraq.Corresponding author: Hussein B.
Ghalib <[email protected]>

Abstract: A quantitative assessment of annual soil erosion by water in the northern part of Kirkuk
Governorate, north of Iraq was investigated through integration of remote sensing, GIS and empirical
RULSE soil erosion model. The five factors of RULSE model (rainfall erosivity R, soil erodibility K,
slope length and steepness LS, crop management C, and practice factor P) were derived from different
resources such as field survey, archival data, digital elevation model, and LANDSAT 8 multi-bands
imagery. The annual soil erosion loss was estimated by multiplying the five factors in raster format using
raster calculator of ArcGIS 10.2 software. The estimated annual soil losses rate for the study area ranges
from 0 to 245 (t ha-1 yr-1) with an average of 2 (t ha-1 yr-1). The value ranges were classified into four
categories: minimal, low, moderate, high soil erosion hazard zones using four classification schema:
quantile, natural breaks, geometric, and standard deviation. Due to the similarity of results, the
comparison was carried out between two schemas: natural breaks and geometric. The area covered by
minimal-low soil hazard zones extends over an area of about 88% and 99% based on geometric and
natural breaks schema, respectively. In turn, the moderate-high soil hazard zones cover only very small
area (0.3%) based on natural breaks and relatively small area (12%) depending on geometric scheme. In
general, both method results indicate that hazard of soil erosion is low in the study area. The spatial
pattern of classified soil erosion rate indicates that the areas at moderate to high risk is located in the
northeast and very small area in the east, while the minimal to low zones cover the other parts. The
obtained results of could be useful to implement soil conservation practices in the study area.

Key words: RUSLE, GIS, NDVI, Kirkuk, Iraq, soil erosion.

1. INTRODUCTION Erosion Model (Morgan et al., 1998) and many

others (Pervoić et al., 2013). Among these known
Soil erosion by water is a world major problem models, the USLE and revised RUSLE version are
that results in decrease of soil fertility, land the most widely used models for assessing soil
degradation, and also affecting the sustainability and erosion hazard around the world, especially in the
productivity of agricultural areas. In fact, soil developing countries due to simplified structures and
erosion is a geomorphological process driven by data demand. In recent years, the combined use of
natural forces, and it could be accelerated by human satellite imagery, geographic information system
activities such as deforestation, construction, with RUSLE model makes soil erosion estimation
agriculture, and mining. To quantify soil rate, on regional scale feasible with reasonable costs and
techniques such as erosion modeling can be helpful. better accuracy (Millward & Mersey, 1999; Wang et
Soil erosion modeling consists in mathematically al., 2003; Lu et al., 2004; Jasrotia & Singh, 2006;
describing of the soil particle detachment, sediment Krishna Bahadur, 2009 in Prasannakumar et al.,
transport and deposition on land surface (Lal, 1994). 2011). Many researchers applied USLE/RUSLE
For assessing the soil erosion, many empirical linked with remote sensing and GIS techniques for
models were developed in the past such as USLE, assessing soil erosion and sediments yields with
Universal Soil Loss Equation (Wischmeier & Smith, successful results (Angima et al., 2003; Lee, 2004;
1978); WEPP, Water Erosion Prediction Project Lu et al., 2004; Kim et al., 2005; Onyando et al.,
(Nearing et al., 1989), EUROSEM, European Soil 2005; Onori et al., 2006; Pandey et al., 2007;

153
Erdogan et al., 2007; Ismail & Ravichandran, 2008; concerning the spatial rate of soil erosion. This study
Yue-Qing et al., 2008; Ugur Ozcan et al., 2008; integrates the empirical RUSLE model with remote
Krishna Bahadur, 2009; Terranova et al., 2009; sensing and GIS techniques to investigate the spatial
Pandey et al., 2009; Arekhi et al., 2010; Jain & Das, distribution of annual soil loss potential in the
2010; Wang et al., 2010; Park et al., 2011; Chen et northern Kirkuk Governorate, Iraq. Delineation of
al., 2011; Kefi et al., 2011; Prasannakumar et al., soil erosion zones is very important to protect soil
2012; Ozosy et al., 2012; Perović et al., 2012; and manage fertile land in a sustainable manner.
Mhangara et al., 2012; Demirci & Karaburun 2012;
Ashiagbor et al., 2013). 2. MATERIALS AND METHODS
The quantity and quality of water and land
available for agricultural production are two 2.1 Study area
fundamental factors on which the economy of the
country is mainly based. The selected area for this The study area located in the northeast Kirkuk
study, the northern part of Kirkuk Governorate, Iraq, Governorate, north of Iraq, along the left side of
is characterized by an abundance of fresh Alton Kopri Basin between latitudes (35°30' –
groundwater and arable land. Despite of the 35°51') and longitude (44°04' – 44°37') and occupies
agricultural importance of the study area, and to the an area of 1,940 km2 (Fig. 1).
best knowledge of the authors, there are no studies

Study area

Figure 1. The northeast Kirkuk Governorate, along the left side of Alton Kopri Basin location in north of Iraq

154
 The Lesser Zab River divides the whole area northeast of the study area. A small valley covered
into two parts, right bank (northern part) belongs to with silt and clay is permeating between these two
Erbil Governorate, and left bank (southern part) lies chains (Al-Sayab et al., 1983). The southwest side of
in Kirkuk Governorate. The first series is the area is particularly a very intense decline
represented by Kanydomlan which is considered a compared to the northeast side of the basin. The
part of baba dome with a height of 450 m (msl) to central area is almost a flat land with the interference
the southwest of the area and the second is a series of some bends and torsions due to the present of
mountain (Kalkalan Dagh) with a height of about several simple wadis, which are filled with water
800 m (msl) to the northeast of the study area. through the seasonal rainfall and discharged toward
Generally, the region elevation ranges between 227 - the Julak valley. This is the main portion of the
852 m above mean sea level (msl), (Fig. 2). The area natural discharge to the Little Zab River (Al-Sayab
shows symmetrical shape surrounded by two parallel et al., 1983).
chains of mountains. The first series is represented The climate is hot and dry in summer, and cold,
by Kanydomlan which is considered a part of Baba humid in winter. The rainy season is from October to
dome with height of 450 m (msl) to the southwest of May. The average annual precipitation is 342.7mmbut
the area. The second is a series mountain (Kalkalan the evapotranspiration amounts 1662.9 mm. The
Dagh) with height of about 800 m (msl) to the dominant climate is semiarid.

Figure 2. Spatial representation of elevation (m) in the study area

155
 Four land use/land cover classes are prevailing folded zone within the hills where the deep valleys
in the study area, namely: herbaceous rangeland descend spreading from the hills. These are shallow to
(1.5%), agricultural farms (10.2%), crop land (44.8%), moderate soils. They are environmentally classified as
barren land (32%), and urban (11.2%). The grasslands being suitable for grazing (Buringh, 1960).
distinguished physiographic units in the area are high The fourth type is represented by brown lithosol soils
folded, foothill zone, flat, and river terraces (Haddad et with moderate to deep thickness. They occur both in
al., 1971). the eastern and western side of Kanydomlan Mountain
From a geological point of view, the study area series an in the centre of the basin. These soils are
is part of the foothill zone in the folded area of unstable likely to be formed from wind erosion, river sediment
shelf of Iraq. The unstable shelf had been most strong transport and deposition. They consist of alluvial clay,
subsiding part of the Arabian plate since the opening of silt and sand, and also contain a small amount of
the southern neo- Tethys in the late Jurassic (Buday & gypsum especially near the Kanydomlain Mountain.
Jassem, 1987). Maximum subsidence occurred during They are considered to be a good area for the
the late Cretaceous ophiolite obduction onto the NE cultivation of winter crops depending on rainfall
margin of the Arabian Plate and during Mio-Pliocene availability (Buringh, 1960).
continental collision. The unstable shelf is thus
characterized by structural trend and facies changes 2.2. Methods
that are parallel to the Zagros-Taurus suture belts.
Surface folds are a characteristic feature of the unit. In The RULSE has been widely used to predict
the study area, outcrops formations are ranging in age the average annual soil loss by introducing improved
between Miocene to Holocene. The Miocene means of computing the soil erosion factors.
formations are represented by Injana and Fatha Basically, the magnitude of soil erosion depends on
Formations, consisting in clay layer. Pliocene layers two factors, namely: the detachment of soil particles
occur in the Northeast of the basin and represented by by the impact of rainfall energy, called the erosivity
Muqdadyi and Bi-Hassan Formations along Kalkalan of rain, and the ability of the soil to resist the
Dagh series consisting of successive layers of gravel, detachment of its particles by this force, called the
conglomerate and thin red silty clayey seam. The erodibility of soil (Wischmeier & Smith, 1960 and
Quaternary layers (loess, loess like deposits, alluvial) 1978). The RULSE equation states that:
cover both the centre of the basin; called Julak (Wischmeier & Smith, 1978; Renard et al., 1997)
Basinand the sides of the Little Zab Valley represented A = R × K × LS × C × L (1)
by fluvial terraces (Parson, 1955; Haddad et al., 1971).
The soil of the study area formed during the where A is the annual average of soil erosion rate
Quaternary period. Most of soils show brownish to factor [t ha-1 yr-1]; R is the rainfall erosivity factor
yellow colour on top. According to Buringh (1960) [MJ mm ha-1 h-1 yr-1]; K is the soil erodibility factor
there are four types of soil in the study area: lithosol [t ha-1 h-1 MJ mm-1); LS is the slope – steepness
soil with sandstone and gypsum, lithosol soil with factor (dimensionless); C is the crop management
limestone, moderately brown to shallow soils, and factor (dimensionless, ranging between 0 and 1); and
thick brown soils. The first type of soil was found in P is the conservation support practice
highland of Kirkuk and hills in northwest and west of (dimensionless, ranging between 0 and 1). The
Kirkuk city. These are often shallow and have been erosivity of rain is represented by the factor R, and
exposed, in places, to severe erosion to form water the erodibility of the soil surface system by the
composed grooves as a result of rill erosion along the multiples of factors KLSCP.
slopes. Buringh (1960) described them as poorly For this study, the RULSE five factors were
distributed, shallow in thickness at the highland slopes prepared by means of ArcGIS 10.2 commercial
while the thickness increases at the foothills. The software. Input five factors belongs to monthly and
second type is brown in colour and is located within annual rainfall data, soil data, digital elevation
folded zones in areas of Bi- Hassan and Muqdadyi model, and satellite imagery were acquired from
Formations. They are deposited at the foothills of different resources such as Iraqi Meteorological
Kalkalan Dagh mountain series in the north east of the Commission archives, field surveys, and web
area). Generally, the soils of this area are shallow and resources for DEM and LANDSAT 8 images. All
do not exceed 30 cm in thickness. They are five factors were prepared as raster from with 30m ×
Aluviosoils, mixed with gravels and accumulated 30m cell size in spatial analysis extension of ArcGIS
limestone separate between them. They are proper for software. The projected coordinate system for all
grazing, when small little grass spreads in the area thematic layers was (UTM WGS 1984- 38N).
(Buringh, 1960). The third type of soils extends in the

156
 3. GENERATING OF RULSE THEMATIC rainfall total (mm). Since there is no any
LAYER FACTORS meteorological station in the study area, data from the
following nearby stations were used for estimating R
3.1 Rainfall erositivity (R factor) values: Erbil, Sulaymaniyah in Iraqi Kurdistan region,
Kirkuk and Baiji. Table 1 shows the estimated R
The rainfall erosivity (R) factor is defined as the values along with other relevant information. The R
potential ability of rain to cause erosion and given as values ranges between 108.26 - 716.28 with average
the product of a given rainfall storms’ maximum 30- value of 368.40 (MJ mm ha-1 h-1 yr-1). It is obvious
min intensities (I30) and the kinetic energy of from table 1 that the calculated R is directly related
rainstorm (E) (Wischmeier & Smith, 1958, 1960 and with annual rainfall; high rainfall leads to high R
1978). The calculated erosion potential of rainfall values, and vice versa. The spatial interpolation of R is
storm is usually written as EI30. The total R is impossible with only four values or could be
therefore the sum of individual EI30 values for each interpolated but with large errors. Therefore, the
rainfall storm event. Unfortunately, these average value of R was assumed to be uniform over the
measurements are rarely available at standard study area and was used for the subsequent analysis.
meteorological stations in Iraq. The alternative
practical option is by using monthly and annual rainfall 3.2 Soil erodibility (K factor)
data collected from a single weather station
(Wischmeier & Smith, 1958, 1978; Arnoldus, 1980): The K factor is the product of susceptibility of soil
  2  
particles to erosion per unit of rain erosivity factor (R),
 1.5× log  Pi  − 0.08188 
12  10    for a specified soil on a unit plot having a 90% uniform
R = ∑1.735 × 10   P  
(2) slope and a slope length of 22.13 m over a
i =1 continuously clean shallow land with up and down
where R is the rainfall erosivity (MJ mm ha-1 h-1 yr-1), slope farming (Wischmeier & Smith, 1960 and 1978).
Pi is the monthly rainfall (mm), and P is the annual
Table 1. Calculated rainfall erosivity factor R from monthly and annual rainfall for considered stations
4

rainfall
annual
Meteorological erosivity
Location (UTM) average of monthly rainfall (mm) rainfall
station (MJ mm ha-1
(mm)
h-1 year-1)
Jan Feb Mar Apr May Jun
74.5 74.4 75.2 54.7 11.9 1.3
Erbil 410274.86 4005904.21 241.22 387.83
Jul Aug Sep Oct Nov Dec
0.5 0.1 0.7 30.8 52.1 80.3
Jan Feb Mar Apr May Jun
122 110 107 87.2 39.7 1.9
Sulaymaniyah 539259.49 3934466.56 449.78 716.28
Jul Aug Sep Oct Nov Dec
0.0 0 1.4 33.2 96.5 118
Jan Feb Mar Apr May Jun
68.7 66.8 50 43.5 13.8 0.2
Kirkuk 444670.76 3925206.21 181.21 261.23
Jul Aug Sep Oct Nov Dec
0.3 0.1 1 14.7 45.6 56.9
Jan Feb Mar Apr May Jun
36.2 35.8 30.5 20 11.5 0.5
Baiji 361326.84 3865674.76 72.42 108.26
Jul Aug Sep Oct Nov Dec
0.0 0 1 8.9 26.9 29.8

Table 2 Soil structure degree criteria (Chen et al., 2011)

a (%) ≤ 0.5 0.5- 1.5 1.51-4.0 ≥ 4.0
Soil structure code b 1 2 3 4

Table 3 Soil permeability degree criteria (Chen et al., 2011)

Clay % ≤ 10 10- 15.9 16-21.6 21.7-27.4 27.5-39 ≥ 39.1
Soil permeability code b 1 2 3 4 5 6

157
 It reflects the ease with which the soil is (Table 4). The most dominant soil texture was sandy
detached by splash during rainfall and/ or surface loam. The K values were then calculated using eq. 3
flow (Dumas et al., 2010). The K factor is closely and interpolated using weighted inverse distance
related to soil texture (percent of clay, silt, and sand), deterministic interpolation method to create thematic
organic matter content (%), soil texture and soil layer raster (Fig. 3). The calculated K values were in
permeability. The K factor for the study area was range 0.04 to 0.38 with K average equals to 0.14.
estimated by using the following equation
(Wishmeier & Smith, 1978; Foster et al., 1980) 3.3 Slope-Steepness (LS) factor
K = 2.8 × 10 −7
M 1.14
(12 − a ) + 4.3 × 10 (b − 2) + 3.3 × 10 (c − 3) (3)
−3 −3

The LS factor represents erodibility due to

where M is the particle size parameter (% silt +
combinations of slope length and steepness relative to
% very fine sand) (100 - % clay); a is organic content
a standard unit plot (Wischmeier & Smith, 1960 and
of soil %; b is the soil structure code (Table 2); c is
1978; Ashiagbor et al., 2013). Slope length (L) is
the profile permeability class (Table 3).
defined as the horizontal distance from the point of
For calculating K (t ha-1 h-1 MJ mm-1), a total of
origin of the overland flow to the point where either
22 samples of soil were collected at a depth of about
the slope gradient decreases enough that deposition
50 cm below the surface after removing the top soil
begins or runoff is concentrated in a defined channel
cover. The soil samples were collected in clean
(Wischmeier & Smith, 1978 in Ozosy et al., 2012).
polyethylene containers and transported to soil
On the other hand, the influence of slope gradient on
laboratory of Civil Engineering/ Engineering
soil erosion could be described by slope steepness (S)
College/University of Basra to carry out grain size
factor. An increase in slope length and steepness
analysis. Locations of these samples were selected
causes an increase in the LS factor. For estimation of
after many criteria, such as easiness to access, even
the LS factor, the following relationship was adapted:
distribution over the study area, and ease to dig
(Jain & Das, 2010).
through the soil surface. The collected soil samples n m
were assigned texture names based on the web-based  A   sin β 
LS =  s    (4)
USDA soil texture calculator. Organic matter content  22.12   0.0896 
was measured by using analytical technique. Four soil where As is the specific area (A/b), defined as
textures were found in the study area, namely: sandy the upslope contributing area for overland grid (A)
loam, loam, sandy clayey loam, and loamy sand
Table 4. The calculated soil erodibility (K) values for the soil samples
Sample K
Easting Northing Soil texture
No. (t ha h ha-1 MJ mm-1)
S1 442815 3936960 Sandy loam 0.13
S2 436742 3943197 Sandy loam 0.13
S3 424462 3951489 Loam 0.38
S4 421648 3957213 Sandy loam 0.13
S5 425314 3957151 Sandy clay loam 0.2
S6 443813 3943612 Sandy loam 0.13
S7 446972 3937708 Sandy loam 0.13
S8 453022 3934686 Loamy sand 0.04
S9 453470 3943402 Loamy sand 0.04
S10 445277 3948471 Loam 0.38
S11 442713 3952430 Sandy loam 0.13
S12 440817 3950748 Sandy loam 0.13
S13 434243 3952089 Sandy loam 0.13
S14 431197 3957904 Sandy loam 0.13
S15 440357 3964462 Sandy loam 0.13
S16 444025 3965270 Loam 0.38
S17 427326 3957442 Loamy sand 0.04
S18 428102 3963382 Sandy loam 0.13
S19 447587 3956621 Sandy loam 0.13
S20 445588 3958297 Sandy loam 0.13
S21 428027 3957251 Sandy loam 0.13
S22 431236 3946750 Loamy sand 0.04

158
 Figure 3. Spatial distribution of soil erodibility factor (K), for the studied area

Legend

Figure 4. Spatial representation of Slope (degree) categories.

159
per unit width normal to flow direction (b); β is the  NDVI 
C = exp − α
− NDVI ) 
(6)
slope gradient in degree; n and m are constants,  (β
usually taken as 0.4 and 1.3 for n and m, where α and β are unitless parameters that
respectively (Lee, 2004).
determine the shape of the curve relating to NDVI
For this study, the LS factor was estimated
from the Advanced Spaceborne Thermal Emission and the C-factor and usually taken as 2 and 1 for α
and Reflection Radiometer (ASTER) Global Digital and β , respectively (Van der Knijff et al., 2000;
Elevation Model(GDEM) The ASTER-GDEM was Perović et al., 2012). NDVI is one of the different
developed by the Ministry of Economy of Japan and indices developed to identify vegetated areas and
the United States National Aeronautics and Space their "condition" and it remains the most well-
Administration (NASA). The spatial resolution of known and used index to detect live green plant
the ASTER-GDEM is approximately 30 m. The eq. canopies in multispectral remote sensing data. The
2 can be written from hydrological parameters NDVI is calculated as follows:
derives from DEM as: NDVI =
(NIR − VIS ) (7)
LS = Power (Flow accumulation × cell size / 22.13, 0.4 )×
(NIR + VIS )
× Power (sin (slope ) / 0.0896,1.4 )
(5) where NIR and VIS are the spectral reflectance
measurements acquired in the visible (red) and near-
The original ASTER-GDEM was firstly re- infrared regions, respectively.
projected to (UTM WGS 1984 38N) projected For this study, LANDSAT 8 multiband image
coordinate system, clipped for the study area, acquired in 26/5/2014 was used for calculating C
reconditioning, and fill sinks. The flow direction and factor. The image was downloaded from USGS web
flow accumulation layers were derived from filled location (http://earthexplorer.usgs.gov/). The bands
ASTER-GDEM by using the Arc Hydro extension in were composed together to create composed image,
ArcGIS 10.2. The slope layer was derived from clipping for the study area, radiometric
filled DEM using spatial analyst extension (Fig. 4). enhancement, and then the NDVI was automatically
The minimum, maximum, and mean of slope calculated using Image analysis extension in
(degree) were 0, 44.3, and 4.6, respectively. The ArcMap. Eq. 7 then used for estimating C factor and
slope raster layer was classified into four categories: is presented graphically as a raster thematic layer in
gentle flat, flat, moderately steep and steep (Fig. 6). The minimum, maximum, and mean values
(Pourghasemi et al., 2013). The relative area (%) of C were 0.38, 0.76, and 0.55, respectively.
covered by each of these categories were 65.8%,
31.2%, 3%, and 0.1%, respectively. 3.5 Practice Management (P) factor
The extent of gentle-flat and flat categories ≈
79% mean that affecting of this factor on soil loss is The P factor is the ratio of soil loss with a specific
fairly small. The raster calculator of ArcGIS 10.2 support practice to the corresponding soil loss with up
was used to calculate LS using eq. 4 to derive the and down cultivation (Wishmeier & Smith, 1960 and
thematic layer of LS of the study area (Fig. 5). The 1978). The lower P value, the more effective the
range of LS factor is from 0 to 10.76 with an conservation practice, is deemed to be at reducing soil
average value of 0.02. erosion (Lee, 2004). The numerical value of P is
always less than 1. The P value is 1 if there is no
3.4 Crop management (C) factor management practice. The P factor value is taken as 1
for the whole study area because no data available
The C factor is defined as the ratio of soil loss concerning the management practice for the
from land cropped under specified conditions to the concerning area.
corresponding loss from clean-tilled, continuous,
fallow land (Wischmeier & Smith, 1960 and 1978; 4. RESULTS AND DISCUSSION
Das, 2002). It reflects the effect of cropping and
management practices on soil erosion rate (Renard et The annual soil loss was estimated by equation
al., 1997). The easiest way for calculating C factor is (1) implemented in raster calculator of spatial analyst
through using remote sensing data. The Normalized extension and presents graphically in (Fig. 7). The
Difference Vegetation Index (NDVI) is normally annual soil loss rate estimated for the study area ranges
used for estimating C factor through the following from 0 to 245 (t ha-1 yr-1). There is no Iraqi standard for
formula: (Zhou et al., 2008; Kouli et al., 2009; the classification of this range. Therefore, four
Prasannakumar et al., 2012) classification schema were used for comparing results,
namely: natural breaks, quantile, geometric, and

160
standard deviation. The detailed description of these categories: minimal, low, moderate, and high using
schema was found with ArcGIS help (ESRI, 1999). those different classification schemas.
The soil potential values were grouped into four

Figure 5. Spatial distribution of LS factor (dimensionless)

Figure 6. Spatial distribution of C factor as a raster thematic layer

161
 Table 5 shows that it is difficult to choose about 88% and 99% based on geometric and natural
which classification scheme fits best for the study breaks schema, respectively.
area, but in general, the geometric and quantile The moderate-high soil hazard zones cover only
approximately have the same results if compared very small area (0.3%) based on natural breaks and
with natural breaks and standard deviation schema relatively small area (12%) depending on geometric
which also have the same figures. Therefore, the scheme. In general, the results obtained through both
comparison was done only by using two schemas: methods indicate that hazard of soil erosion in the
the geometric and natural breaks, (Fig. 8). The study area is low.
obtained results presented in table 5 were also The spatial pattern of classified soil erosion rate
further reduced into two categories to facilitate indicates that the areas with moderate to high risks
comparing the process (Table 6). This table also are located in the northeast and just a few in the east,
contains a correlation between slope categories and while the minimal to low zones comprises other
soil hazard risk zones. The area covered by minimal- parts of the study area. From tables 4 and 5 it is
low soil hazard zones extend over an relative area of obvious that there is very strong relationship
between slope and soil loss categories.

Legend

-1 -1
Figure 7. Spatial distribution of annual soil erosion (t ha yr ), the annual soil loss rate estimated for the study
area ranges from 0 to 245 (t ha-1 yr-1).

Table 5. Comparison of calculated soil loss using four different classification schemes
Geometric Quintile Natural breaks Standard deviation
Soil risk
Pixels % Pixels % Pixels % Pixels %
Minimal 1147762 0.530 1038949 0.480 2045609 0.944 1904913 0.879
Low 761225 0.351 864820 0.399 115170 0.053 168769 0.078
Moderate 255036 0.118 194906 0.090 5425 0.003 47764 0.022
High 2454 0.001 67802 0.031 273 0.000 45031 0.021

Table 6. The relation between slope and soil erosion categories

Slope categories % Soil risk Geometric Natural breaks

gentle flat - flat 0.970 minimal-low 0.881 0.997
moderately steep-steep 0.031 moderate-high 0.119 0.003

162
 Legend Legend

Figure 8. Comparison between (a) natural breaks and (b) geometric classification schema for soil loss hazard zones
identification.

5. CONCLUSIONS were classified into four soil erosion hazard zones:

minimal, low, high, and very high by using different
The main objective of this study was to assess classification schema. The area covered by minimal-
the spatial distribution of annual soil loss over the low soil hazard zones covers a relative area of 88% -
northern part of Kirkuk Governorate, Iraq using 99% based on geometric and natural breaks
RULSE model. The five factors of RULSE were schemes, respectively .In turn, the moderate-high
derived from different resources. The R factor was soil hazard zones only cover very small area (0.3%)
estimated from annual and monthly rainfall data of based on natural breaks and relatively small area
meteorological stations close and around the study (12%) depending on geometric scheme. The areas at
area. The calculated R values range from 108.26 to moderate to high risk are found primarily in the
716.28 with average value of 368.40 (MJ mm ha-1 h- north-eastern part of the study area and strongly
1
yr-1). related with slope. Soil conservation practices have
The K factor was estimated from collected to be implemented to reduce soil erosion. Generally,
soil samples and the calculated K values vary from the obtained results indicate that hazard of soil
0.04 to 0.38 with an average of 0.14 (t ha-1 h-1 MJ erosion in the study area is low. To get more
mm-1). The topographic factor (LS) was estimated accurate soil erosion status in the study area, it is
from DEM with 30 m resolution. The range of LS recommended to derive land use/land cover map and
factor was from 0 to 10.76 with an average value of relate it to soil erosion map.
0.02. The C factor was estimated from remote
sensing data and by using Normalized Difference REFERENCES
Vegetation Index (NDVI). Minimum, maximum,
Al-Sayab A.A., Hassan H.A., Ayob M.S., Taha S.H.,
and mean values of estimated C were 0.38, 0.76, and
Salih A.Y. & Faizi K.N., 1983. Water- salt
0.55, respectively. The P factor was assumed to be balance and supplementary irrigation of Alton
one because there is no apparent practice Copry Basin, Tech Rep No 14811 SRC. 89p.
management in the study area. The map of spatial Angima S.D., Stott D.E., O’Neill M.K., Ong C.K. &
distribution of soil loss revealed that the average of Weesies G.A., 2003. Soil erosion prediction using
soil loss was found between 0 - 245 (t ha-1 yr-1) with RUSLE for Central Kenyan highland conditions.
an average of 2 (t ha-1 yr-1). The soil erosion values Agric Ecosyst Environ 97(1–3):295–308.

163
Arekhi S., Niazi Y. & Kalteh A., 2010. Soil erosion and watershed prioritization using GIS and remote
sediment yield modeling using RS and GIS sensing. Water Resour Manage 24:2091–2112.
techniques: a case study, Iran. Arab J Geosci. 5(2): doi: 10.1007/s11269-009-9540-0.
285-297. doi: 10.1007/s12517-010-0220-4. Jasrotia A.S. & Singh R., 2006. Modeling runoff and
Arnoldus H.M.J., 1980. An approximation of the rainfall soil erosion in a catchment area, using the GIS, in
factor in the universal soil loss equation. In: De the Himalayan region, India. Environ Geol 51:29–
Boodt M, Gabriels D (eds) Assessment of erosion. 37.
Wiley, Chichester, pp 127–132. Kefi M., Yoshino K. & Setiawan Y., 2011. Assessment
Ashiagbor G., Forkuo E.K., Laari P. & Aabeyir R., and mapping of soil erosion risk by water in
2013. Modeling soil erosion using RUSLE and Tunisia using time series MODIS data. Paddy
GIS tools. International journal of remote sensing Water Environ. 10(1): 59-73. doi: 10.1007/s10333-
& geosciences 2(4): 7-17. 011-0265-3.
Buday T. & Jassim S.Z., 1987. The Regional Geology of Kim J., Saunders P. & Finn J., 2005. Rapid assessment
Iraq, Vol.2: Tectonism, Magmatism, and of soil erosion in the Rio Lempa basin, central
Metamorphism. GEOSURV, Baghdad, 352 p. America, using the universal soil loss equation and
Buringh D.P., 1960. Soils and Soil Condition of Iraq. GIS. Environ Manage 36: 872–885. doi:
Ministry of Agriculture, Baghdad, Iraq, 337 p. 10.1007/s00267-002-0065-z.
Chen T., Niu R., Wang Y., Li P., Zhang L. & Du B., Kouli M., Sopios P. & Vallianatos F., 2009. Soil erosion
2011. Assessment of spatial distribution of soil prediction using the Revised Universal Soil Loss
loss over the upper basin of Miyun reservoir in Equation (RUSLE) in a GIS framework, Chania,
China based on RS and GIS techniques. Environ Northwestern Crete, Greece. Environ Geol
Monit Assess 179:605–617. doi: 10.1007/s10661- 57:483–497.
010-1766-z. Krishna Bahadur K.C., 2009. Mapping soil erosion
Das G., 2002. Hydrology and soil conservation susceptibility using remote sensing and GIS: a case
engineering. Prentic-Hall of India, 459 p. of the upper Nam Wa watershed, Nan Province,
Demirci A. & Karaburun A., 2012. Estimation of soil Thailand. Environ Geol 57:695–705. doi:
erosion using RUSLE in a GIS framework: a case 10.1007/s00254-008-1348-3.
study in the Buyukcekmece Lake watershed, Lal R., 1994. Water management in various crop
northwest Turkey. Environ Earth Sci 66:903–913. production systems related to soil tillage. Soil
doi: 10.1007/s12665-011-1300-9. Tillage Res 30: 169-185.
Dumas P., Printemps J., Mangeas M. & Luneau G., Lee S., 2004. Soil erosion assessment and its verification
2010. Developing erosion models for integrated using the universal soil loss equation and
coastal zone management: a case study of the new geographic information system: a case study at
Caledonia West Coast. Marine Pollution Bull Boun, Korea. Environ Geol 45:457–465. doi:
61(7): 519-529. 10.1007/s00254-003-0897-8.
Erdogan E.H., Erpul G & Bayramin İ., 2007. Use of Lu D., Li G., Valladares G.S. & Batistella M., 2004.
USLE/GIS methodology for predicting soil loss in Mapping soil erosion risk in Rondonia, Brazilian
a semiarid agricultural watershed. Environ Monit Amazonia: using RUSLE, remote sensing and
Assess 131:153–161. doi: 10.1007/s10661-006- GIS. Land Degrad Dev 15:499–512.
9464-6. Mhangara P., Kakembo V. & Lim K., 2012. Soil
ESRI., 1999. Getting started with ArcGIS. USA. 260p. erosion risk assessment of the Keiskamma
Foster G.R., Lane L.J., Nowlin J.D., Laflen J.M. & catchment, South Africa using GIS and remote
Young R.A., 1980. A model to estimate sediment sensing. Environ Earth Sci 65:2087–2102. doi:
yield from field-sized areas: development of 10.1007/ s12665-011-1190-x.
model. In: Knisel WG (ed) CREAMS: a field scale Millward A.A. & Mersey J.E., 1999. Adapting the
model for chemicals, runoff, and erosion from RUSLE to model soil erosion potential in a
Agricultural Management Systems, US Dept. of mountainous tropical watershed. Catena
Agric., Sci. and Educ. Admin., Conser. Rep. No. 38(2):109–129. doi:10.1016/S0341-
26, pp 36–64. 8162(99)00067-3.
Haddad H.R., Al-Jawad S.B., Haddad I., Younan A.I., Morgan R.P..C, Quinton J.N., Smith R.E., Govers G.,
Salvo A.V., 1971. Hydrogeological investigation Poesen J.W.A., Auerswald K., Chisci G., Torri
in Jolak Basin of the Alton Copry area. Tech Rep D. & Styczen M.E., 1998. The European soil
No 25 Institute for Applied Research on Natural erosion model (EUROSEM): a dynamic approach
Resources, 38p. for predicting sediment transport from fields and
Ismail J. & Ravichandran S., 2008. RULSE2 model small catchments. Earth Surf Proc Land 23:527–
application for soil erosion assessment using remote 544. doi:10.1002/(SICI) 1096-9837(199806).
sensing and GIS. Water Resour Manage (2008) Nearing M.A., Foster G.R., Lane L.J. & Finkner S.C.,
22:83–102. doi: 10.1007/s11269-006-9145-9. 1989. A process based soil erosion model for
Jain M.K. & Das D., 2010. Estimation of sediment yield USDA-water erosion prediction project
and areas of soil erosion and deposition for technology. T ASAE 32(5): 1587–1593.

164
Onori F., De Bonis P. & Grauso S., 2006. Soil erosion information technology. Geoscience Frontier 3(2):
prediction at the basin scale using the revised 209-215.
universal soil loss equation (RULSE in a Renard K.G., Foster G.R., Weesies G.A., McCool D.K.
catchment of Sicily (southern Italy). Environ Geol & Yoder D.C., 1997. Predicting Soil Erosion by
50: 1129–1140. doi: 10.1007/s00254-006-0286-1. Water: A Guide to Conservation Planning with the
Onyando J.O., Kisoyan P. & Chemelil M.C., 2005. Revised Universal Soil Loss Equation (RUSLE).
Estimation of potential soil erosion for River US Department of Agriculture, Agriculture
Perkerra catchment in Kenya. Water Resour Handbook No. 703: 404 pp.
Manage 19: 133–143. doi: 10.1007/s11269-005- Terranova O., Antronico L., Coscarelli R .& Iaquinta.,
2706-5. 2009. Soil erosion risk scenarios in the
Ozosy G., Aksoy E., Sabri Dirim M. & Tumsavas Z., Mediterranean environment using RULSE and
2012. Determination of soil erosion risk in the GIS: an application mode for Calabria (southern
Mustafakemalpasa river basin, Turkey, using the Italy). Geomorphology 112: 228–245.
revised universal soil loss equation, geographic Ugur Ozcan A., Erpul G., Basaran M. & Emrah
information system, and remote sensing. Environ Erdogan H., 2008. Use of USLE/GIS technology
Manage 50:679–694. doi: 10.1007/s00267-012- integrated with geostatistics to assess soil erosion
9904-8. risk in different land uses of Indagi Mountain
Pandey A., Chowdary V. & Mal B., 2007. Identification Pass-Çankırı, Turkey. Environ Geol (2008)
of critical erosion prone areas in the small 53:1731–1741. doi: 10.1007/s00254-007-0779-6.
agricultural watershed using USLE, GIS and Van der Knijff J.M., Jones R.J.A. & Montanarella L.,
remote sensing. Water Resour Manage (2007) 2000. Soil Erosion Risk Assessment in Europe.
21:729–746. doi: 10.1007/s11269-006-9061-z. EUR 19044 EN. Office for Official Publications of
Pandey A., Mathur A., Mishra S. & Mal B., 2009. Soil the European Communities, Luxembourg, p. 34.
erosion modeling of a Himalayan watershed using Wang G., Gertner G., Fang S. & Anderson A.B., 2003.
RS and GIS. Environ Earth Sci (2009) 59:399– Mapping multiple variables for predicting soil loss
410. doi: 10.1007/s12665-009-0038-0. by geostatistical methods with TM images and a
Park S., Oh C., Jeon S., Jung H. & Choi C., 2011. Soil slope map. Photogramm Eng Remote Sens
erosion rick in Korean watersheds, assessed using 69:889–898.
the revised universal soil loss equation. J Hydrol Wang G., Yu J., Shrestha S., Ishidaira H. & Takeuchi
399: 263–27. K., 2010. Application of a distributed erosion
Parson R.M., 1955. Groundwater resource of Iraq, vol.4 mode for the assessment of spatial erosion patterns
Kirkuk Liwa. Development Board, Ministry of in the Lushi catchment, China. Environ Earth Sci
Development, Government of Iraq, 142 p. 61:787–797. doi: 10.1007/s12665-009-0391-z.
Perović V., Životić L., Kadović R., Ðorđević A., Wischmeier W.H., Smith D.D. (1958) – Rainfall energy
Jaramaz D., Mrvić V. & Todorovic M., 2012. and its relationship to soil loss. Amer.
Spatial modeling of soil erosion potential in a Geophysical Union Transactions of the ASAE, 12.
mountainous watershed of south-eastern Serbia. Wischmeier W.H., Smith D.D. (1960) – A universal soil
Environ Earth Sci 68: 115-128. doi: loss estimation equation to guide conservation
0.1007/s12665-012-1720-1. farm planning. Trans. 7th Congr. Int. Soil Sci. Soc.
Pourghasemi H.R., Pradhan B., Gokceoglu C. & 1.
Moezzi KD., 2013. A comparative assessment of Wischmeier W.H. & Smith D.D., 1978. Predicting
prediction capabilities of Dempster-Shafer and rainfall erosion loss: a guide to conservation
weights-of-evidence models in landslide planning. Agricultural Handbook No. 537, US
susceptibility mapping using GIS. Geomatics, Nat Department of Agriculture, Agricultural Research
Hazards and Risk 4(2): 93-118. Service, Washington, DC. 407 p.
Prasannakumar V., Shiny R., Geetha N. & Vijith H., Yue-Qing X., Xiao-Mei S., Xinag-Bin K., Jian P. &
2011. Spatial prediction of soil erosion risk by Yun-long C., 2008. Adapting the RULSE and GIS
remote sensing, GIS, RULSE approach: a case to model soil erosion risk in a mountains karast
study of Siruvani river watershed in Attapady watershed, Guizhou Province, China. Environ
valley, Kerala, India. Environ Earth Sci (2011) Monit Assess (2008) 141:275–286. doi:
64:965–972. doi: 10.1007/s12665-011-0913-3. 10.1007/s10661-007-9894-9.
Prasannakumar V., Vijith H., Abinod S. & Geetha N., Zhou P., Luukkanen O., Tokola T. & Nieminen J.,
2012. Estimation of soil erosion risk within a small 2008. Effect of vegetation cover on soil erosion in
mountains sub-watershed in Kerala, India, using a mountainous watershed. Catena 75:319–325.
revised universal loss equation (RULSE) and geo-

Received at: 12. 02. 2015

Revised at: 09. 09. 2015

165
 Accepted for publication at: 06. 11. 2015
Published online at: 09. 11. 2015

166

View publication stats