Environmental Earth Sciences (2021) 80:511

[Link]

ORIGINAL ARTICLE

Risk modelling of soil erosion in semi‑arid watershed of Tamil Nadu,

India using RUSLE integrated with GIS and Remote Sensing
P. Sandeep1 · K. C. Arun Kumar1 · S. Haritha2

Received: 2 July 2020 / Accepted: 30 July 2021 / Published online: 9 August 2021
© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract
Water-induced soil erosion is one of the challenging threats in various parts of the world, thus systematic investigations of
soil loss risk are more crucial for sustainable agricultural production and water management. The present study was carried
out to estimate the average soil loss in the Amravati watershed of Tamil Nadu state in South India using Revised Universal
Soil Loss Equation (RUSLE) along with Geographical Information System (GIS) and Remote Sensing techniques. In this
study, the rain gauge-based rainfall data were completely substituted with the CHIRPS datasets to compute the R factor, since
it gives continuous surface rainfall data rather than location-based measurement. The estimated soil loss in the watershed
was found to be in the range of 0–280.2 t/ha−1/yr−1, in which about 64.7% of the watershed under very low erosion risk
whilst about 12.9% of the area prone to moderately high-to-very high erosion risk. The maximum soil loss rate was identi-
fied in the Degraded forest with a mean loss of 59.51 t/ha−1/yr−1 followed by the Degraded plantation (32.17 t/ha−1/yr−1),
Scrubland/Wasteland (17.75 t/ha−1/yr−1), Current fallow (12.08 t/ha−1/yr−1), and Rainfed cropland (10.03 t/ha−1/yr−1). The
study shows that the present land use / land cover and the landscape of the watershed have a great influence on soil loss. To
check the efficiency of the RUSLE model for the assessment of soil erosion risk, the final derived output was validated with
the use of 4k UHD Google Earth images.

Keywords RUSLE · Soil erosion · CHIRPS · GIS and remote sensing · Amravati watershed

Introduction 2019); in addition, anthropogenic activities, such as inten-

sive farming, inadequate land management, clearance of for-
Water-induced soil erosion is considered as the critical envi- est, and agricultural activities on steep slopes further elevate
ronmental issue leading to land degradation by removing the rate of erosion (Jayasekara et al. 2018). The soil loss rate
the fertile rich topsoil (Gelagay and Minale 2016; Ganasri is highly controlled by many factors, such as the amount of
and Ramesh 2016; Chen et al. 2017). Soil erosion leads to rainfall and its intensity, anthropogenic activities, soil char-
the loss of soil nutrients and degrades the productivity of acteristics, and the landscape of the region (Thomas et al.
cultivable land (Dutta et al. 2015; Shit et al. 2015; Srinivasan 2018; Duarte et al. 2016; Galdino et al. 2015; Amin and
et al. 2019; Qin et al. 2018); moreover, the transported mate- Romshoo 2018). The rill and sheet erosion are the most com-
rials contaminate the downstream water (Zhu et al. 2013; mon types of soil erosion in many parts of the globe, during
Wang et al. 2016). As a natural process, soil erosion refers sheet erosion, soil particles get detached by the impact of
to the detachment, transportation, and accumulation of soil water droplets and transported through surface runoff on
particles through erosive agents, such as surface water, wind, the other hand rill erosion occurred by the action of surface
and gravity (Gelagay and Minale 2016; Panditharathne et al. runoff (Steinmetz et al. 2018). About 85% of the land deg-
radation across the globe is due to soil erosion (Tang et al.
* P. Sandeep 2015), thus it has become a worldwide concern in modern
deepugeo93@[Link] times especially in those countries mainly dependent on the
agricultural sector (Samantha et al. 2016; Saha et al. 2018).
1
ICAR-National Bureau of Soil Survey and Land Use The assessment and monitoring of soil erosion have great
Planning, Amravati Road, Nagpur 440 033, India
importance to take better mitigation and prevention plans for
2
Department of Geography, Bharathidasan University, sustainable agricultural production and water management.
Tiruchirappalli, Tamil Nadu 620 024, India

13
Vol.:(0123456789)
511 Page 2 of 20 Environmental Earth Sciences (2021) 80:511

As per the report of the National Bureau of Soil Survey Amravati river, one of the main tributaries of river Cauvery
and Land Use Planning (NBSS&LUP), in India, about and that flows from south to north through the study area.
146.8 million hectares (Mha) of the land has been affected The altitude of the study area varies from 130 to 2414 m from
by soil degradation (NBSS&LUP 2005). Soil from semi- the mean sea level (MSL) with varied topography, such as
arid and arid regions is more easily eroded naturally due to valley, plain, uplands, dissected hills, hillocks, etc. The hill
the absence of thick protective vegetative cover (Miller and ranges are mostly covered with mixed forest, while the plain
Donahue 1990). Tamil Nadu state in South India experiences and the lower valley of the study area are under intensive cul-
a semi-arid climate, where about 4.9 Mha soil has degraded tivation. The geological unit of the study area encompasses
by water (NBSS&LUP 2005). The traditional field-based Hornblende–biotite gneiss and Charnockite (Thirunavukkar-
assessment of soil erosion risk is relatively expensive and asu and Ambujam 2020). Geomorphologically the study area
time-consuming; furthermore, it fails to give an accurate comes under two different origins, namely, structural and
estimation for a wide area (Teng et al. 2016). Recently, sev- denudation origins (Thirunavukkarasu and Ambujam 2020).
eral convenient and scientifically advanced models (Physical The major crops grown in the study area are paddy, veg-
and Empirical) have been used to quantify soil erosion risk etables, groundnut, sugarcane, banana, finger millet, maize,
in different parts of the world. Among these models Revised coconut, dry crops, etc.
Universal Soil Loss Equation (RUSLE) (Renard et al. 1997)
is one of the most popular and widely applied empirical
models especially for the assessment of soil loss through rill Materials and methods
and inter-rill erosion (Rellini et al. 2019). Unlike all other
models, RUSLE is a simple, flexible, cost- and time-effec- Datasets used
tive mathematical model, besides all the five inputs for the
RUSLE model can be developed only from four datasets, viz The datasets used in the present study include Landsat 8
Soil, Rainfall, Slope, and Land Use/Land Cover (LU/LC) OLI/TIRS multispectral satellite image, The Shuttle Radar
(Zerihun et al. 2018; Thlakma et al. 2018). The application Topography Mission Digital Elevation Model (SRTM
of Geospatial techniques facilitates detailed spatio-temporal DEM). The Climate Hazards Group Infrared Precipitation
analysis of soil loss with reasonable cost (Prasannakumar with Stations (CHIRPS) rainfall data, and the Soil character-
et al. 2011; Toubal et al. 2018; Jazouli et al. 2017). The istic data. The satellite image and the elevation model for the
integration of RUSLE with Geographic Information System study area were downloaded from the United States Geologi-
(GIS) and Remote Sensing techniques offers more precise cal Survey (USGS) website ([Link]
and accurate cell by cell estimation of soil loss in a large with a spatial resolution of 30 m. The LU/LC layer of the
area (Zerihun et al. 2018; Beyene 2019; Kayet et al. 2018; study area was prepared from the atmospherically corrected
Adongo et al. 2019). Today, various soil erosion models are Landsat 8 OLI/TIRS image. Subsequently, the downloaded
being carried out using medium-to-fine resolution remote- DEM was filled using the ‘Fill tool’ in ArcGIS software,
sensing datasets (Cunha et al. 2017; Ali and Hagos 2016), the process involves the filling of all the depressions in the
while GIS allows analyzing, processing, and mapping of DEM ESRI (2001), later which were used to generate the
the RUSLE input and output datasets (Anees et al. 2018; terrain parameters, namely, slope and flow accumulation for
Belayneh et al. 2019). The present study is aimed to estimate the computation of slope length and steepness (LS) factor
and map the average soil loss rate of the Amravati watershed of the study area. Moreover, the boundary of the watershed
by using the RUSLE model. was delineated from the same dataset by using ‘Arc hydro
extension tool’ in ArcGIS software. The CHIRPS a hybrid
Study area gridded time series (Geleta and Deressa 2019; Sulugodu and
Deka 2019) monthly rainfall data for the period of 30 years
The study was conducted in the Amravati watershed of (1990–2019) were downloaded from [Link]
Tamil Nadu state in South India and its geographical location edu. with a spatial resolution of 5 km which were used to
extends from 10° 10′ 51.65ʹʹ N to 10° 47′ 23.8″ N latitudes develop the R factor. In addition, the detailed soil data for
and 77° 2′ 13.07″ E to 77° 34ʹ 5.69″ E longitudes, as shown the study area was obtained from the Tamil Nadu Agricul-
in Fig. 1. The watershed spreads the Thirupur and Dindugal tural University (TNAU) with a scale of 1: 50,000 km which
districts of Tamil Nadu state with an area of 1675.8 sq. km. consists the information, such as percentage of sand, silt, and
The study area experiences a semi-arid climate, receives a clayey, soil texture, and percentage of organic matter (OM),
mean annual rainfall of 720 mm, in which about 60% of the structural and permeability codes, etc., these information
rainfall is contributed by a retreating monsoon and about 40% are very crucial for the computation of soil erodibility factor
is received during the southwest monsoon season (Senthil- (R) in RUSLE model (chadli 2016; Gianinetto et al. 2019).
kumar et al. 2019). The watershed is mainly drained by the Table 1 shows the data sets used in the study.

13
Environmental Earth Sciences (2021) 80:511 Page 3 of 20 511

Fig. 1  Location map of the

study area

Generation of land use/land cover map (LU/LC)

Table 1  Datasets used in the study
The LU/LC layer of the study area was generated from Land-
SI. No Datasets Resolution Source
sat 8 OLI/TIRS data by using the ‘multiresolution segmenta-
1 Landsat 8 OLI/ 30 m [Link] tion technique’ in eCognition software which helps to sys-
TIRS tematically differentiate the spectral values and accordingly
2 SRTM DEM 30 m [Link] vectorize them (Yuhendra 2019; Lebourgeois et al. 2017).
3 CHIRPS 5 km [Link] Later the vectorized layer was opened in ArcGIS software
4 Soil data 1:50,000 Tamil Nadu Agriculture and each LU/LC class in the watershed was identified by
University
visual interpretation technique subsequently, the LU/LC

13
511 Page 4 of 20 Environmental Earth Sciences (2021) 80:511

classes were properly fed into the attribute table. Mainly 13 Quite number of equations have been developed for the com-
LU/LC classes were identified in the watershed, namely, Irri- putation of the R factor; however, in the present study, the
gated cropland, Rainfed cropland, Current fallow, Plantation, erosivity index was calculated using Eq. 2 by Singh et al.
Scrubland/Wasteland, Mining area, Industrial area, Built-up (1981), as it is more applicable for the Indian context. The
land, River/Waterbodies, Degraded forest, Degraded planta- same equation has been adopted to compute the R factor by
tion, Deciduous forest, and Evergreen/Semi-evergreen for- Bera 2017; Saha et al. 2018; Dutta et al. 2015 Sangeetha
est. For the purpose of checking the accuracy level of the and Ambujam 2021. The equation shows the relationship
created LU/LC layer, 400 sample points were randomly col- between the rainfall erosivity index and the mean annual
lected from the LU/LC layer and those layers were opened in rainfall. Due to the limited availability of rain gauge data for
Google Earth Pro to confirm that whether the created LU/LC the study area, the CHIRPS monthly satellite-based gridded
layer of the watershed match with the respective LU/LC in rainfall data were used in the present study to compute the R
the Google Earth high-definition natural color image or not. factor, it is a continues raster rainfall data available at 5 km
With the help of this validation result, a confusion matrix spatial resolution for the whole world. As a first task, the
was created and obtained an overall accuracy of 87.98% with monthly rainfall data from 1990 to 2019 were downloaded
a Kappa coefficient value of 0.82 and the result indicates that and resampled to 30 × 30 m grid size by using the ‘bilinear
the classification was well performed. interpolation technique’ in ArcGIS software. Following this,
the annual rainfall raster of each year from 1990 to 2019
RUSLE model description was produced from the monthly CHIRPS rainfall datasets
by using the ‘Cell Statistics tool’ in ArcGIS. Subsequently
RUSLE is an empirical model that is widely used to estimate the 30-year long-term mean single rainfall raster was devel-
the annual average soil loss in an area. The RUSLE model oped from the processed annual rainfall raster by using the
considers 5 different parameters that are directly related ‘Cell Statistics tool’ and finally the R factor was developed
to climate, soil properties, terrain characteristics, and LU/ by using the following equation:
LC (Ruiz et al. 2015; Vijith et al. 2017). The estimation of
annual average soil loss per unit area per year of the study
R = 79 + 0.363 × RN (2)
area was assessed as per the following equation developed where RN is the long-term mean annual rainfall in mm.
by Wischmeier and Smith (1978):
A = R × K × LS × C × P (1)
K factor
−1 −1
where A represents mean annual soil loss (t ­ha ­yr ), R
refers to rainfall erosivity factor (MJ mm h­ a−1 ­h−1 yr-1), K K factor indicates the erosion vulnerability of soil by rainfall
denotes the soil erodibility factor (t ha h ­MJ−1 ­ha−1 ­mm−1), and associated surface runoff (Islam et al. 2018; Mustefa
LS is the slope length and steepness factor, C and P are the et al. 2019; Lee et al. 2017; Pradhan et al. 2018). The differ-
crop management and support practice factors, respectively. ent types of soil have different levels of resistance in terms
Among these factors LS, C, and P factors are dimensionless. of erosion. The erodibility rate of the soil highly depends
Table 2 shows RUSLE parameters and their parent datasets. on its physical and chemical properties, such as soil texture,
soil depth, organic matter, structure, and permeability (Wis-
RUSLE parameter estimation chmeier and Smith 1978). The soil erodibility factor was
computed by using the equation given by Wischmeier and
R factor Smith 1978, which determines the K factor, as a function
of the complex interaction of percent of sand, silt, and clay,
R factor implies the rainfall effect on soil loss, since its vol- organic matter, structural, and permeability codes of a given
ume, intensity, and duration trigger the rate of rill and inter- soil. The K factor values are ranging from 0 to 1, where a
rill erosion (Kayet et al. 2018; Gelagay and Minale 2016). value close to 1 indicates the soil is highly vulnerable to

Table 2  Parent datasets and SI. No Parent datasets Derived parameters

derived parameters
1 Long-term average rainfall Rainfall erosivity factor (R)
2 Soil properties Soil erodibility factor (K)
3 Digital Elevation Model (DEM) Slope length and steepness factor (LS)
4 LU/LC or NDVI Crop management factor (C)
5 LU/LC or NDVI Support practice factor (P)

13
Environmental Earth Sciences (2021) 80:511 Page 5 of 20 511

erosion, while values approaching 0 denote soil has a good cover significantly reduces the rate of erosion by decreasing
resistance capacity on erosion (Mustefa et al. 2019). In the runoff velocity and control the raindrop impact on the
this study, the K factor values were computed from the soil soil surface (Ranzi et al. 2012), thus the vegetation cover is
characteristic data of the study area by using the following considered as the second most significant factor after topog-
equation: raphy that controls soil loss rate (Thiaw and Dacosta 2017;

K = 2.1 × 10−4 (12 − OM) × M 1.14 + 3.25 (s − 2) + 2.5 (p − 3)∕759.4 (3)

where M is the particle size which expressed as Li et al. 2019). The C factor values are ranged from 0 to 1,
where 0 represents the area completely free from erosion due
M = (% of silt + %of very fine sand) × 100 − % of clay
to the presence of thick vegetative cover, while 1 indicates
(4)
the area is highly prone to soil erosion due to the lack of suf-
where OM is the organic matter content, ficient vegetative cover. The present study adopted LU/LC
S and P are the structural and permeability codes, based method for assessing C factor values of the study area,
respectively. since it gives more accurate values as compared to all other
After the computation of the K factor value of each soil approaches (Lin et al. 2017), this method has recently been
type in the study area, a new field was created in the attribute adopted by many researchers (Belayneh et al. 2019; Cunha
table of the soil shapefile. Subsequently, the derived K factor et al. 2017; Paul et al. 2019; Pritam et al. 2018; Gashaw et al.
values were assigned to corresponding soil types in the layer 2019). Following this, the C factor values were assigned
accordingly this particular field was converted into raster by to each LU/LC class of the study area (Table 3) as per the
using the ‘Feature to Raster tool’ in ArcGIS software with previous literature (USDA-SCS 1972; Shit et al. 2015; Bal-
a 30 × 30 m grid size. asubramani et al. 2015). Finally, the particular field which
has the R factor values were converted into raster format
LS factor with a 30 × 30 m grid size.

LS factor reflects the influence of landscape on soil erosion, P factor

steep slope and heavy rainfall make an area more vulnerable
to soil erosion. If the length and steepness of the slope are The P factor reflects the favorable effect of soil–water
bigger the volume of runoff and its velocity increase and conservation measures on soil erosion. To put it another
lead to erosion (Reddy et al. 2016; Galdino et al. 2015; Paul way, it is the proportion of erosion rate in conditions of
et al. 2019). The DEM could offer a better measurement of protective operations to the erosion take place in the nor-
the LS factor for the given area (Panagos et al. 2015). For the mal condition (Wischmeier and Smith 1978). The P fac-
computation of LS factor, the flow accumulation and slope tor value is ranging from 0 to 1; a value standing close to
raster of the study area were generated from the processed 0 indicates the area under well soil–water conservation
30 m SRTM DEM afterwards the LS factor for the study area
was successively generated by executing the Eq. 5 by Moore
and Burch 1986 in ‘Raster Calculator tool’ (ArcGIS): Table 3  C factor values for corresponding LULC classes

LS = ([Flow accumulation]) × [Cell size]∕22.130.4 SI. No LULC classes C values

1.3
(5)
× (sin slope∕0.0896) 1 Irrigated crop land 0.28
2 Rainfed crop land 0.5
where LS is the slope length and steepness factor, Cell size is
3 Plantation 0.28
the pixel size of the DEM, and sin slope is the slope degree
4 Current fallow 0.5
in sine.
5 Scrub land/wasteland 0.8
6 Degraded plantation 0.8
C factor 7 Degraded forest 0.8
8 Evergreen/semi evergreen forest 0.004
The C factor expresses the relationship between the cover 9 Deciduous forest 0.008
management practices and soil erosion rate (Koirala et al. 10 Mining area 0.18
2019; Reddy et al. 2016). It is the ratio of soil loss from the 11 Industrial area 0
cropped land to the corresponding loss from the bare soil or 12 Built-up land 0
unprotected soil (Wischmeier and Smith 1978). Vegetative 13 River/waterbodies 0

13
511 Page 6 of 20 Environmental Earth Sciences (2021) 80:511

practices, whereas values approach 1 indicate there are Results

no proper conservation measures adopted. In this study,
the P factor layer was developed based on the LU/LC of R factor
the study area and the appropriate P factor values for cor-
responding LU/LC types were collected from the previous The erosivity rate is always correlated with the amount
literature (Dutta et al. 2015; Jena et al. 2018; Chandramo- of rainfall received by an area. The R factor values of the
han et al. 2015; Brema and Hauzinger 2016). Later the watershed vary from 256.54 to 637.35 MJ mm ­ha−1 ­h−1
values were assigned to each LU/LC class according to ­y−1 and they were categorized into 5 classes based on
the control practices that exist in the different LU/LC cat- their intensity as < 400, 400–450, 450–500, 500–550,
egories (Table 4) after that the layer was converted into and > 550 MJ mm ­ha−1 ­h−1 ­y−1 (Fig. 3) The southern most
a raster with a 30 × 30 m grid size. Figure 2 shows the tip of the watershed is distributed by highest R value class,
complete methodology of the study. i.e., 550 MJ mm ­ha−1 ­h−1 ­y−1 as this region experiences
high rainfall during the retreating monsoon season. At
the same time, the maximum hilly portion of the water-
Table 4  P factor values of different land use/land cover types
shed in the southern part is distributed by the next two
SI. No LU/LC classes P values highest R value classes, 500 –550 mm ­ha−1 ­h−1 ­y−1 and
450–500 mm ­ha−1 ­h−1 ­y−1 respectively. As a result, these
1 Irrigated crop land 0.5
highest R factor values play a significant influence on soil
2 Rainfed crop land 1
erosion in this part of the study area. While the small-
3 Plantation 0.8
est values (< 400) were identified in the northern part of
4 Current fallow 1
the study area for the reason that the rainfall distribution
5 Scrub land/wasteland 1
of the study area decreases towards its north. The center
6 Degraded plantation 1
part of the watershed has R factor values between 400 and
7 Degraded forest 1
500 MJ mm ­ha−1 ­h−1 ­y−1.
8 Evergreen forest 1
9 Deciduous forest 1
10 Mining area 1
K factor
11 Industrial area 1
The main soil types in the Amravati watershed are Clay,
12 Built-up land 1
Clay loam, Loamy sand, Sandy clay, Sandy clay loam, and
13 River/waterbodies 1

Fig. 2  Flow chart of the complete methodology

13
Environmental Earth Sciences (2021) 80:511 Page 7 of 20 511

Fig. 3  R factor

Sandy loam. About 26.54% of the total geographical area 0.03 to 0.05 t ha ­MJ−1 ­mm−1 and are mostly located in the
(TGA) of the watershed is covered by Clay soil followed southern and eastern portions of the watershed covered
by Sandy loam (20.41%), Sandy clay loam (18.86%), Clay in sandy clay loam soil. The K value classes 0.05–0.07
loam (16.07%), Sandy clay (10.79), and Loamy sand (2%). and > 0.07 t ha ­M J −1 ­m m −1 were found few patches in
The estimated K factor values of the watershed range from the north, northeast, northwest, and some pockets of the
0 to 0.09 (Fig. 4). The low K factor values are associated southern part of the watershed, because these regions are
with Clay and Clay loam soils (< 0.03 ha ­M J −1 ­m m −1) covered by sandy loam and loamy sand soils. These soils
primarily located in the central, few patches in the south, are less cohesive and unstable thus easily subjected to ero-
north, and northwestern portions of the watershed. Major sion by water.
parts of the watershed have K factor values ranging from

13
511 Page 8 of 20 Environmental Earth Sciences (2021) 80:511

Fig. 4  K factor

LS factor associated with plains and valleys because of the low slope
gradient. On the contrary, the hilly portions in the southern
The elevation of the Amravati watershed varies from 130 to part of the study area have very high (> 20) LS factor values,
2414 m (Fig. 5), the major part of the watershed falls under this is because when the slope gradient increases the value
plain and valley. The LS factor of the study area ranges from of the LS factor also increases.
0 to 37.16, with a mean value of 4.08 and a standard devia-
tion of 7.71. The developed LS factor was classified into C factor
5 distinct class viz very low (< 5), low (5–10), moderate
(10–15), high (15–20), and very high (> 20). The major- The C factor values of the watershed vary from 0 to
ity of the watershed has a very low LS factor value (< 5), 0.8, the highest value (0.8) was assigned to Scrubland/

13
Environmental Earth Sciences (2021) 80:511 Page 9 of 20 511

Fig. 5  LS factor

Wasteland, Degraded plantation, and Degraded forest, soil is well-protected by an adequate vegetative cover,
where erosion rate is probably higher due to insufficient mainly identified along the valley (Fig. 6). The forest
vegetative cover to protect soil from erosion by the severe usually protects the surface soil by regulating the run-
runoff and raindrop effect. Similarly, the chance of ero- off velocity moreover, it prevent the raindrop effect on
sion will be more in the Rainfed cropland and Current soil. The southern part of the study area is dominated by
fallow land, since it is not under cultivation for the whole forest which comprises both Deciduous and Evergreen/
season in a year, thus assigned a second highest C value Semi-evergreen types, a C factor value of 0.004 assigned
(0.5) to these LU/LC classes. The Plantation and Irrigated to Evergreen/Semi-evergreen forest and the value 0.008
croplands reflect a very small C factor value (0.28) as the was assigned to the Deciduous type.

13
511 Page 10 of 20 Environmental Earth Sciences (2021) 80:511

Fig. 6  C factor

P factor Degraded forest, Mining area, Industrial area, Built-up

land, and River/Waterbodies.
The P factor value of the watershed is ranged from 0.5
to 1 (Fig. 7). The lowest value, i.e., 0.5, is given to Irri- Estimation of mean annual soil loss using RUSLE
gated cropland, where practices intensive agriculture with
proper conservation measures, while the P factor value 0.8 The final pixel-by-pixel mean annual soil loss raster was
is assigned to land under plantation crops. The rest of the developed in ArcGIS software by using empirical Eq. 1. The
land is not subjected to proper soil–water conservation estimated soil loss in the watershed was found to be in the
measures, so the maximum P value, i.e.,1 was assigned, range of 0–280.2 t/ha−1/yr−1, based on the NBSS&LUP soil
these areas are mainly occupied by forest (Deciduous loss classification schema the final output raster was clas-
and Evergreen/Semi-evergreen), Scrubland/Wasteland, sified into 6 different categories (Fig. 8), namely, very low,
Rainfed cropland, Current fallow, Degraded plantation, low, moderate, moderately high, high, and very high. The

13
Environmental Earth Sciences (2021) 80:511 Page 11 of 20 511

Fig. 7  P factor

study illustrates that about 64.7% and 15.1% of the water- forest also comes under these same erosion risk categories.
shed comes under very low (< 5 t/ha−1/yr−1) and low erosion Whilst approximately 7.3% and 4% of the area was exposed
risks (5–10 t/ha−1/yr−1) respectively, were mostly found in to moderate (10–15 t/ha−1/yr−1) and moderately high ero-
the northern, central and few patches in the eastern, and sion risk (15–20 t/ha−1/yr−1), respectively, mainly observed
southern parts of the watershed mainly associated with lower in the southwestern, southeastern, and northeastern parts
valley and plain. These areas of the watershed are character- of the watershed. About 5.9 and 3.0% of the TGA of the
ized by low R factor values (256.54–500 MJ mm ­ha−1 ­h−1 ­ watershed fall under high (20–40 t/ha−1/yr−1) and very high
y−1) and very low LS factor values (< 10); moreover, the soil (> 40 t/ha−1/yr−1) erosion risk respectively, prominently dis-
is protected by Irrigated crops. Similarly, the hilly portions tributed in the southern part of the watershed associated with
of the study area which cover Evergreen/Semi-evergreen foothills, this part of the study area has very high (> 20) LS

13
511 Page 12 of 20 Environmental Earth Sciences (2021) 80:511

Fig. 8  Mean annual soil loss

values because of the presence of very steep slope (> 35%) analyzed, the statistics are shown in Table 6. The analysis
as well as the region is recorded R factor values between reveals that the present LU/LC practices in the watershed
400 and 500 MJ mm ­ha−1 ­h−1 ­y−1. The eastern and western exert a strong impact on erosion vulnerability and soil loss
margins of the watershed also were observed under a high from the watershed. The study area, mostly covered with
to a very high erosion risk category (Table 5). Deciduous forest, accounts for 30.3% of the TGA, followed
by Irrigated cropland (18.3%), Rainfed cropland (17.5%),
Soil loss over various LU/LC classes Current fallow land (13.5%), and Evergreen/Semi-ever-
green forest (10. 3%). The spatial distribution of distinct
To analyze the effect of LU/LC on soil erosion risk, the LU/LC classes of the watershed is shown in Fig. 9. Among
mean soil loss rate of each LU/LC class was computed and these LU/LC classes Degraded forest, Degraded plantation,

13
Environmental Earth Sciences (2021) 80:511 Page 13 of 20 511

Table 5  Area under different soil erosion degree in Amravati water- Semi-evergreen forest protects the soil surface from the rain-
shed drop impact by its thick vegetation canopies.
Soil erosion rate (t/ha−1/yr−1) Area (ha.) Area (%)

Very low (< 5) 108,422.0 64.7

Discussion
Low (5–10) 25,233.0 15.1
Moderate (10–15) 12,202.8 7.3
Inadequate vegetative cover, isolated high rainfall intensity
Moderately high (15–20) 6785.8 4.0
during the rainy season, unfavorable soil characteristics,
High (20–40) 9926.5 5.9
steep slopes, improper cropping systems, and inadequate
Very high (> 40) 5012.6 3
soil-water conservation practices make the watershed more
Total 167,582.63 100.00
vulnerable to soil erosion. The estimated soil loss in the
watershed was found to be on a scale of 0–280.2 t/ha−1/yr−1.
In this estimation, about 5.9 and 3.0% of the TGA of the
Table 6  Mean soil loss rate (t/ha−1/yr−1) under various LULC classes watershed lies under high (20–40 t/ha−1/yr−1) and very high
of Amravati watershed (> 40 t/ha−1/yr−1) erosion risks respectively, prominently
LULC classes Area (ha) Area (%) Mean soil Mean soil distributed in the southern part of the watershed associated
loss rate loss rate with foothills. The present study exhibits that the high and
(t/ha−1/ (%) very highly vulnerable area to soil erosion is found in the
yr−1)
areas are having higher LS and relatively higher R factor
Built-up land 2565.6 1.5 1.48 0.89 values, as the slope length and its steepness rises the soil
Current fallow 22,652.1 13.5 12.08 7.27 runoff to the downslope direction also increases (Koirala
Deciduous forest 50,753.3 30.3 5.99 3.60 et al. 2019; Senanayake et al. 2020). In addition, the runoff
Degraded forest 401.1 0.2 59.51 35.80 rate is often triggered by isolated high rainfall intensity in
Degraded plantation 1443.3 0.9 32.17 19.36 the semi-arid areas (Thomas et al. 2018). Similar studies
Evergreen forest 17,333.0 10.3 3.82 2.30 regarding high erosion rates associated with steeper slopes
Industrial area 28.1 0.0 7.14 4.30 and isolated high rainfall intensity have been reported in
Irrigated crop land 30,647.6 18.3 3.32 2.0 the various watershed of southern India by Balasubramani
Mining area 322.4 0.2 5.80 3.49 et al. 2015; Markose and Jayappa 2016; Kolli et al. 2021;
Plantation 4492.0 2.7 5.48 3.30 Saravanan et al. 2021. In addition, the soil texture of the
Rainfed crop land 29,364.2 17.5 10.03 6.03 watershed also hikes the soil erosion rate especially from
Scrub land/waste- 5381.7 3.2 17.75 10.68 the croplands, since about 39.27% of the study area is dis-
land tributed by sandy loam and sandy clay loam textures, as
River/waterbodies 2198.3 1.3 1.64 0.99 per the USDA (1983) these two texture classes fall under
Total 167,582.63 100 166.21 100 moderate to rapid and moderate to slow permeability classes
respectively (codes 2 and 4). Consequently, the particle can
be detached and eroded quickly by runoff. A similar result
Scrubland/Wasteland, Current fallow, and Rainfed croplands has been reported from the semi-arid region of Maharashtra
are more vulnerable to erosion with a mean soil loss rate of by Maury et al. 2019, in this study, it was found that the
59.51 t/ha−1/yr−1, 32.17 t/ha−1/yr−1, 17.75 t/ha−1/yr−1, 12.08 areas with steep slopes and sandy loam texture are exposed
t/ha−1/yr−1, and 10.03 t/ha−1/yr−1 respectively, due to the to higher soil loss.
absence of adequate vegetative cover to protect their topsoil. As the present result compared with the previous studies
Rainfed croplands are left exposed until the next growing which have been carried out by using the RUSLE model in
season after annual crops were harvested, during this time different watersheds with similar topography and climatic
likely rate of soil erosion by water is more. Deciduous for- conditions shows that the obtained estimation of soil loss
est on the steep slope (slope > 35%) also contributes much from the present study is quite higher. A study from the
erosion risk in the watershed because of its topographic Kummattipatti Nadi watershed of Tamil Nadu estimated the
influence and relatively high surface runoff caused by high mean annual soil loss ranges from 0–153 t/ha−1/yr−1 (Shaikh
rainfall during the northeast monsoon season. The Irrigated et al. 2020), which shows a relatively lower estimation rate
cropland, Evergreen/Semi-evergreen forest, and Plantation in comparison to the erosion rate estimated from the Amra-
experience the lowest annual soil loss this is because the vati watershed. Similarly, the study conducted in the Sil-
Irrigated cropland and Plantation are mostly laid in the plain lahalla watershed, Nilgiri district of Tamil Nadu state (San-
region (slope 0–8%), as well as the soil, is well-protected by geetha and Ambujam 2021) for the period of 2011–2019,
a sufficient vegetative cover, at the same time, the Evergreen/ also shows a lower rate (51.89 t/ha−1/yr−1) of mean annual

13
511 Page 14 of 20 Environmental Earth Sciences (2021) 80:511

Fig. 9  Land use/land cover of

the study area

soil loss as compared to the rate from Amravati water- shows a more or less similar estimation of annual soil loss
shed. Whereas, the assessed soil loss rate from the adjacent to the estimation of the present study with the range from
watershed (Palayakottai watershed, Pritam et al. 2018) was 0–319.55 t/ha−1/yr−1 (Behera et al. 2020), At the same time,
found to be quite higher than that of the Amravati watershed the present result is comparable to the erosion rates reported
and the analysis reveals that about 17% of the watershed from the various semi-arid environment by using the RUSLE
area is classified under severe to very severe erosion risks model across the world including Morocco (Bou-imajjane
(20–80 t/ha−1/yr−1) which may be attributed to inadequate et al. 2020, 0–160.3 t/ha−1/yr−1), Brazil (Silva et al. 2018,
vegetative cover, isolated high rainfall, unsuitable cropping 0–500 t/ha−1/yr−1), Ethiopia (Bekele 2021, 0–255 t/ha−1/
pattern, and the lack of soil-water conservation measures. yr−1), Saudi Arabia (Mallick et al. 2014, 0–278.65 t/ha−1/
A study carried out in the semi-arid river basin of Odisha yr−1), and Iran (Sardari et al. 2019, 0–254.1 t/ha−1/yr−1).

13
Environmental Earth Sciences (2021) 80:511 Page 15 of 20 511

Whereas, the estimation from the semi-arid region of Alge- land use types which are highly sensitive to soil erosion. The
ria (Djoukbala et al. 2018, 0–17 t/ha−1/yr−1) reveals a com- contour bunding with vegetative barriers could reduce the
paratively lower soil erosion rate than the current result. The soil erosion rate from Rainfed cropland; tillage conserva-
analysis of soil erosion rate under each LU/LC class of the tion and contour bunding are the best conservation meas-
study area shows that the existing LU/LC practices without ures to reduce the soil loss from the Current fallow land.
having proper conservation measures in the watershed have Forest plantation or dry crop cultivation in the Scrubland/
a strong impact on soil erosion vulnerability and soil loss Wasteland will help to control the soil loss rate to some
from the watershed. Among 13 LU/LC classes, Degraded extent. Staggered contour trench in the Deciduous forest
forest (59.51 t/ha−1/yr−1), Degraded plantation (32.17 t/ha−1/ on the foothills will help to reduce the soil erosion rate;
yr−1), Scrubland/Wasteland (17.75 t/ha−1/yr−1), Current fal- Rejani et al. 2016 suggested the same conservation measure
low (12.08 t/ha−1/yr−1) and Rainfed croplands (10.03 t/ha−1/ for a semi-arid watershed cluster of Telangana state. The
yr−1) are more vulnerable to erosion process mainly due to afforestation in the degraded forests will keep these areas
the absence of adequate vegetative protection. Since the rate away from further erosion risk. Figure 10 shows the sug-
of erosion is highly influenced by poor vegetative protection gested soil–water conservation measures for the watershed.
and lack of soil conservation measures more particularly Pushpalatha et al. (2017) have conducted a similar study
in semi-arid regions (Thomas et al. 2018). The study con- in the semi-arid watershed of Telangana state and highly
ducted in Perambalur taluk of Tamil Nadu state (Karthick recommend contour bunding for the Agricultural land in the
et al. 2017) also estimated a nearly comparable amount of watershed to check the further erosion risk. Another study
soil erosion from the forest area (52.25 t/ha−1/yr−1), Thomas has been conducted by Das et al. 2020 at Ri-Bhoi district,
et al. 2018 estimated soil loss for the nearby semi-arid water- Meghalaya and has recommended the contour bunding and
shed and obtained comparable estimation (24.28 t/ha−1/yr−1) terracing for Agricultural land to protect the land from future
from the Open scrubland. The R factor is considered as one erosion risk.
of the important parameters in RUSLE and its computa- The validation of the final output is an integral part of
tion assumes greater importance in this model. Researchers every GIS and Remote Sensing-based study. In the present
have been depending on rain gauge rainfall data to compute study, the final soil loss raster was validated through Google
the R factor for many years. However, developing countries Earth 4k Ultra-High Definition (UHD) natural color images,
like India having only limited numbers of rain gauge sta- as presented in Fig. 11. It is clearly illustrated that Google
tions; moreover, small geographical areas such as water- Earth Images of erosion risk areas in the study area are well
sheds may not have a sufficient number of rain gauge sta- in agreement with the final output of the study. Present
tions. For this reason, many scholars forced to compute the days, the Google Earth images play a vital role in validat-
R factor from the least number of rainfall observation data ing the models as it offers true color, high resolution, real-
and it could offer less constituency in the R factor. To over- time images of the given area. Hence the researchers from
come this issue scholars have started using high-resolution various countries rely on Google Earth images to validate
consistent satellite-based rainfall data. Dutta et al. 2015 used their models. A number of researchers have validated their
TRMM rainfall data of 12 years (2001–2012) to generate R results from the RUSLE model with the help of Google
factor by using the formula developed by Singh et al. 1981 Earth images (Behera et al. 2020; Balasubramani et al. 2015;
for the Sanjal watershed of Jharkhand and have obtained Moses 2017; Khan and Govil 2020).
a better result. Similarly, Thapa 2020 has also deployed
TRMM rainfall data to compute the R factor for the Dolakha
district of Nepal. However, due to its coarse spatial resolu- Conclusion
tion (27 km), the TRMM satellite rainfall product is not pref-
erable for small watersheds; therefore, Das et al. 2020 used The RUSLE model has proved as an ideal tool along with
fine resolution (5 km) and more reliable CHIRPS gridded geospatial technology for the assessment of soil erosion risk
rainfall data to compute the R factor for the west Kameng and its mapping, even where field-based assessment is dif-
watershed, Arunachal Pradesh. Hence, the present study also ficult to conduct. And also, the study points out that Remote
adopted the CHIRPS rainfall data to compute the R factor Sensing datasets play a significant role in the quantification
of the study area mainly due to two reasons: (1) availability of soil erosion risk. The CHIRPS rainfall data are more pre-
of the limited number of rain gauge stations in the study cise for the computation of R factor than rain gauge data,
area and (2) the free availability of CHIRPS high-resolution since it gives continuous surface rainfall data rather than
daily and monthly wise rainfall data for more than 30 con- location-based measurement. In this study, RUSLE input
secutive years. parameters were generated from various parent datasets
To control the soil loss risk in the watershed, it is very and integrated into GIS platform to estimate soil loss rate
crucial to implement soil–water conservation practices on in Amravati watershed. The resultant raster was categorized

13
511 Page 16 of 20 Environmental Earth Sciences (2021) 80:511

Fig. 10  Suggested soil–water

conservation measures

into six classes based on its severity level viz, very low, low, ha−1/yr−1 followed by Degraded plantation (32.17 t/ha−1/
moderate, moderately high, high, and very high. The study yr−1), Scrubland/Wasteland (17.75 t/ha−1/yr−1), Current fal-
reveals that about 4%, 5.9%, and 3% of the watershed are low (12.08 t/ha−1/yr−1), and Rainfed cropland (10.03 t/ha−1/
susceptible to moderately high, high, and very high soil loss yr−1). The LU/LC and topography of the watershed have a
rates respectively. The maximum erosion rate was identi- great influence on soil erosion rates in the study area.
fied in the Degraded forest with average soil loss of 59.51 t/

13
Environmental Earth Sciences (2021) 80:511 Page 17 of 20 511

Fig. 11  Validation of final output with Google earth UHD natural colour images

References Gumara watershed using RUSLE and GIS techniques. Environ

Syst Res 8:1–17. [Link]
Bera A (2017) Assessment of soil loss by universal soil loss equation
Adongo TA, Agyare WA, Abagale FK, Buffour NK (2019) Spatial soil
(USLE) model using GIS techniques: a case study of Gumti River
loss estimation using an integrated GIS-based revised universal
Basin, Tripura, India. Model Earth Syst Environ 3:29. [Link]
soil loss equation. Int J Sci Tech 11(4):58–71. [Link]
org/​10.​1007/​s40808-​017-​0289-9
4314/​ijest.​v11i4.6
Beyene AA (2019) Soil erosion risk assessment in Nashe Dam reser-
Ali SA, Hagos H (2016) Estimation of soil erosion using USLE and
voir using remote sensing, GIS and RUSLE model techniques in
GIS in Awassa catchment, Rift valley, Central Ethiopia. Geoderma
Horro Guduru Wollega zone, Oromia region, Ethiopia. J Civil
Reg 7(2):159–166. [Link]
Constr Environ Eng 4(1):1–18. [Link]
Amin M, Romshoo SA (2018) Comparative assessment of soil ero-
20190​401.​11
sion modelling approaches in a Himalayan watershed. Model
Bou-imajjane L, Belfoul MA, Elkadiri R, Stokes M (2020) Soil ero-
Earth Syst Environ 5:175–192. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 007/​
sion assessment in a semi-arid environment: a case study from the
s40808-​018-​0526-x
Argana Corridor, Morocco. Environ Earth Sci 79:409. [Link] oi.​
Anees MT, Abdullah K, Nawawi MNM, Norulaini NAN, Syakir MI,
org/​10.​1007/​s12665-​020-​09127-8
Omar AKM (2018) Soil erosion analysis by RUSLE and sediment
Brema J, Hauzinger J (2016) Estimation of the soil erosion in Cauvery
yield models using remote sensing and GIS in Kelantan state,
watershed (Tamil Nadu and Karnataka) using USLE. J Environ
Peninsular Malaysia. Soil Res 56(4):356–372. [Link]
Sci Toxicol Food Tech 10:1–11. [Link]
1071/​SR171​93
10120​30111
Balasubramani K, Veena M, Kumaraswamy K, Saravanabavan V
Chadli K (2016) Estimation of soil loss using RUSLE model for Sebou
(2015) Estimation of soil erosion in a semi-arid watershed of
watershed (Morocco). Model Earth Syst Environ. [Link]
Tamil Nadu (India) using revised universal soil loss equation
10.​1007/​s40808-​016-​0105-y
(rusle) model through GIS. Model Earth Syst Environ 1:10.
Chandramohan T, Venkatesh B, Balchand AN (2015) International
[Link]
conference on water resources, coastal and ocean engineering
Behera M, Sena DR, Mandal U, Kashyap PS, Dash SS (2020) Inte-
(icwrcoe 2015), Evaluation of three soil erosion models for small
grated GIS-based RUSLE approach for quantification of potential
watersheds. Aquatic Procedia 4:1227–1234. [Link]
soil erosion under future climate change scenarios. Environ Monit
1016/j.​aqpro.​2015.​02.​156
Assess 192:733. [Link]
Chen H, Oguchi T, Wu P (2017) Assessment for soil loss by using a
Bekele M (2021) Geographic information system (GIS) based soil loss
scheme of alterative sub-models based on the RUSLE in a Karst
estimation using RUSLE model for soil and water conservation
Basin of Southwest China. J Integr Agric 16(2):377–388. [Link]
planning in anka_shashara watershed southern, Ethiopia. Int J
doi.​org/​10.​1016/​S2095-​3119(16)​61507-1
Hydro 5(1):9–21. [Link]
Cunha ERD, Bacani VM, Panachuki E (2017) Modeling soil erosion
Belayneh M, Yirgu T, Tsegaye D (2019) Potential soil erosion estima-
using RUSLE and GIS in a watershed occupied by rural settlement
tion and area prioritization for better conservation planning in

13
511 Page 18 of 20 Environmental Earth Sciences (2021) 80:511

in the Brazilian Cerrado. Nat Hazards 85:851–868. [Link] oi.o​ rg/​ universal soil loss equation model (RUSLE) and geo information
10.​1007/​s11069-​016-​2607-3 technology. Int Res J Earth Sci 5(8):8–14
Das B, Bordoloi R, Thungon LT, Paul A, Pandey PK, Mishra M, Kayet N, Pathak K, Chakrabarty A, Satiprasad S (2018) Evaluation of
Tripathi OPJ (2020) An integrated approach of GIS, RUSLE soil loss estimation using the RUSLE model and SCS-CN method
and AHP to model soil erosion in West Kameng watershed, in hillslope mining areas. Int Soil Water Conserv Res 6:31–42.
Arunachal Pradesh. Earth Syst Sci 129:94. [Link] [Link]
1007/​s12040-​020-​1356-6 Khan A, Govil H (2020) Evaluation of potential sites for soil erosion
Djoukbala O, Mazour M, Hasbaia M, Benselama O (2018) Estimating risk in and around Yamuna River flood plain using RUSLE. Arab
of water erosion in semiarid regions using RUSLE equation under J Geosci 13:707. [Link]
GIS environment. Environ Earth Sci 77:345. [Link] Koirala P, Thakuri S, Joshi S, Chauhan R (2019) Estimation of soil
1007/​s12665-​018-​7532-1 erosion in Nepal using a RUSLE modeling and geospatial tool.
Duarte L, Teodoro AC, Gonçalves JA, Soares D, Cunha M (2016) Geosci 9(4):147. [Link]
Assessing soil erosion risk using RUSLE through a GIS open- Kolli MK, Opp C, Groll M (2021) Estimation of soil erosion and
source desktop and web application. Environ Monit Assess sediment yield concentration across the Kolleru Lake catchment
188:351–366. [Link] using GIS. Environ Earth Sci 80:161. [Link]
Dutta D, Das S, Kundu A, Taj A (2015) Soil erosion risk assessment s12665-​021-​09443-7
in Sanjal watershed, Jharkhand (India) using geo-informatics, Lebourgeois V, Dupuy S, Vintrou É, Ameline M, Butler S, Bégué A
RUSLE model and TRMM data. Model Earth Syst Environ (2017) A combined random forest and obia classification scheme
1:37–45. [Link] for mapping smallholder agriculture at different nomenclature
Galdino S, Sano EE, Andrade RG, Grego CR, Nogueira SF, Bragantini levels using multisource data (simulated sentinel-2 time Series,
C, Flosi AHG (2015) Large-scale modeling of soil erosion with VHRS and DEM). Remote Sens 9(3):259. [Link] oi.o​ rg/1​ 0.3​ 390/​
RUSLE for conservationist planning of degraded cultivated Bra- rs903​0259
zilian Pastures. Land Degrad Develop 27(5):773–784. [Link] Lee E, Ahn S, Im S (2017) Estimation of soil erosion rate in the Demo-
org/​10.​1002/​ldr.​2414 cratic People’s Republic of Korea using the RUSLE model. For-
Ganasri BP, Ramesh H (2016) Assessment of soil erosion by RUSLE est Sci Technol 13(3):100–108. [Link] oi.o​ rg/1​ 0.1​ 080/2​ 15801​ 03.​
model using remote sensing and GIS—a case study of Nethravathi 2017.​13414​35
Basin. Geosci Front 7(6):953–961. [Link] Li Y, Qi S, Liang B, Ma J, Cheng B, Ma C, Qiu Y, Chen Q (2019)
2015.​10.​007 Dangerous degree forecast of soil loss on highway slopes in moun-
Gashaw T, Tulu T, Argaw M, Worqlul AW (2019) Modeling the tainous areas of the Yunnan-Guizhou Plateau (China) using the
impacts of land use–land cover changes on soil erosion and revised universal soil loss equation. Nat Hazards Earth Syst Sci
sediment yield in the Andassa watershed, upper Blue Nile basin, 19(4):757–774. [Link]
Ethiopia. Environ Earth Sci 78:679. [Link] Lin D, Gao Y, Wu Y, Shi P, Yang H, Wang J (2017) A conversion
s12665-​019-​8726-x method to determine the regional vegetation cover factor from
Gelagay HS, Minale AS (2016) Soil loss estimation using GIS and standard plots based on large sample theory and TM images: a
Remote sensing techniques: a case of Koga watershed, Northwest- case study in the eastern farming-pasture ecotone of northern
ern Ethiopia. Int Soil Water Conserv Res 4(2):126–136. [Link] China. Remote Sens 9:1035–1055. [Link]
doi.​org/​10.​1016/j.​iswcr.​2016.​01.​002 1035
Geleta CD, Deressa TA (2019) Evaluation of climate hazards group Mallick J, Alashker Y, Al-Deen MS, Ahmed M, Abul HM (2014) Risk
infrared precipitation station (CHIRPS) satellite-based rainfall assessment of soil erosion in semi-arid mountainous watershed in
estimates over Finchaa and Neshe watersheds. Ethiopia Eng Saudi Arabia by RUSLE model coupled with remote sensing and
Report. [Link] GIS. Geocarto Int 29(8):915–940. [Link]
Gianinetto M, Aiello M, Polinelli F, Frassy F, Rulli MC, Ravazzani 049.​2013.​868044
G, Bocchiola D, Chiarelli DD, Soncini A, Vezzoli R (2019) Markose VJ, Jayappa KS (2016) Soil loss estimation and prioritization
D-RUSLE: a dynamic model to estimate potential soil erosion of sub-watersheds of Kali River basin, Karnataka, India, using
with satellite time series in the Italian Alps. Eur J Remote Sens RUSLE and GIS. Environ Monit Assess 188:22. [Link]
52(4):34–53. [Link] 10.​1007/​s10661-​016-​5218-2
[Link] Maury S, Gholkar M, Jadhav A, Rane N (2019) Geophysical evaluation
[Link] of soils and soil loss estimation in a semiarid region of Maha-
Islam RM, Jaafar WZW, Hin LS, Osman N, Din MAM, Zuki FM, Sriv- rashtra using revised universal soil loss equation (RUSLE) and
astava P, Islam T, Adham MI (2018) Soil erosion assessment on GIS methods. Environ Earth Sci 78:144. [Link]
hillslope of GCE using RUSLE model. J Earth Syst Sci 127:1–16. s12665-​019-​8137-z
[Link] Miller RW, Donahue RL (1990) Soils: an introduction to soils and plant
Jayasekara MJPTM, Kadupitiya HK, Vitharana UWA (2018) Map- growth, 6th edn. Prentice Hall, Englewood Cliffs, New Jersey,
ping of soil erosion hazard zones of Sri Lanka. Trop Agric Res USA, p 768
29(2):135–146. [Link] Moore I, Burch F (1986) Physical basic of the length-slope factor in
Jazouli AE, Barakat A, Ghafiri A, Moutaki SE, Ettaqy A, Khellouk R the universal soil loss equation. Soil Sci Soc Am J 50:1294–1298
(2017) Soil erosion modeled with USLE, GIS, and remote sensing: Moses AN (2017) GIS-RUSLE interphase modelling of soil erosion
a case study of Ikkour watershed in Middle Atlas (Morocco). Geo- hazard and estimation of sediment yield for river Nzoia Basin
sci Lett 4(25):1–12. [Link] in Kenya. J Remote Sens GIS 6(3):205. [Link]
Jena RK, Padua S, Ray P, Ramachandran S, Bandyopadhyay S, Roy DP, 2469-​4134.​10002​05
Reddy GPO, Singh SK, Ray SK (2018) Assessment of soil ero- Mustefa M, Fufa F, Takala W (2019) GIS estimation of annual average
sion in sub-tropical ecosystem of Meghalaya, India using remote soil loss rate from Hangar River watershed using RUSLE. J Water
sensing, GIS and RUSLE. Indian J Soil Conserv 46(3):273–282 Clim Change. [Link]
Karthick P, Lakshumanan C, Ramki P (2017) Estimation of soil ero- National Bureau of Soil Survey and Land Use Planning (NBSS&LUP),
sion vulnerability in Perambalur Taluk, Tamil Nadu using revised Nagpur, India (2005) Annual Report.

13
Environmental Earth Sciences (2021) 80:511 Page 19 of 20 511

Panagos P, Borrelli P, Meusburger K (2015) A New European slope Samantha S, Koloa C, Kumar PD, Pals matas B (2016) Estimation
length and steepness factor (LS Factor) for modeling soil erosion of potential soil erosion rate using RUSLE and E ­ 30 model.
by water. Geosci 5:117–126. [Link] Model Earth Syst Environ 2(149):1–11. [Link]
s5020​117 s40808-​016-​0206-7
Panditharathne DLD, Abeysingha NS, Nirmanee KGS, Mallawatantri Sangeetha R, Ambujam NK (2021) The overshadow of the human
A (2019) Application of revised universal soil loss equation evolvement process in the dynamics of soil drift of an agricultural
(Rusle) model to assess soil erosion in “Kalu Ganga” River basin watershed in Nilgiri Hills, India. Soil Water Res. [Link]
in Sri Lanka. Appl Environ Soil Sci 1:1–15. [Link] 10.​17221/​105/​2020-​SWR
1155/​2019/​40373​79 Saravanan S, Jennifer JJ, Singh L, Thiyagarajan S, Sankaralingam S
Paul SS, Li J, Li Y, Lei S (2019) Assessing land use-land cover change (2021) Impact of land-use change on soil erosion in the Coonoor
and soil erosion potential using a combined approach through Watershed, Nilgiris Mountain Range, Tamil Nadu, India. Arab
remote sensing, RUSLE and random forest algorithm. Geocarto J Geosci 14:407. [Link]
Int. [Link] Sardari MRA, Bazrafshan O, Panagopoulos T, Sardooi ER (2019)
Pradhan RK, Srivastava PK, Maurya S, Singh SK, Patel DP (2018) Modeling the impact of climate change and land use change
Integrated framework for soil and water conservation in Kosi scenarios on soil erosion at the Minab Dam watershed. Sustain-
River Basin. Geocarto Int 35:391–410. [Link] ability 11:3353. [Link]
10106​049.​2018.​15209​21 Senanayake S, Pradhan B, Huete A, Brennan JA (2020) Review on
Prasannakumar R, Shiny N, Geetha H, Vijith H (2011) Spatial pre- assessing and mapping soil erosion hazard using geo-informat-
diction of soil erosion risk by remote sensing, GIS and RUSLE ics technology for farming system management. Remote Sens
approach: a case study of Siruvani river watershed in Attapady 12(24):4063. [Link]
valley, Kerala, India. Environ Earth Sci 64:965–972. [Link] Senthilkumar M, Gnanasundar D, Arumugam R (2019) Identify-
org/​10.​1007/​s12665-​011-​0913-3 ing groundwater recharge zones using remote sensing & GIS
Pritam KM, Reddy GPO, Hegde R (2018) Mapping and assessment techniques in Amaravathi aquifer system, Tamil Nadu, South
of soil erosion by RUSLE model using remote sensing and GIS India. Sustain Environ Res 29(15):1–9. [Link]
techniques—a case study of Palayakottai Watershed of Tirupur s42834-​019-​0014-7
district in Tamil Nadu. Bull Environ Sci Res 8(1):1–7 Shaikh S, Palanisamy M, Mohideen ARS (2020) Estimate the annual
Pushpalatha K, Kumar KS, Rao KV, Rejani R (2017) Spatial and soil loss in Kummattipatti Nadi watershed using rusle model
temporal variation Of C-factor and soil erosion in a semi-arid through geospatial technology. Geodesy Cartogr 46(2):75–82.
watershed: a case study in Mahabubnagar District. Int J Agric [Link]
Sci 7(5):175–188 Shit PK, Arup SN, Bhunia GS (2015) Soil erosion risk mapping
Qin W, Qiankun G, Wenhong C, Zhe Y, Qinghong Y, Zhijie S, Fenlic using RUSLE model on jhargram sub-division at West Bengal
Z (2018) A new RUSLE slope length factor and its application to in India. Model Earth Syst Environ 1(28):1–9. [Link]
soil erosion assessment in a Loess Plateau watershed. Soil till Res 10.​1007/​s40808-​015-​0032-3
182:10–24. [Link] Silva RM, Santos CAG, Maranhão KUA, Silva AM, Lima VRP
Ranzi R, Le TH, Rulli MC (2012) A RUSLE approach to model sus- (2018) Geospatial assessment of eco-environmental changes in
pended sediment load in the Lo river (Vietnam): effects of reser- desertification area of the Brazilian semi-arid region. Earth Sci
voirs and land use changes. J Hydrol 422–423:17–29. [Link] Res J 22(3):175–186. [Link]
org/​10.​1016/j.​jhydr​ol.​2011.​12.​009 Singh G, Babu R, Narain P, Bhushan LS, Abrol IP (1981) Soil loss
Reddy GPO, Kurothe RS, Sena DR, Harindranath CS, Niranjana KV, prediction research in India. Bull. T-12/D-9, Central Soil and
Naidu LGK, Singh SK, Sarkar D, Mishra PK, Sharda VN (2016) Water Conservation Research Training Institute, Dehradun
Assessment of soil erosion in tropical ecosystem of Goa, India Srinivasan R, Singh SK, Nayak DC, Hegde R, Ramesh M (2019)
using universal soil loss equation, geostatistics and GIS. Indian J Estimation of soil loss by USLE model using remote sensing
Soil Conserv 44(1):1–7 and GIS techniques—a case study of coastal Odisha, India.
Rejani R, Rao KV, Osman M, Srinivasa R, Reddy KS, Chary GR, Eurasian J Soil Sci 8(4):321–328. [Link] oi.​o rg/​1 0.​1 8393/​
Pushpanjali SJ (2016) Spatial and temporal estimation of soil ejss.​598120
loss for the sustainable management of a wet semi-arid watershed Steinmetz AA, Cassalho F, Caldeira TL, de Oliveira VA, Beskow S,
cluster. Environ Monit Assess 188:143. [Link] Timm LC (2018) Assessment of soil loss vulnerability in data-
s10661-​016-​5143-4 scarce watersheds in southern Brazil. Agric Sci 42(6):575–587.
Rellini C, Scopesia S, Olivarib M, Firpoa MM (2019) Assessment of [Link]
soil erosion risk in a typical Mediterranean environment using a Sulugodu B, Deka PC (2019) Evaluating the performance of
high-resolution RUSLE approach (portofino promontory, NW- CHIRPS satellite rainfall data for streamflow forecasting.
Italy). J Maps 15(2):356–362. [Link] Water Resour Manag 33:3913–3927. [Link]
2019.​15994​52 s11269-​019-​02340-6
Renard KG, Foster GR, Weesies GA, McCool DK, Yoder DC (1997) Tang Q, Xu Y, Sean J, Bennett LY (2015) Assessment of soil erosion
Predicting soil erosion by water: a guide to conservation planning using RUSLE and GIS: a case study of the Yangou watershed
with the revised universal soil loss equation (RUSLE). USDA in the Loess Plateau, China. Environ Earth Sci 73:1715–1724.
Agric Handb 703:1–251 [Link]
Ruiz GJM, Beguer S, Romero NE, Hidalgo GJC, Renault LN, Sanju Teng H, Rossel RV, Shi Z, Behrens T, Chappell A, Bui E (2016) Assim-
Y (2015) A meta-analysis of soil erosion rates across the world. ilating satellite imagery and visible near infrared spectroscopy to
Geomor 239:160–173. [Link] model and map soil loss by water erosion in Australia. Environ
03.​008 Modell Softw 77:156–167. [Link] oi.o​ rg/1​ 0.1​ 016/j.e​ nvsof​ t.2​ 015.​
Saha A, Ghosh P, Mitra B (2018) GIS based soil erosion estimation 11.​024
using Rusle model: a case study of upper Kangsabati watershed, Thapa P (2020) Spatial estimation of soil erosion using RUSLE mod-
West Bengal, India. J Environ Sci Nat Resour 13(5):119–126. eling: a case study of Dolakha district, Nepal. Environ Syst Res
[Link] 9:15. [Link]

13
511 Page 20 of 20 Environmental Earth Sciences (2021) 80:511

Thiaw I, Dacosta H (2017) Mapping of soil erosion risk in the Diarha Malaysia, Northern Borneo. Environ Dev Sustain 20:1365–1384.
watershed using Rusle, RS and GIS. Am J Remote Sens 5(4):30– [Link]
42. [Link] Wang X, Zhao XL, Zhang ZX, Yi L, Zuo LJ, Wen QK, Liu F, Xu JY,
Thirunavukkarasu P, Ambujam NK (2020) Zoning of groundwater Hu SG, Liu B (2016) Assessment of soil erosion change and its
potential in Amaravathy river basin, south India, by integrating relationships with land use/cover change in China from the end
remote sensing and GIS. Indian J Geo Mar Sci 49(10):1686–1692 of the 1980s to 2010. CATENA 137:256–268. [Link]
Thlakma SR, Iguisi EO, Odunze AC, Jeb DN (2018) Estimation of soil 1016/j.​catena.​2015.​10.​004
erosion risk in Mubi South watershed, Adamawa State, Nigeria. J Wischmeier WH, Smith DD (1978) Predicting rainfall erosion losses:
Remote Sens GIS 7(1):1–10. [Link] a guide to conservation planning, vol 537. USDA, Agriculture
10002​26 Handbook, Washington, DC
Thomas J, Joseph S, Thrivikramji KP (2018) Estimation of soil ero- Yuhendra YE (2019) Multi-temporal sentinel-2 images for classifica-
sion in a rain shadow river basin in the southern Western Ghats, tion accuracy. J Comput Sci 15(2):258–268. [Link]
India using RUSLE and transport limited sediment delivery func- 3844/​jcssp.​2019.​258.​268
tion. Int Soil Water Conserv Res 6(2):111–122. [Link] oi.o​ rg/1​ 0.​ Zerihun M, Mohammed S, Mohammedyasin SD, Adem AA, Lakew M
1016/j.​iswcr.​2017.​12.​001 (2018) Assessment of soil erosion using RUSLE, GIS and remote
Toubal AK, Achite M, Ouillon S, Dehni A (2018) Soil erodibility sensing in NW Ethiopia. Geoderma Reg 12:83–90. [Link] oi.o​ rg/​
mapping using the RUSLE model to prioritize erosion control in 10.​1016/j.​geodrs.​2018.​01.​002
the Wadi Sahouat basin, North-West of Algeria. Environ Monit Zhu AX, Wang P, Zhu TX, Chen LJ, Cai QG, Liu HP (2013) Modeling
Assess. [Link] runoff and soil erosion in the Three-Gorge Reservoir drainage area
USDA. Natl. Soil. Surv. Handbook (1983) Depart. Agr. 430 of China using limited plot data. J Hydrol 492:163–175. [Link]
USDA-SCS (U.S. Department of Agriculture-Soil Conservation doi.​org/​10.​1016/j.​jhydr​ol.​2013.​03.​038
Service) (1972) SCS national engineering handbook, section 4,
hydrology. Chapter 10, estimation of direct runoff from storm Publisher’s Note Springer Nature remains neutral with regard to
rainfall. U.S. Department of Agriculture, Soil Conservation Ser- jurisdictional claims in published maps and institutional affiliations.
vice, Washington, D.C, pp 10.1-10.24
Vijith H, Seling LW, Dodge WD (2017) Estimation of soil loss and
identification of erosion risk zones in a forested region in Sarawak,

13