3.1. Models of Soil Erosion Estimation
Soil erosion estimation models can be categorized into three commonly used categories: conceptual, physically-based, empirical, and empirical models. These categories are based on approaches to data generation, data dependence, and the physical processes simulated by the model (Hajigholizadeh 2018; Luvai, Obiero, and Omuto 2022; Tsegaye and Bharti 2021).
Physically Based Models
Simulate soil erosion processes based on the fundamental laws of physics and mechanics. They consider factors such as topography, rainfall intensity, vegetation cover, soil characteristics, and other relevant parameters. Physically based models typically use mathematical equations to represent erosion mechanisms, such as rainfall-runoff processes, sediment transport, and detachment of soil particles. They require detailed input data and are often used for more complex and precise erosion predictions. Examples of physically based models include the SWAT (Soil and Water Assessment Tool), KINEROS (Kinematic Runoff and Erosion Model), WEEP (Watershed Erosion Prediction Project) (Luvai et al., 2022; Regasa et al., 2021).
Empirical Models are based on observed associations between erosion and easily measurable input variables. These models derive their equations or relationships from field measurements and empirical data rather than explicitly considering the underlying physical processes. Empirical models are easier to use, require less input data, and are suitable when detailed information on erosion processes is unavailable (Hajigholizadeh 2018). However, they may have limited transferability to different locations or conditions. Empirical models include the RUSLE, the Universal Soil Loss Equation (USLE), and the Modified Universal Soil Loss Equation (MUSLE). The USLE model is frequently mentioned in the literature as one of the empirical models. Despite being primarily developed using the spatial data collected in the United States, this model, along with its revised and modified versions, has gained widespread global application. Numerous subsequent models have been developed based on the foundation provided by the USLE model (Luvai et al., 2022; Regasa et al., 2021).
Conceptual Models
Conceptual models represent a simplified representation of the soil erosion process using conceptual components or modules (Hajigholizadeh 2018). They often rely on expert knowledge and qualitative descriptions of erosion processes rather than accurate mathematical formulations. Conceptual models are helpful in understanding erosion processes and detecting the main factors controlling erosion. They are typically less data-intensive but may want the detailed predictive capabilities of physically based or empirical models. Conceptual erosion models include the MMF (Morgan-Morgan-Finney model) and the European Soil Erosion Model (EUROSEM), WATEM (Water and Tillage Erosion Model) (Luvai et al., 2022; Regasa et al., 2021).
3.2. Historical background of RUSLE in Ethiopia
The historical background of the RUSLE model in Ethiopia can be traced back to the 1970s and 1980s when soil erosion emerged as a significant ecological problem in the country (Hurni 1985). The Ethiopian government and international organizations and researchers recognized the need for practical soil erosion assessment and management strategies. During this period, efforts were made to adapt and apply erosion models, including RUSLE, to Ethiopian conditions. The RUSLE model, originally developed in the United States by Wischmeier and Smith in the late 1960s, gained attention for its simplicity and practicality in estimating soil loss (Monjezi et al. 2017). In the 1980s, the Ethiopian Agricultural Research Organization (EARO) conducted research and studies to evaluate the applicability of RUSLE in Ethiopian contexts. Hurni (1985) conducted an inspiring study on soil erosion in Ethiopia. The study utilized the RUSLE and field measurements to estimate soil loss rates. The findings highlighted the alarming rates of soil erosion in Ethiopia, particularly in highland areas.
They emphasized the importance of soil conservation measures to mitigate the adverse impacts on agriculture, water resources, and the environment. The study provided valuable insights into the dynamics of soil erosion in Ethiopia, contributing to subsequent research and developing erosion control strategies in the country.
According to different studies (Abebaw et al., 2022; Almaw et al., 2021; Anteneh, 2021; Negese et al., 2021; Sinshaw et al., 2021; Tadesse et al., 2017; Yeneneh et al., 2022; Yirgu, 2022) collected data on soil erodibility, rainfall erosivity, land cover, slope characteristics, and management practices across different regions of Ethiopia and tried to estimate the annual soil loss rate. RUSLE provided a helpful framework for quantifying soil loss and prioritizing areas requiring intervention (Gashaw et al. 2020; Kanito et al. 2023; Masha, Yirgu, and Debele 2021; Tesfaye et al. 2019; Yiferu, Girma Taddese, and Mebrate 2018). Over the years, researchers and institutions in Ethiopia continued to refine and customize the RUSLE model to suit local conditions. Localized studies were conducted to calibrate the model parameters based on field observations, soil analysis, rainfall data, and erosion monitoring (Girmay et al. 2020; Tsegaye, Kendie, and Esa 2020). The application of RUSLE in Ethiopia has played a vital role in generating awareness about soil erosion, guiding land management decisions, and implementing soil erosion control measures (Woldesenbet, Wudmatas, and Denboba 2020; Yeneneh et al. 2022; Yirgu 2022). It has contributed to developing soil and water protection strategies, such as terracing, agroforestry, and improved farming practices, to moderate the impact of erosion and preserve soil fertility(Masha et al. 2021; Melese et al. 2021). Today, RUSLE and its modified versions are commonly used by researchers, government agencies, and NGOs in Ethiopia for soil erosion assessment, land use planning, and the development of sustainable land management practices. The model continues to evolve with ongoing research and advancements in erosion modeling techniques.
3.3. RUSLE model parameters used in Ethiopia
RUSLE is considered an empirical model (Luvai et al. 2022). It estimates soil loss based on observed relationships between soil erosion and various easily measurable factors without explicitly incorporating the underlying physical processes of erosion (Regasa et al. 2021). RUSLE uses a mathematical equation that considers slope length, slope steepness, rainfall erosivity, soil erodibility, cover management practices, and erosion control practices (Hurni 1985). These factors are determined based on field and empirical data collected from various locations. While RUSLE incorporates some physical aspects, such as slope characteristics and land management practices, its primary approach is empirical. It calculates soil loss based on observed correlations between erosion and the input parameters. This empirical nature allows RUSLE to provide relatively simple and practical estimates of soil erosion that can be applied to different locations and conditions (Hurni 1985; Luvai et al. 2022; Regasa et al. 2021).
In Ethiopia’s context, the RUSLE model utilizes specific parameters relevant to the country's soil and climatic conditions (Almaw et al. 2021). The annual average soil loss is calculated by multiplying the following five parameters (Eq. 1):
“A = R * K* LS* C* P --------------------------------- (1)
where, A stands for annual soil loss (t ha − 1 year − 1);
R stands for rainfall erosivity factor (MJmmha − 1h − 1 year − 1);
K stands for soil erodibility factor (t ha − 1MJ − 1mm − 1);
L stands for slope length and S is slope steepness factor(dimensionless);
C stands for cover and management factor(dimensionless) and P is support and conservation practice factor(dimensionless) (Hurni 1985)”.
The above parameters are reviewed in detail to show how researchers used them in each context:
3.3.1. Rainfall Erosivity (R-factor):
It is a precondition for soil erosion (Luvai et al. 2022). The R-factor represents the erosive power of rainfall and is typically derived from long-term rainfall data. The R-factor is quantified by multiplying the total energy of a storm by the maximum 30-minute measurement of rainfall intensity recorded using autographic recorders. Kinetic energy is employed to represent the raindrop's capacity to detach soil particles from the soil mass as a whole. Because of the limited availability of rainfall intensity data in Ethiopia, numerous studies have relied on empirical equations that estimate R-values using easily accessible rainfall totals. This approach allows researchers to estimate the erosivity factor (R-factor) in the absence of rainfall intensity data. Forty studies among the reviewed articles employed a simple linear regression formula to estimate the rainfall erosivity factor (R) based on mean annual rainfall (P) in millimeters (mm). The formula, R = -8.12 + (0.562 × P), allows for a straightforward calculation of the erosivity factor using readily available mean annual rainfall data. This linear regression model assumes a linear relationship between mean annual rainfall and erosivity, where the intercept (-8.12) represents the base erosivity value, and the coefficient (0.562) indicates the rate at which the erosivity factor increases with each unit increase in mean annual rainfall. One of the articles reviewed used the following erosivity formula R = 1.24 *P 1.36, where R = Rainfall erosivity factor (MJmm ha − 1 h − 1 yr − 1) P Annual mean precipitation (mm) (Wagari and Tamiru 2021). The other study considered the formula R = 0.55 ∗ P − 4.7 (Balabathina et al. 2020). In a study conducted at the national level, the rainfall erosivity formula is not mentioned (Almaw et al. 2021).
In Ethiopia, rainfall erosivity can vary across different regions due to variations in rainfall intensity and distribution patterns. A review of research conducted by Ethiopian researchers on the calculation of R-factor erosivity in the RUSLE model reveals varying durations of rainfall data used. Among the studies reviewed, 6 studies utilized 10 years or less of rainfall data, six researchers incorporated rainfall data spanning between 11 and 20 years, eight researchers considered data covering a duration of 21 to 30 years, while nine researchers utilized rainfall data spanning over 31 years. However, the remaining 13 studies did not explicitly mention the number of years considered (Table 1). The impact of the duration of years on soil erosion estimation is an important aspect to consider. Using shorter durations of rainfall data, such as ten years or less, may not adequately capture the full range of erosive events, potentially leading to underestimation or overestimation of erosivity.
On the other hand, studies utilizing longer durations, such as 21 to 30 years or beyond, are more likely to provide a comprehensive understanding of erosivity patterns over time. By considering a broader range of rainfall intensities and durations, these studies are better equipped to estimate erosivity more accurately. However, it is essential to note that the availability of reliable long-term records specific to the study area should guide the selection of the duration of rainfall data.
Furthermore, the specific impact of using different durations of rainfall data may vary depending on the region's climatic characteristics and the specific equations or coefficients employed within the RUSLE model. The RUSLE model recommends using long-term rainfall data of 10–30 years to estimate annual soil loss, accounting for inter-annual variations, seasonal patterns, and extreme events, and establishing a more robust estimate of rainfall erosivity(Kimberlin and Moldenhauer 1977). Therefore, it is advisable to consult regional guidelines, existing studies, and experts to determine the most appropriate duration of rainfall data when estimating the R-factor and assessing soil erosion.
Table 1
Number of Years considered for rainfall erosivity (In the reviewed papers)
| Number of Years | Number of Studies |
| Unknown | 13 |
| 1 to 10 | 6 |
| 11 to 20 | 6 |
| 21 to 30 | 8 |
| 31 to 40 | 9 |
| Total | 42 |
3.3.2. Soil Erodibility (K-factor)
It signifies the erodibility of soil as affected by soil properties (Jahun et al. 2015; Luvai et al. 2022) and is determined based on different soil properties such as organic matter content, texture, permeability, and structure (Almaw et al. 2021; Endalamaw et al. 2021; Girma and Gebre 2020; Negese et al. 2021). Soil erodibility can vary due to variations in soil types across different regions (Jothimani, Getahun, and Abebe 2022).
Research conducted by various researchers in Ethiopia on calculating the erodibility factor (K) in the RUSLE model reveals the significance of considering soil color, texture, organic matter content, soil type, and soil structure (Table 2,3,6). Among the reviewed studies, 14 articles focused on soil texture, organic matter, and soil structure as key factors influencing erodibility, whereas the rest 28 studies used soil type and color (Table 6).
Two (2) studies among the reviewed articles tried to collect soil data from the field and conducted laboratory analysis to determine soil erodibility based on soil texture, encompassing the relative proportions of clay, silt, and sand particles, influences the soil's ability to retain water and withstand erosive forces (Desalegn et al., 2023; Girmay et al., 2020). By incorporating the soil parameters, researchers gain insights into the soil's physical characteristics and stability, which are vital in determining its susceptibility to erosion.
The soil organic matter content is crucial in enhancing its structure, stability, and water infiltration capacity, thereby mitigating erosion risks. Moreover, soil structure, including factors such as compaction and aggregation, contributes to the overall resistance of the soil against erosion processes (Endalamaw et al. 2021; Gashaw, Tulu, and Argaw 2017; Haregeweyn et al. 2017; Luvai et al. 2022; Moisa 2022)
Table 2
K-values of different Soil types
| Soil type | Soil color | K factor | References |
| Regoslos (Eutric) | Brown | 0.2 | (Negese et al. 2021), |
| Cambisols (Eutric) | Brown | 0.2 |
| Lithosols | Brown | 0.2 |
| Cambisols (Vertic) | Black | 0.15 |
| Cambisols (Eutric) | Brown | 0.2 | (Yeneneh et al. 2022) |
| Leptosols (Eutric) | Brown | 0.2 | |
| Vertisols (Eutric) | Black | 0.15 | |
| Nitosols (Haplic) | Red | 0.25 | |
| Luvisols (Haplic) | Brown | 0.2 | |
| Leptosols (Rendzic) | Brown | 0.2 | |
| Vertisol | Black | 0.15 | (Woldesenbet et al. 2020) |
| Luvisol | Brown Red | 0.2 |
| Cambisol | Red | 0.25 |
| Fluvisol | Brown | 0.2 |
| Leptosol | Yellow | 0.3 |
| Water | - | 0.4 |
| Andosols | * | 0.2 | (Degife, Worku, and Gizaw 2021) |
| Cambisols | * | 0.13 |
| Luvisols | * | 0.11 |
| Leptosols | * | 0.22 |
| Pellic vertisols | * | 0.2 | (Bekele 2021) |
| Chromic luvisols | * | 0.15 |
| Eutric nitisols | * | 0.15 |
| Chromic vertisols | * | 0.2 |
| Calcic cambisols | * | 0.05 | (Moisa 2022) |
| Calcic xerosols | * | 0.2 |
| Dystric gleysols | * | 0.35 |
| Chromic cambisols | * | 0.28 |
| Dystric nitisols | * | 0.25 |
| Eutric Cambisols | * | 0.34 |
| Eutric nitisols | * | 0.25 |
| Haplic aerosols | * | 0.2 |
| Leptosols | * | 0.3 |
| Orthic acrisols | * | 0.22 |
| Orthic luvisols | * | 0.2 |
| Orthic solonchaks | * | 0.15 |
| Phaeozems | * | 0.20 |
| Vertic Cambisols | * | 0.24 |
| Haplic Alisols | Brown | 0.20 | (Sinshaw et al. 2021) |
| Nitisols(haplic), Cambisols (Eutric), | Red | 0.25 |
| Luvisols(Haplic),Luvisols(Chromic) | Yellow | 0.30 |
| haplic acrisols, Eutric leptosols | Gray | 0.35 |
| Fluvisols(Eutric), Vertisols(Eutric) | Black | 0.15 | |
Table 3
Soil type and texture-based Soil erodibility (K factor) value
| Soil classes | Soil texture | Soil color | Erodibility (k-factor) | References |
| Leptosols (Eutric) | Silt clay loam | Grey to yellow | 0.35 | (Endalamaw et al. 2021) |
| Luvisols (Haplic) | Coarse soils | Yellow | 0.30 |
| Alisols (Haplic) | Friable clay loam to clay | Brown | 0.20 |
| Vertisols (Eutric) | Heavy clay | Black | 0.15 |
| Acrisols (Haplic) | Rocky to sandy soil | Grey | 0.35 |
| Fluvisols (Eutric) | Silt clay to clay | Black silt clay to clay | 0.15 |
| Nitosols (Haplic) | Sandy Clay Loam | Dark Red | 0.25 |
| Lithic Leptosols | Clay Loam | White | 0.49 |
| Eutric Leptosols | Loam | * | 0.21 | (Abebaw et al. 2022) |
| Chromic Luvisols | Coarse Sandy Clay Loam | * | 0.40 |
| Humic Nitisols | Clay | * | 0.10 |
| Eutric Vertisols | Light Clay | * | 0.14 |
| Haplic Luvisols | Sandy Clay Loam | * | 0.26 | |
3.3.3. LS-factor (Slope Length and Steepness):
The rise in L and S causes an increase in the LS factor and soil erosion, for which numerous methodologies of estimating the LS factor have been developed (Woldesenbet et al. 2020). Among the articles reviewed ,28 articles used DEM (30 m resolution), 3 articles used DEM 12.5m, 2 articles used 90m resolution DEM whereas 9 articles did not explicitly mention the resolution of DEM data used to calculate LS factor (Table 6). Several academics used different ways to compute the LS factor. These methods used to calculate LS factor considered flow direction, flow accumulation, slope, flow length and slope steepness (Borrelli et al. 2021; Endalamaw et al. 2021; Woldesenbet et al. 2020).
Table 4
Formulas used to calculate LS factor
| Ls factor formula | References |
| LS = (𝐹𝑙𝑜𝑤 𝑎𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑥 cell size / 22.13 )0.4 x (sin(𝑠𝑙𝑜𝑝𝑒) 0.0896 )1.3 | (Yadeta et al., 2021) |
| LS = Power (FA * cell size / 22.13, 0.4) * (0.065 + 0.045 * Slope% + 0.0065 * Slope% * Slope%) | (Duguma 2022) |
| LS = ([flow accumulation] *[cell size]/22.13)0.4*sin[slope gradient]/0.0896)1.3 | (Jothimani et al. 2022) |
| LS = ((A*cell size)/ 22.13 )0.4 * ((sin(slope) 0.01745) / 0.0896)) 1.3 * 1.4 | (Bekele, Gella, and Ejigu 2022) |
| LS = Power (Flow accumulation ∗ cell size / 22.1, 0.6) ∗ Power (Sin(Slope ∗ 0.01745) / 0.09, 1.3) | (Olika, Fikadu, and Gedefa 2023) |
3.3.4. C-factor (Cover Management):
The C-factor denotes the influence of land cover and land management methods on soil erosion. It considers vegetation cover, crop type, agricultural residue management, and conservation practices, among other things. The C-factor varies based on land use practices and vegetation cover. The value of the C- factor ranges between 1 and 0, where a value of 1 indicates a lack of cover and a value near zero indicates a solid cover. Representative C-factor values for various land cover types can be chosen from tables provided by Wischmeier and Smith (1965) or Hurni (1985)
Most studies used land sat data captured during the winter(dry season in Ethiopia), but exceptionally (Negese et al. 2021) tried to use Landsat data captured during the summer season when the crop cover is the highest in Ethiopia. High resolution spot 6 image is used to have a quality land use landcover classification (Endalew and Biru 2022). Among the papers reviewed, 37 of the articles used Land use land cover and 5 articles used NDVI (Normalized Difference Vegetation Index) to compute the c factor values (Table 4 and Table 6).
Table 5
C factor value based on land cover
| Land cover | C- value | References |
| Cropland | 0.15 | (Abathun et al. 2022) |
| Forest | 0.001 |
| Shrubland | 0.014 |
| Grassland | 0.010 |
| Settlement | 0.090 |
| Waterbody | 0.00 |
| Bare land | 1.00 |
| Grass land | 0.05 | (Moisa 2022) |
| Crop land | 0.18 |
| Forest | 0.001 |
| Bare land | 0.05 |
| Shrub land | 0.014 |
| Forestland | 0.001 | (Olika et al. 2023) |
| Waterbody | 0.00 |
| Cultivated land | 0.15 |
| Bare land | 0.05 |
| Grassland | 0.08 |
| Town (Urban) | 0.24 |
| Fallow Land | 0.00 |
3.3.5. P-factor (Support Practice):
It accounts for the impact of erosion control practices or conservation measures implemented on a given geographic area. It indicates the ratio of soil loss from managed fields to bare ground or cultivated fields up and down the slope. The p values range from 0 to 1, with 0 representing well-managed fields and 1 representing unmanaged fields(Yeneneh et al. 2022). It considers practices such as contour plowing, terracing, bunds, and other soil conservation techniques. The P-factor can vary depending on the adoption and effectiveness of conservation practices in different areas of Ethiopia (Hurni 1985).
Table 6
| Land Use | Slope (%) | P-Value | References |
| Bare land | * | 0.73 | (Moisa 2022) |
| Cultivated land | * | 0.9 |
| Forest | * | 0.53 |
| Grassland | * | 0.63 |
| Shrubland | * | 0.6 |
| Agricultural land | 0–5 | 0.1 | (Sinshaw et al. 2021) |
| 5–10 | 0.12 |
| 10–20 | 0.14 |
| 20–30 | 0.19 |
| 30–50 | 0.25 |
| 50–100 | 0.33 |
| Other land use | All | 1 |
| Forestland | * | 0.5 | (Olika et al. 2023) |
| Waterbody | * | 1 |
| Cultivated land | * | 0.9 |
| Bare land | * | 1.0 |
| Grassland | * | 0.9 |
| Town (Urban) | * | 1 |
| Fallow Land | * | 0 |
3.4. The Use of the RUSLE Model for Estimating Soil Loss:
The reviewed literature provides valuable insights into the use of the RUSLE model for estimating soil loss in various geographical contexts. Numerous studies have explored different versions and modifications of the RUSLE model, focusing on input parameters, data sources, model validation, and calibration, as well as the accuracy and uncertainty of model outputs. By organizing the literature based on key themes, trends, and findings, this section highlights the pertinent aspects of the RUSLE model's application for soil loss estimation.
3.5. Validation and Calibration of the RUSLE Model:
Validation and calibration of the RUSLE model play a crucial role in assessing its reliability and accuracy. Studies have employed field measurements, erosion plots, and comparison with observed data to validate and calibrate the model outputs. The literature reveals that the RUSLE model generally performs well at plot and small catchment scales, with reasonable agreement between predicted and observed soil loss. However, the model's performance may vary across different geographic locations and land management practices, emphasizing the need for site-specific calibration and validation.
3.6. Accuracy and Uncertainty of Model Outputs:
The accuracy and uncertainty of the RUSLE model outputs have been a subject of investigation in the literature. Studies have evaluated the reliability of soil loss estimates by comparing RUSLE results with measured erosion data. While the model generally provides reasonable estimates, there are inherent uncertainties associated with input data, parameter selection, and spatial variability. The literature suggests that the accuracy of the model can be improved by incorporating more detailed and localized data, refining parameterization methods, and considering the effects of uncertainties in the model's inputs.
Table 7
RUSLE parameters and the amount of soil loss estimated in the reviewed 42 articles
| Articles Number | Area (km2) | Study Year | Soil loss(t/ha/year) | R factor | C factor | LS factor | K factor | P factor |
| (Negese et al. 2021) | 125.82 km² | 1995–2018 | 38.70 | R = -8.12 + (0.562 × P) | NDVI | DEM 30 | Soil type | LC (Field observation) |
| (Woldesenbet et al. 2020) | 2110.4 km² | 1987 to 2017 | 31.910 | R = 0.562P − 8.12 | LC | DEM 30 | Soil type (color) | LC (Field observation) |
| (Degife et al. 2021) | 142,661 ha | * | 37.00 | R = -8.12 + (0.562 × P) | LC | DEM 30 | Soil unit/type | LULC |
| (Almaw et al. 2021) | 1,119,683 km2 | 1981–2016 | 16.50 | * | LC | DEM 90 | Soil texture | Slope class and LU |
| (Bekele et al. 2022) | 1591.56 km2 | 1988–2019 | 24.20 | R = EI30 = 100 R = -8.12 + (0.562 × P), | LC | DEM 30 | field observation, soil color | LULC, slope |
| (Moisa 2022) | 2,613.4km 2 | * | 83.70 | R = -8.12 + (0.562 × P), | LC | DEM 30 | soil type | land-use categories |
| (Jothimani et al. 2022) | 353 km2 | 2001–2019 | 68.47 | R = -8.12 + (0.562 × P), | LC | DEM 30 | soil type | LU, field observations |
| (Masha et al. 2021) | 100 km² | 1998–2020 | 29.62 | R = (P × 0.562) − 8.12 | NDVI | DEM 30 | soil color/type/ | slope class and LULC |
| (Wagari and Tamiru 2021) | 2148 ha | * | 76.50 | R = 1.24 *P 1.36 | LC | DEM 12.5m | Soil types | LULC |
| (Girmay et al. 2020) | 183.3 ha | * | 25.00 | R = (P × 0.562) − 8.12 | LC | DEM 30 | soil texture, OM, (lab result) | LULC and Slope |
| (Girma and Gebre 2020) | * | 2018 | 69.00 | R = -8.12 + (0.562 × P), | LC | DEM 30 | Texture | LULC with slope class |
| (Balabathina et al. 2020) | 292,230 ha | 1986–2018 | 37.89 | R = 0.55 ∗ P − 4.7 | NDVI LULC | DEM 30 | Texture and OM | LU. with slope class |
| (Tsegaye and Bharti 2021) | 1165 ha | * | 17.30 | R = -8.12 + (0.562 × P), | LC | DEM 30 | soil types with field data | slope and land use |
| (Abebaw et al. 2022) | 427 km2 | 2000–2020 | 32.84 | R = -8.12 + (0.562 × P), | LC | DEM 30 | Texture and OM | LULC and slope gradient |
| (Belay and Mengistu 2021) | 423 km2 | 1985 to 2017 | 19.70 | R = -8.12 + (0.562 × P), | LC | DEM 30 | soil types | LULC |
| (Yeneneh et al. 2022) | 80,343 ha | 1985 to 2019 | 31.40 | R = − 8.12 + 0.562P | LC | DEM 30 | Soil types(color) | LULC with slope class |
| (Abathun et al. 2022) | 48,348.4 ha | 1986–2018 | 64.20 | R = − 8.12 + (0.562 × P) | LC | DEM 12.5m | Soil texture | LU, slope class |
| (Ayele et al. 2022) | 26,093.7 ha | * | 37.72 | R = -8.12 + (0.562 × P) | NDVI | DEM* | Soil types(color) | LULC |
| (Yirgu 2022) | * | 2019 to 2020 | 30.60 | R = -8.12 + (0.562 × P) | LC | DEM 90m | Soil types | conservation measures(survey) |
| (Endalew and Biru 2022) | 43,317.45 ha | * | 30.00 | R = -8.12 + (0.562 × P) | LC (SPOT6) | DEM 30m | Texture and OM | LU, slope class |
| (Yebyo 2022) | 14,965.3 ha | 2017 | 110.06 | R = -8.12 + (0.562 × P) | LC | DEM* | Soil type | Conservation per slope class |
| (Atoma, Suryabhagavan, and Balakrishnan 2020) | 185.8 km² | 1998 to 2018 | 27.00 | R = -8.12 + (0.562 × P) | LC | DEM 30 | Soil texture | Conservation per LU |
| (Mahmud Mustefa 2020) | 7,790 km2 | * | 32.00 | R = -8.12 + (0.562 × P) | LC | DEM * | Soil type(color) | Conservation per slope class |
| (Olika et al. 2023) | 867 km2 | * | 13.27 | R = -8.12 + (0.562 × P) | LC | DEM 30 | Soil type(color) | LU and practices |
| (Getnet and Mulu 2021) | 859.2 km² | * | 27.7 | R = -8.12 + (0.562 × P) | LC | DEM * | soil type | conservation practices |
| (Yared et al, 2020) | 14,055 km2 | 2000 to 2016 | 31.00 | R = -8.12 + (0.562 × P) | LC | DEM * | soil type | LU, slope class |
| (Endalamaw et al. 2021) | 770 km² | 1993 to 2017 | 28.68 | R = -8.12 + 0.562∗ P | LC | DEM 30 | Soil Texture and soil type | LU, slope class |
| (Kanito et al. 2023) | 20,440.89 ha | 2011 to 2016 | 22.31 | R = -8.12 + 0.562∗ P | LC | DEM 30 | soil type-color | LU, slope class |
| (Dawit, 2022) | 8188 km2 | 1986 to 2020 | 64.00 | R = -8.12 + 0.562∗ P | LC | DEM 30 | soil color and types | LULC |
| (Melese et al. 2021) | 960 km2 | 1986–2016 | 24.30 | R = -8.12 + 0.562∗ P | LC | DEM 30 | soil type-color | LULC |
| (Usman et al. 2023) | 1277 km2 | 1987 to 2021 | 45.35 | R = 8.12 + (0.562 ∗ p) | LC | DEM 20 | Soil texture and OM | LU slope class (field data) |
| (Gashaw et al. 2020) | 25,609 ha | 2007 to 2015 | 23.7 | R = − 8.12 + (0.562 × P) | LC | DEM 30 | soil color | LU slope class |
| (Assefa et al. 2022) | * | 1986–2015 | 20.01 | R = − 8.12 + (0.562 × P) | LC | DEM 30 | soil texture, OM | conservation measures |
| (Lemma et al. 2022) | 374.98 km2 | * | 43.21 | R = (P × 0.562) − 8.12 | LC | DEM 30 | Soil texture | conservation practices |
| (Desalegn et al. 2023) | 7465 ha | 1989 to 2018 | 12.94 | R = − 8.12 + (0.562 × P) | LC | DEM 30 | Texture and OM (lab result) | field surveys and secondary sources |
| (Tessema, Simane, and Angassa 2023) | 33,376 ha | * | 37.54 | R = − 8.12 + (0.562 × P) | LC | DEM * | soil color | LU and-site observation |
| (Duguma 2022) | 6785 km2 | 2011–2020 | 25.23 | R = − 8.12 + (0.562 × P) | LC | DEM * | soil texture, OM | LULC |
| (Asmamaw, Mohammed, and Mathias 2021) | 2,148.72 km2 | 2016 | 20.00 | R = -8.12 + 0.562∗ P | LC and NDVI | DEM * | Soil type | LU slope class |
| (Gitima et al. 2023) | 59,901.80 ha | 1985 to 2021 | 40.93 | R = − 8.12 + 0.562Z | LULC | DEM 30 | Soil type | LU slope class |
| (Ejegu and Yegizaw 2021) | 11088.42 ha | 2000–2020 | 41.07 | R = − 8.12 + (0.562 × P) | LULC | DEM * | Soil type | LU slope class |
| (Yadeta et al, 2021) | * | 1992–2016 | 13.20 | R = − 8.12 + (0.562 × P) | LC | DEM 30 | Soil type | conservation practices |
| (Eniyew et al. 2021) | 20,205 ha | 2005 to 2019 | 576 | R = 0.562 × P − 8.12 | LC | DEM 12.5 | soil color | conservation practices |
“R is erosive power of rainfall, p is mean annual precipitation in millimeters, Z mean yearly rainfall, EI 30 = EI is kinetic energy (30 min rainfall intensity) of raindrops, * no available data, DEM (Digital Elevation Model), ha(hectare), LULC (Land Use Land Cover), NDIV (Normalized Difference Vegetation Index)”.
As stated above in Table 6, the RUSLE model considers five parameters, such as rainfall erosivity (R), soil erodibility (k), cover management (c), slope length and steepness (LS), and support practice (p). Even though the five parameters were used in the reviewed 42 articles, they needed to be moreonsistent based on the data used, the methods and techniques applied, and the mathematical algorithms. Besides, the articles were different in considering the number of years considered in rainfall data, which greatly impacts the expected result. There is also diversity in using the data type, for instance, the resolution of satellite data, which impacts the qualities of C factor and LS factor maps
Table 6 information presents the reviewed studies that focus on soil loss estimation in various regions and time periods. These studies employ different factors and methodologies to estimate the average annual soil loss in tons per hectare per year (t/ha/yr.). The research encompasses diverse spatial scales, ranging from small-scale regions like 183.3 hectares to large-scale national levels of 1,119,683 km2. The reported annual average soil loss estimates range from 12.94 t/ha/yr to a significantly high estimation of 576-t/ ha − 1 year − 1, reflecting the influence of climatic, topographic, and land use factors.
To estimate soil loss, the studies employed factors like R, C, LS, K, and P factors, which were calculated based on NDVI, DEM, soil texture, organic matter content, and land use/land cover data. The variation in methodologies, which includes laboratory analyses, remote sensing, field observations, and expert knowledge, contributes to the diverse results. Long-term studies spanning several decades offer valuable insights into the temporal trends and changes in soil erosion patterns.
Several studies discuss soil conservation practices and their potential impact on mitigating soil loss (Abebaw et al., 2022; Alemu & Melesse, 2020; Atoma et al., 2020; Ayele et al., 2022; Balabathina et al., 2020; D. A. Bekele et al., 2022; Getnet & Mulu, 2021; Girmay et al., 2020; Negese et al., 2021; Yeneneh et al., 2022). Understanding the effectiveness of these measures is crucial for implementing appropriate strategies and policies to maintain sustainable land use practices. Policymakers, environmental agencies, and researchers can utilize the data to identify vulnerable areas, prioritize conservation efforts, and promote sustainable land management practices. However, to draw more comprehensive conclusions and identify overarching trends, further analysis of the whole research paper is necessary, considering methodologies, data sources, and potential limitations. Overall, the information underscores the significance of soil conservation efforts and ongoing research to address soil erosion and its potential environmental implications.
Table 8
Summary of Soil Loss Estimation Results
| Soil loss (t/ha/yr.) | Severity class | Number of Studies |
| 0 to 5 | Low | 0 |
| 5 to 20 | moderate | 6 |
| 20 to 50 | High | 28 |
| 50 to 100 | very high | 6 |
| 100 to 150 | Severe | 1 |
| 150 to 716 | Extreme | 1 |
| Total | | 42 |
Source for Severity classes (Molla and Sisheber 2017).
3.7. Strengths and Limitations of the RUSLE Model for Estimating Soil Loss:
Based on the evidence and arguments presented in the literature, the strengths and limitations of the RUSLE model for estimating soil loss can be identified. The model's strength lies in its simplicity, integrating multiple factors influencing soil erosion into a comprehensive framework (Hurni 1985). The RUSLE model has demonstrated its applicability across diverse geographic regions and scales, making it a widely adopted tool for soil loss estimation(Morgan 2005). However, limitations exist, such as the reliance on average and generalized values for input parameters, which may overlook site-specific characteristics. Additionally, the model's sensitivity to input data and assumptions and uncertainties in parameter estimation can introduce variability and affect the accuracy of soil loss predictions (Melese et al. 2021).
Overall, the review underscores the significance of the RUSLE model in estimating soil loss and provides a comprehensive understanding of its strengths, limitations, and potential improvements. The findings of the reviewed studies contribute to the knowledge base surrounding the RUSLE model's application and highlight areas that warrant further research and refinement to enhance its accuracy and applicability in soil conservation practices.
The review article discusses some factors contributing to soil erosion, including climate, topography, soil properties, land use practices, and vegetation cover. The article also identifies some strategies that can be used to prevent or mitigate erosion, including the implementation of conservation tillage, the use of cover crops, the establishment of buffer strips, and the adoption of agroforestry practices. The article emphasizes that effective soil conservation strategies need to be tailored to the specific conditions of each region, and should take into account the local socio-economic and cultural context.
3.8. Key Findings and their Significance:
Summarizing the key findings, it is evident that the RUSLE model has undergone considerable development and modification to enhance its accuracy and applicability. The integration of advanced features, such as GIS, remote sensing data, and erosion control practices, has contributed to improved soil loss estimations. The model has shown satisfactory performance at plot and small catchment scales, providing reasonable agreement with observed data. Its simplicity and ability to encompass various factors influencing soil erosion have contributed to its wide adoption in soil loss estimation.
3.9. Comparison with Other Soil Erosion Models:
Comparing the RUSLE model with other soil erosion models, it becomes apparent that each model has its own set of advantages and disadvantages. The RUSLE model's strength lies in its simplicity, which allows for quick estimations and broad-scale assessments. In contrast, more complex models may offer higher accuracy and precision but require extensive data inputs and computational resources. However, the RUSLE model's reliance on average and generalized values for input parameters may overlook site-specific characteristics, which could limit its accuracy in certain contexts. Therefore, the choice of soil erosion model should be based on the specific research or practical objectives, considering the trade-off between simplicity and precision.
3.10. Gaps and Challenges in the Use of the RUSLE Model:
Despite its wider use and significant advancements, several gaps and challenges persist in the application of the RUSLE model for estimating soil loss. One of the key challenges is the availability and quality of input data. Accurate estimations require detailed and up-to-date information on parameters such as rainfall erosivity, soil erodibility, and land cover, which may not always be readily available or easily accessible. Furthermore, the spatial and temporal variability of erosion processes poses a challenge in capturing localized erosion patterns and predicting erosion dynamics accurately. The model's simplicity, while advantageous in some respects, may limit its ability to account for these complex spatiotemporal variations.
3.11. Suggestions for Future Research and Practice:
The findings presented in the previous section offer valuable insights into the use of the RUSLE model for estimating soil loss. These findings significantly advance knowledge and understanding of the RUSLE model's application and its implications for soil conservation practices. Based on the insights and recommendations presented in the literature, several avenues for future research and practice in soil erosion estimation using the RUSLE model can be identified. Firstly, efforts should be directed toward improving data availability and accessibility by leveraging advancements in GIS and remote sensing technologies, data-sharing platforms, and collaborative initiatives. This would enable more accurate parameterization and validation of the model, enhancing its reliability across different geographic regions and scales.
Additionally, research should focus on refining the model's input parameters for site-specific characteristics, variability, and uncertainty. This can be achieved through field-based measurements, data assimilation techniques, and improved modeling of erosion processes. Moreover, investigations into the integration of the RUSLE model with other models and tools, such as hydrological models and decision support systems, would facilitate a more comprehensive understanding of the soil erosion processes and their impacts.
Furthermore, the validation and calibration of the RUSLE model should be extended to a broader range of contexts, considering different land use practices, erosion control measures, and climate conditions. This would enhance the model's performance and reliability across diverse landscapes. Additionally, studies should focus on quantifying and addressing the uncertainties associated with the model's outputs to provide users with a better understanding of the confidence level and potential limitations of the estimations.
The findings underscore its strengths, such as simplicity and broad-scale applicability, while acknowledging its limitations, particularly in capturing site-specific characteristics and accounting for spatiotemporal variability. To advance research and practice in this field, future efforts should focus on improving data availability and accessibility, refining input parameters, and validating the model across diverse contexts. Additionally, exploring the integration of the RUSLE model with other erosion models and tools would offer a more comprehensive understanding of soil erosion processes. Furthermore, addressing uncertainties and quantifying the confidence level of model outputs would enhance its practical utility. By pursuing these avenues, researchers and practitioners can contribute to the continuous improvement and effective utilization of the RUSLE model in soil erosion estimation and inform soil conservation practices for sustainable land management.