Application of Revised Universal Soil Loss Equation (Rusle) Model to Assess Soil Erosion in “Kalu Ganga” River Basin in Sri Lanka

Soil erosion is one of the main forms of land degradation. Erosion contributes to loss of agricultural land productivity and ecological and esthetic values of natural environment, and it impairs the production of safe drinking water and hydroenergy production. *us, assessment of soil erosion and identifying the lands more prone to erosion are vital for erosion management process. Revised Universal Soil Loss Equation (Rusle) model supported by a GIS system was used to assess the spatial variability of erosion occurring atKalu Ganga river basin in Sri Lanka. Digital ElevationModel (30× 30m), twenty years’ rainfall data measured at 11 rain gauge stations across the basin, land use and soil maps, and published literature were used as inputs to the model. *e average annual soil loss in Kalu Ganga river basin varied from 0 to 134 t ha 1 year 1 and mean annual soil loss was estimated at 0.63 t ha 1 year . Based on erosion estimates, the basin landscape was divided into four different erosion severity classes: very low, low, moderate, and high. About 1.68% of the areas (4714 ha) in the river basin were identified with moderate to high erosion severity (>5 t ha 1 year ) class which urgently need measures to control soil erosion. Lands with moderate to high soil erosion classes were mostly found in Bulathsinghala,Kuruwita, and Rathnapura divisional secretarial divisions. Use of the erosion severity information coupled with basin wide individual RUSLE parameters can help to design the appropriate land use management practices and improved management based on the observations to minimize soil erosion in the basin.


Introduction
Soil erosion is a natural process of removal of soil material and transportation through the action of erosive agents such as water, wind, gravity, and human disturbance [1], and it has been accelerated by human activities such as intensive agriculture, improper land management, deforestation, and cultivation on steep slopes [2]. It is a serious and continuous environmental problem when combined with climate-induced high-intensity rainfall [3]. Degradation of agricultural land by soil erosion is a worldwide phenomenon leading to loss of nutrient rich surface soil, increased runoff from more impermeable subsoil, and decreased water availability to plant [4]. Worldwide, the average soil erosion rate in crop lands is about 30 t ha − 1 yr − 1 ranging from 0.5 to 400 t ha − 1 yr − 1 [5]. Estimates indicate that about 85% of land attenuation, globally, is due to soil erosion reducing crop productivity by about 17%, affecting the soil fertility initially and in the long term resulting land desertion [6].
ough soil erosion is a naturally occurring process, this has been accelerated by human activities such as intensive agriculture, improper land management, deforestation, and cultivation on steep slopes [2]. Removal of vegetation cover and shaping of surface topography induce or accelerate soil displacement and movement [7]. Serious soil erosion is occurring in most of the world's major agricultural regions [5] due to the expansion of agriculture without adequate soil conservation practices.
Several areas in Sri Lanka are subjected to severe soil erosion [2,8,9]. Soil loss in agricultural areas in steep central highlands of Sri Lanka is 10-100 times more compared to natural rates of erosion [8]. Soil erosion in intensively cultivated vegetable and tea lands with poor land management is higher than that in the forested landscapes or in well-managed tea and home gardens in Kurudu Oya subcatchment in upper Mahaweli area [9].
Moreover, crop residues are removed for fodder, biofuel, and industrial uses leaving the soil surfaces bared from a protective cover enhancing the vulnerability to lands to erosion. e resulting runo ultimately transports sediments, organic material, nutrients, and pesticide residues o -site impacting both water and soil quality. When lands are left as fallow to recover, the erosion problem is worsened due to minimal vegetative cover [10]. Soil erosion is reported to increase with higher magnitude of rainfall and frequent occurrences of heavy precipitation [11].
Soil erosion and degradation of lands in Sri Lanka are a ecting the national food production and sustainability of natural eco systems [2]. Erosion rates are high in highlands as well as in low lands due to land use changes including the removal of vegetative cover and urbanization [7]. Several studies conducted in Sri Lanka to assess soil erosion, based on both numerical modeling and actual quanti cation, have shown that soil erosion is a severe problem in Sri Lanka [12][13][14][15]. erefore, it is useful to identify the areas more prone to soil erosion, so that farmers and land managers can incorporate appropriate soil conservation measures to minimize extensive soil erosion. ere are several erosion prediction approaches widely used [16] comprised of empirical, conceptual, and physical based models [6,17].
e Universal Soil Loss Equation (USLE), Modi ed Universal Soil Loss Equation (MUSLE), and Revised Universal Soil Loss Equation (RUSLE) are the most popular empirical models used globally for erosion prediction and control [14,18]. RUSLE developed by the U.S. Department of Agriculture is used as a decision support system in soil conservation and land use planning [19]. It uses a set of mathematical equations to describe ecological processes related to conservation practices and erosion in a given landscape [20].
RUSLE is a exible tool that has been adapted to landscape and watershed scales combined with Geographic Information Systems (GIS) [21][22][23][24][25][26] in soil erosion assessments. e study of annual soil loss using GIS-based RUSLE to the Pamba watershed in a mountain landscape demonstrated the applicability of RUSLE in investigating erosion hazard where soil erosion rate is minimum in natural forest areas and maximum in places with human in uence [22]. In Sri Lanka, a few studies have been conducted to assess soil erosion using the RUSLE model [14,15].
Kalu Ganga is the 4 th longest river (129 km) in Sri Lanka and entirely located in the wet zone of the country [27]. As a result, it accounts for the largest amount of kilometers  discharge to the sea in the country [28]. ere are drinking water scheme and mini and medium scale hydropower generation stations associated with the river [29]. In addition, the main land use of the basin is agriculture. Kalu Ganga basin is in the area where the climate-induced rainfall expected to increase with an elevated potential for oods and landslides [30]. us, it is worth analyzing the spatial variation of soil loss in the basin so that planner can take precautions to minimize soil erosion. e present study estimates the average annual soil loss in the Kalu Ganga river basin using the Revised Universal Soil Loss Equation (RUSLE) in combination with an ArcGIS interface to develop the soil erosion hazard map of the Kalu Ganga river basin.

Study Area.
e Kalu Ganga river basin (2,766 km 2 ) is located in the Southwestern part of Sri Lanka.
e river length is about 130 km long and extends from 80.00°to 80.67°E and 6.42°to 6.83°N. e Kalu Ganga starts in the central hills of the country at an altitude of 2,250 m and runs into the Indian Ocean near the town Kalutara after travelling through one of the highest rainfall areas of the country. Upstream area of the basin is higher in gradient while downstream area of the basin is more or less at and elevation varied from 2149 m to 0 m [28] (Figure 1). e annual rainfall average is about 4,000 mm and the annual water ow is about 4,000 million m 3 [28]. Average  Applied and Environmental Soil Science 3 annual rainfall variation of the basin is shown in Figure 2. e maximum rainfall value of 4,466 mm was recorded in Kudawa and the lowest average annual rainfall value of 2,613 mm was recorded in Kaluthara rain gauge station. Compared to average annual rainfall of Sri Lanka, this river basin receives higher rainfall [27].   small holder tea. e forest cover of the basin is more than 30% with a 77% canopy cover. e soil map of the basin ( Figure 4) indicates that Red Yellow Podzolic is the main soil type found in the basin and Alluvial soil is found in the ood areas of the river ( Figure 4). Bog and Half Bog and Lithosols are the other soil types found in the basin but to a lesser extent.

RUSLE Parameter Estimation.
e RUSLE was used as the model and it was interfering with the ArcGIS 10.2.1. RUSLE [19] can be expressed as where A average annual soil loss per unit area (t ha − 1 yr − 1 ), R rainfall-runo erosivity factor (MJ mm ha − 1 h − 1 yr − 1 ), K soil erodibility factor (t ha h MJ − 1 mm − 1 ), L slope length factor, S slope steepness factor, C cover and management factor, and P support and conservation practices factor.
Major soil type      Figure 6: Map of estimated R factor. 6 Applied and Environmental Soil Science

Rainfall-Runo Erosivity Factor (R).
e erosive power of rainfall can be estimated by calculating the erosivity factor for a particular location [31]. It depends on the amount and the intensity of rainfall [2]. e original equation proposed by Wischmeier and Smith [32] and Renard et al. [33] was not used due to nonavailability of rainfall intensities. As a substitute, regression equations are developed for di erent regions to calculate the R values. We used the rainfall-erosivity relationship developed by Premalal [34] as in equation (2). e same equation has been used in Kirindi Oya [14], in Kelani river basins [15], and for mapping soil erosion hazard in Sri Lanka [2]. In our study, rainfall-runo erosivity factor was determined from the average annual rainfall from 1996 to 2016 along with the rainfall erosivity factor equation developed for Sri Lanka conditions by [34] (equation (2)). us, this study concerns the average soil erosion assessment for the period from 1996 to 2016.
where R rainfall-runo erosivity factor (MJ mm ha − 1 h − 1 yr − 1 ) and F average annual rainfall (mm). e rain gauge stations shape le was created by using latitude and longitude values with ArcGIS 10.2.1.
Among the available interpolation techniques, two types of interpolation techniques were used to select the suitable technique to create rainfall and R factor maps. One method was simple Kriging in geostatistical wizard with spherical semivariogram model. e other one was IDW method, with power 2. According to the cross validation results, IDW provides the least error (RMSE 163.39) for mapping rainfall data over Kriging method (RMSE 190.25). erefore, both annual rainfall variability map and rainfall erosivity factor map for Kalu Ganga river basin were generated using IDW interpolation.   Applied and Environmental Soil Science the susceptibility of soil towards erosion and measures the contribution of soil types [24]. K factor raster map was generated from vector soil map with 30 m resolution using the Feature to Raster tool in ArcGIS (Table 1).

Slope Length and Steepness Factor (LS).
Slope length and steepness has the greatest in uence on soil loss and describes the e ect of topography on soil erosion. S factor measures the e ect of slope steepness and was calculated by equations (3) and (4) using raster calculator based on the relationship given by [33] S 10.8 sin θ + 0.03 for slope percent < 9%, S 16.8 sin θ − 0.50 for slope percent ≥ 9%, where S slope steepness factor and θ slope angle in degree. L factor was calculated by equation (5) using raster calculator based on equation given by [32] where L slope length factor, λ horizontal projected slope length (m) (λ flow accumulation * cell size), and m slope length exponent. In this equation, "m" is the slope length exponent that varies based on slope steepness. Slope length exponent equals 0.5 if the slope is 4.5% or more, 0.4 on slopes between 3% and 4.5%, 0.3 on slopes between 1% and 3%, and 0.2 for atter terrains with gradient less than 1% [25]. L and S factors were calculated by using ow accumulation and slope in degree as inputs and nally LS factor map was generated by the multiplication of both L and S factors in raster calculator in ArcGIS.

Cover and Management Factor (C).
e land use and land cover map of the basin was extracted from the land use map produced by the survey department in 2016, and this study assumed that the land use and land cover was not changed signi cantly during the past 20 years. C factor values ( Figure 5) for di erent land uses were obtained from available literature [14,15,35,36] and matched with the land use map. Based on these C factor values, C factor raster

Support and Conservation Practices Factor (P)
. P value varies from 0 for good conservation practices to 1 for poor conservation practices. e P factor map was created based on the land use map of the Kalu Ganga river basin. P values ( Figure 5) were added to the attribute table of land use map and P factor raster map was created with the conversion tool in ArcGIS 10.2.1.

Creation of Soil Erosion Severity
Map. Soil erosion severity map was created using a raster overlay analysis with rainfall-runo erosivity factor, soil erodibility factor, slope length and steepness factor, cover and management factor, and support and conservation practices factor.

Sediment Delivery Ratio (SDR) and Sediment Yield (SY) of the Main Stream. Sedimentary Delivery Ratio (SDR)
is a fraction of gross erosion that is transported from a given area at a given time interval. It indicates the capability of a catchment for storing and transporting the eroded soil [37]. According to [38], the average slope of the stream channel is more signi cant than other parameters in estimating sediment delivery ratio, and it is a function of slope of main stream channel (equation (6)). e papers [39,40] also used this equation to estimate the sediment delivery ratios in di erent river basins.
where SCS is main stream channel slope measured in percent unit. After determining the ow direction, ow accumulation, and stream network using DEM, main stream channel slope was computed using ArcGIS HEC-GeoHMS extension. en using those SCS values, SDR for the main river of the Kalu Ganga river basin was calculated in raster calculator. Finally, sediment yield was calculated by overlaying the mean annual soil loss raster layer and sediment delivery ratio of the main stream using raster calculator.

Results and Discussion
e soil erosion hazard map was created by using all the ve RUSLE layers, rainfall-runo erosivity factor, soil erodibility factor, slope length and steepness factor, cover and management factor, and support and conservation practices factor. ese factors were estimated on a 30 × 30 m grid.  Applied and Environmental Soil Science 9 the mean R factor recorded is 387.39 MJ mm ha − 1 h − 1 yr − 1 . R factor map developed using Inverse Distance Interpolation (IDW) with a power of 2 is shown in Figure 6. (Figure 7) was evaluated based on di erent soil types ( Figure 4) and their properties found in the Kalu Ganga river basin. e results showed that more than 92% of Kalu Ganga river basin is covered by Red Yellow Podzolic (RYP). Alluvial soil covered more than 4% of Kalu Ganga river basin. K factor of the basin varied from 0.05 to 0.31 t h MJ − 1 mm − 1 (Table 2) with a mean of 0.21 t h MJ − 1 mm − 1 . Alluvial soil has the highest K value and Bog and Half Bog soil has the lowest K value in the basin (Figure 7). K value of RYP (0.22) is less susceptible to soil erosion than Reddish Brown Earth (RBE) (0.27) which is the main type of soil in dry zone, Sri Lanka.

Slope Length and Steepness Factor Map (LS).
e average slope of the Kalu Ganga river basin was 12.06°and it ranged from 0 to 90° (Figure 8). LS factor was calculated by using ow accumulation and slope in degree as inputs. LS factor value ranged from 0 to 56.77 ( Figure 9) and the mean LS factor value was 0.05. When slope and ow accumulation increase, LS factor increases. Higher LS factor was observed in the Northeastern upstream area of the basin.

Cover and Management Factor (C) and Support and Conservation Practices Factor (P).
Based on the di erent land use classes (Figure 3), the C factor map ( Figure 10) was generated. e mean C factor value was 0.28 for the Kalu Ganga river basin. C factor for well-protected land is 0 and for bare land it is 1 [4]. Summarized major land use type of the basin is shown in Table 3. High vegetative cover characterizes the low soil erosion potentials because of its ability to resist high-intensity rains expected as a result of climate change [41].
P factor value indicates the extent of erosion control practices in the Kalu Ganga river basin. Spatial distribution of P factor is shown in Figure 11 and mean P factor value for Kalu Ganga river basin is 0.28 ( Figure 11).

Soil Erosion Estimation.
e soil erosion severity map ( Figure 12) was generated by overlaying all the parameter layers of RUSLE. Results indicated that Kalu Ganga river basin has a mean average annual soil loss of about 0.63 t ha − 1 yr − 1 with a range of 0 to 134 t ha − 1 yr − 1 . Based on soil erosion classi cation used by [42], the estimated soil erosion was classi ed into three classes, namely, low (<5 t ha − 1 ), moderate (5-12 t ha − 1 ), and high (>12 t ha − 1 ) per annum. Due to the dominance of low erosion areas, the severity map was further classi ed into two subclasses as very low (<2 t ha − 1 ) and low areas (2-5 t ha − 1 ), respectively (Table 4).
A large portion of Kalu Ganga river basin is classi ed as very low erosion and it is more than 93% of the total area of the basin. Low to very low erosion were prominent in Kalu Ganga river basin compared to Kelani Ganga river [15] and Kiridi Oya river basins in Sri Lanka [14].
Comparatively recorded lower erosion rates may be due to the higher % of forest lands in Kalu Ganga (31%) river basin compared to Kelani (10%) and Kirindi Oya (27%) river basins. In addition, Red Yellow Podzolic soils are the main soil type in the Kalu Ganga river basin with a lower K value with a lower contribution to soil erosion. High erosion areas in Kirindi Oya river basin were occupied by home gardens and scrub lands [14] while high eroded areas in Kelani river basin were bare lands and lands under coconut with relatively high P factor and C factor values [15].
Moreover, the mean R factor recorded (387.39 MJ mmha − 1 h − 1 yr − 1 ) is lower than the R factor estimated for high rainfall areas elsewhere in the world [43]. It is also observed that R values used in other studies that used the same regression equation are also low [14,15]. is equation which was used in [34] has been developed for the entire country, where 1/3 of the country is in the dry zone. However, Kalu Ganga river basin is located in wet  erefore, there may be underestimation of the rainfall erosivity factor. erefore, we suggest to improve the regression equation for the di erent climatic zone of the country using new rainfall data and validate the proposed equations with the eld observations. e Kalu Ganga basin is characterized as having a potential low susceptibility to soil erosion due to the good cover, lower K values of the existing soils, etc. Based on the RUSLE model estimates, about 1.9% of the area in the basin could be categorized into moderate to high soil erosion hazard class. Lands located in Kuruwita, Bulathsinghala, and some parts of Rathnapura divisional secretarial divisions fall into these moderate to high erosion category. To control the soil erosion, it is suggested to use agronomic soil conservation measures especially to crop lands. However, steep slope areas may need comparatively low-cost mechanical structures such as stone terraces, lock and spill drains, and drop structures to minimize the erosion.

Sediment Yield of the Basin Estimated over the Main
Stream. Sediment yield (Figure 13) of the basin was determined by sediment delivery ratio (Figure 14) of the main stream. e sediment delivery ratio results ranged from 0 to 0.16 for the main stream of Kalu Ganga river. e mean sediment yield for the main stream of the Kalu Ganga river basin was 0.29 t ha − 1 yr − 1 and sediment yield ranged from 0 to 4.66 t ha − 1 yr − 1 . A similar study is reported in central highland of Blue Nile basin, Ethiopia [40]. According to their analysis, sediment delivery ratio of the central highland of Blue Nile basin watershed ranges between 0 and 0.26.    Figure 14: Sediment delivery ratio over the main stream of Kalu Ganga.

Conclusion
e soil erosion severity map at 30 × 30 m resolution for Kalu Ganga river basin based on the RUSLE model integrated with a GIS provided the spatial variation of soil erosion in the Kalu Ganga river basin. e entire basin has a low soil erosion severity and the mean annual soil erosion rate of the basin is low which is 0.63 t ha − 1 yr − 1 . Area under moderate to high risk with 5 to over 12 t ha − 1 yr − 1 erosion severity is about 1.89% of the entire basin. is area is located in Bulathsinghala, Kuruwita, and Rathnapura administrative divisions. e mean sediment yield value for the Kalu Ganga main stream was 0.29 t ha − 1 per annum. e study highlighted a potential underestimation of the R factor and recommended revisiting the regression equations used to compute rainfall erosivity, presently, generalized to the country, based on a climate zone approach, and also considering the climate predictions for rainfall under different scenarios.
Steep topographic features, poor vegetation cover and poor land management practices, and high rainfall rate may contribute to the high erosion and sediment delivery in the areas where there is high rate of erosion (1.89% of the basin).
is identification of the geographic areas critical for effective watershed management helps to target resources to conserve the soil quality and fertility and to meet the increasing erosion potential due to increasing rainfall intensity due to climate change. e best land management practices contributed to the less erosion potential such as enhanced canopy cover observed can be upscaled to other erosive areas of the country after conducting similar erosion analysis using RUSLE.

Data Availability
Rainfall data used are available on the Department of Meteorology, Colombo, and Natural Resources Management Center, Peradeniya, data portals.

Conflicts of Interest
ere are no conflicts of interest among authors.