Introduction

Land degradation caused by soil erosion is one of the most widespread and critical environmental problems in the world (Zuazo and Pleguezuelo 2008; Ganasri and Ramesh 2016; FAO and ITPS 2015; Li et al. 2014; Sharma et al. 2011). At a time when agricultural efforts are focused on increasing food production, soil erosion is worldwide increasing (Pimentel 1993). It is estimated that about 75 billion tons of soil is removed from the world’s terrestrial ecosystem due to erosion (Dabral et al. 2008), which is approximately one-sixth of the world’s total land size (Hurni et al. 2010). The rate and extent of soil erosion is severe in the world’s agricultural regions (Pimentel 1993) where soil is being washed 10–40 times faster than it is being replenished (Pradhan et al. 2012).

The fact that soil is almost non-renewable resource over the human time-scale makes soil erosion a critical problem (Bewket and Teferi 2009). It requires at least 500 years for the formation of 2.5 cm of topsoil under tropical and temperate agricultural conditions (Pimentel 1993). As a result, soil erosion assessment has gained a great attention in twentyfirst century due to its importance as a base for developing effective soil conservation plans (Ali and Hagos 2016). Soil erosion is considered to be one of the greatest environmental problems in Ethiopia where subsistence rain-fed agriculture constitutes the basis of the economy (Reusing et al. 2000; Amsalu and de Graaff 2007). Its severity is more pronounced in the highland areas which are characterized by steep slopes, intensive rainfall, high population, and livestock densities (Kidane 2016; Abate 2011; Shiferaw and Holden 1999; Hailu et al. 2015; Subhatu et al. 2017; Moges and Taye 2017).

The annual rate of soil loss in Ethiopia is estimated at 1.5 billion tons (FAO 1986; Hurni 1993), of which more than 50% occurs in cropland (Assefa and Bork 2015) and 10% of it crosses the national boundary (Ali and Hagos 2016). Many studies have been conducted in the area of soil erosion in Ethiopian highlands and revealed the alarming rate of the problem. For instance, the study by FAO (1986) showed the estimated average annual soil loss of 100 t ha−1 year−1 from the cultivated lands. The average annual soil loss rate of 42 t ha−1 year−1 was also estimated by Hurni (1993). Moreover, recent studies in north-western highland Ethiopia by Bewket and Teferi (2009) and Gelagay and Minale (2016) presented the estimated mean annual soil loss rate of 93 and 47 t ha−1 year−1, respectively. Such excessive rate of soil erosion constitutes a real threat to economic and ecological sustainability in Ethiopia.

The problem of soil erosion in Ethiopia is often the product of complex interactions between human and the ecosystem (Nyssen et al. 2004; Pimentel 1993). The LULU change, which is the result of intended or unintended decisions and the subsequent actions, is one of the easiest indicators of nature–human interaction (Kindu et al. 2015; Teferi et al. 2016). The LULC change in combination with climatic and geomorphologic conditions of the given area has an accelerated impact on environmental degradation in the form of soil erosion, nutrient leaching, soil acidification and organic matter depletion (Sharma et al. 2011). Many researches (Tamene and Vlek 2007; Garedew et al. 2009; Minale and Rao 2012; Kindu et al. 2015; Meshesha et al. 2014; Kidane et al. 2014; Mengistu and Waktola 2016; Teferi et al. 2016) have been conducted in Ethiopia to analyse LULC changes and their consequences. However, only few studies (Mengistu and Melesse 2011; Ciampalini et al. 2012; Tadesse et al. 2017; Tesfaye et al. 2016) have been carried out on the impacts of LULC change on soil erosion, and the knowledge of spatio-temporal interaction between LULC change and soil erosion is limited, particularly in the Rib watershed. Therefore, estimation and mapping of soil loss caused by LULC change is very crucial in the study area to provide scientific basis for land use planning and soil conservation decisions.

Estimation of soil loss is often a difficult task due to the complex interplay of biophysical and socioeconomic components. Therefore, effective estimation of soil loss needs approaches capable of integrating and addressing these components and their spatial and temporal variability. In recent decades, integration of geospatial technologies, primarily remote sensing and geographic information system (GIS), with erosion estimation models has made soil erosion assessment and its spatial distribution attainable at a high accuracy and low cost (Wang et al. 2003). Several models have been developed to estimate soil losses worldwide. For instance, universal soil loss equation (USLE) (Wischmeier and Smith 1965, 1978), revised universal soil loss equation (RUSLE) (Renard et al. 1997), Soil and Water Assessment Tool (SWAT) (Arnold et al. 1998), Water Erosion Prediction Project (WEPP) (Nearing et al. 1989), Chemical Runoff and Erosion for Agricultural Management System (CREAMS) (Knisel 1980), and European Soil Erosion Model (EuroSEM) (Morgan et al. 1998) are some of the models which have been developed and utilized for decades.

Among these prediction models, RUSLE in combination with remote sensing and GIS has been frequently used by several researchers worldwide (Kavian et al. 2017; Shit et al. 2015; Jabbar 2003; Bagherzadeh 2014; Farhan and Nawaiseh 2015; Tanyas et al. 2015; Fagnano et al. 2012; Fu et al. 2005; Ganasri and Ramesh 2016; Kumar et al. 2014). The RUSLE is an empirical model designed to predict annual soil loss carried by runoff from slopes with specified cropping and management conditions (Renard et al. 1997; Le Roux et al. 2005). Even though RUSLE is the blind model in prediction of erosion caused by gullies (Renard et al. 1997), it has been recognized as the most effective empirical model to assess the average annual rate of soil erosion caused by raindrop impact and surface runoff (rill and sheet erosion) (Kouli et al. 2009). Compared to other process-based erosion estimation models, RUSLE model is characterized by numerous benefits. It is more flexible, easy to implement, and compatible with geospatial information technologies (Bhandari et al. 2015; Mhangara et al. 2012). The data required to run RUSLE model are not complex and are easily accessible (Farhan and Nawaiseh 2015; Yue-Qing et al. 2008).

Moreover, RUSLE model enables the researchers to predict erosion rates of ungauged watersheds by using knowledge of the watershed characteristics and local hydroclimatic conditions (Yue-Qing et al. 2008). Unlike models that predict soil losses based on small experimental sites, RUSLE model provides better accurate results for watershed and regional level studies (Prasannakumar et al. 2012). This model also makes soil erosion estimation and observation results feasible at a reasonable cost (Farhan and Nawaiseh 2015). Based on these and other various merits, RUSLE model was selected and used in present study in integration with GIS and remote sensing to analyse the effect of LULC change on soil erosion in the Rib watershed, north-western highland Ethiopia, over the period 1986–2016. The specific objectives of this study were: (1) to examine the spatio-temporal dynamics of LULC changes and soil erosion; (2) to examine the effect of LULC change on soil erosion in the study area.

Materials and methods

Description of the case study site: the Rib watershed

The Rib watershed is located in the north-western highland Ethiopia which extends from the top of Mount Guna to the eastern part of Lake Tana, forming the headwater source of Blue Nile River. It lies between 11°40′N–12°20′N latitude and 37°30′E–38°20′E longitude (Fig. 1) and covers a total area of 1975 square kilometres. In administrative terms, it is located in South Gondar Zone, Amhara Regional State. Its elevation ranges from 1758 m (around Lake Tana, outlet) to 4104 m (at the top, Guna Mountain). The watershed is the main source of water for Lake Tana, which together with Gumara, Megech and Gilgel Abay contributes more than 90% of inflow water (Setegn et al. 2009; Dile et al. 2013). The topography of the study area is generally increasing in elevation from the downstream to upstream (Fig. 1). A mountainous and hilly dissected terrain with steep slopes characterizes significant parts of the watershed. The most elevated areas are found in the south-eastern and some northern parts of the watershed, and it declines towards west and south-west.

Fig. 1
figure 1

Location map of the study area

Full size image

The climate condition of Ethiopia generally ranges from equatorial rainforest in the south and south-west parts of the country to the desert areas of the northeast, east and southeast lowlands (Awange et al. 2014). Based on traditional classification system, it can be classified into five agro-climatic zones (Hurni 1998; Dejene 2003): Bereha (hot arid), Kolla (warm semi-arid), Woyna-Dega (cool sub-humid), Dega (cool and humid) and Wurch (Alphine). The climate of the study area is dominantly humid (Dega) with some areas undergoing sub-humid and Alphine climate. The average annual rainfall and temperature in the study area (1986–2016) was 1503.43 mm and 15.62 °C, respectively. During the study period, about 80.24% of the annual rainfall had occurred between June and September (Fig. 2), which refers to the main rainy season (locally known as Kiremt). The minor rainy season or autumn (locally known as Belg) is also an important season for the farming system of the area. This season usually ranges from March to May.

Fig. 2
figure 2

Mean monthly temperature and rainfall in Rib watershed (1986–2016)

Full size image

According to the population and housing census result obtained from the Ethiopian Central Statistical Authority (CSA 2007), the total population of the watershed was 376,256 with corresponding population density of 190 people per square kilometre. More than 85% of the population lives in rural areas where farming is the main source of food and income. The agriculture in this area is characterized by rain-fed and mixed crop-livestock production with an average farm size of not more than 1 ha. The major crops grown in the watershed include maize (Zea mays L.), barley (Hordeum vulgare), teff (Ergarostis teff Zucc), wheat (Triticum vulgare), beans (Phaseolus vulgaris L.), rice (Oryza galberrima), and Potato (Solanum tuberosum). The common livestock types raised in the area include cattle, equine, and poultry. It is rare to find farmers who are not raising at least one livestock of any kind.

Based on USDA soil taxonomy, eight different soil groups were identified in the study area, which include Alisols, Cambisols, Ferrasols, Fluvisols, Leptosols, Luvisols, Regosols, and Vertisols. Luvisols (30%), Vertisols (23%), and Leptosols (21%) were the dominant soil types. The major LULC types identified in the watershed include cropland, grassland, shrubland, forestland, built-up land, and waterbody. Cropland was the dominant LULC type covering 88% of the total size followed by grassland.

The Rib watershed is endowed with diverse vegetation types due to its location in the highland area combined with topographic variations. Some of indigenous and high value tree species observed in the watershed include Millettia ferruginea (locally known as Birbera), Strychnos spinose (Dokma), Rosa abyssinica (Kega), Hagenia abyssinica (Kosso), Acacia bussei (Girar), Cordia Africana (Wanza), Olea africana (Weira), and Ficus sp. (Warka). However, the coverage of these plant species has indicated significant and continuous reduction in the past few decades mainly due to the expansion of cropland and exotic pant species, for example eucalyptus. Plantation of eucalyptus, usually by removing the natural forests, is the common practice in the watershed because of its fast growing nature for immediate household purpose and better performance in area with limited water and soil fertility.

The RUSLE model

The soil erosion risk in present study was estimated for 1986 and 2016 using RUSLE model. This model predicts the long-term average annual rate of soil erosion on a field slope based on five input factors in raster data format: rainfall erosivity (R); soil erodibility (K); slope length and steepness (LS); cover management (C); and support practice (P). These input data were obtained from the meteorological stations; available soil map, and satellite images (Fig. 3) with intensive field observation. The RUSLE equation is expressed as (Renard et al. 1997):

$$ A = R*K*LS*C*P $$
(1)

where A is average soil loss (t ha−1 year−1); R is rainfall-runoff erosivity factor (MJ mm ha−1 h−1 year−1); K is soil erodibility factor (t h MJ−1 mm−1), LS is slope length and steepness factor (dimensionless), C is crop and management factor (dimensionless) and P is conservation practice factor (dimensionless).

Fig. 3
figure 3

Flowchart of approach used to assess soil erosion risk

Full size image

Rainfall-runoff erosivity factor (R)

The R factor is the ability of rainfall and runoff to detach and transport soil at a particular area (Renard et al. 1997). An increase in the intensity and amount of rainfall results an increase in the values of R. In the original equation of RUSLE, the value of R measures the kinetic energy of the rain, and it requires measuring the rainfall intensity with autographic records continuously. However, due to lack of recorded rainfall intensity data in the study area, the empirical equation developed by Hurni (1985) for Ethiopian conditions was used to estimate R factor value. It depends on the easily available mean annual rainfall and given by the regression equation:

$$ R = - \;8.12 + 0.562*P $$
(2)

where R is the erosivity factor and P is the mean annual rainfall in mm.

Daily rainfall data series for 12 stations in and around the Rib watershed were obtained from the National Meteorological Agency (NMA) of Ethiopia. Rainfall data qualities such as data completeness, length and homogeneity were checked for each station over the period 1986–2016. Table 1 summarizes the name and location of selected stations, elevation, and average annual rainfall during the study period. Kriging interpolation in ArcGIS 10.3 was utilized to compute R factor values and develop the R factor map (Fig. 4a).

Table 1 Summary of meteorological stations in and around the Rib watershed (1986–2016)
Full size table
Fig. 4
figure 4

Spatial distribution of RUSLE factors: rainfall-runoff erosivity factor (a), soil erodibility factor (b), slope (c), and topography factor (d)

Full size image

Soil erodibility factor (K)

The K factor expresses inherent erodibility of the soil or surface materials (Demirci and Karaburun 2012). It shows an average long-term soil and soil profile responses to the impact of raindrop on the soil surface. Soil erodibility basically depends on spoil physical and chemical properties such as soil texture, aggregate stability, shear strength, infiltration capacity, organic content, and several chemical compositions. However, in conditions where there is unavailability of such detail data of soil properties, different researchers (Hellden 1987; Bewket and Teferi 2009) suggested to use soil colours as the base for K factor determination in Ethiopia. A similar method was applied to obtain K values in this study. The soil map of the watershed was extracted from the soil map of Ethiopia (FAO/UNESCO), and the K values were assigned for each extracted soil types based on their respective colours, as suggested by Amsalu and Mengaw (2014), Bewket and Teferi (2009), and Brhane and Mekonen (2009) (Table 2). The identification of soil colours was also supported by field survey results. The K factor map is indicated in Fig. 4b.

Table 2 K factor values (t h MJ−1 mm−1) of different soil types
Full size table

Topography factor (LS)

The LS factor describes the effect of both steepness and slope length of the field on soil erosion (Renard et al. 1997). Slope length (L) defines the distance from the point of origin of the overland flow to the place where either the slope gradient decreases enough that deposition begins or the runoff water enters a well-defined channel that may be part of a drainage network (Smith and Wischmeier 1957), whereas slope steepness (S) refers to the rise or fall of the land surface. In many studies, these variables have been considered as most important parameters of RUSLE model, and they are commonly combined in a single index as LS refer to topography. The LS factor of the study area was estimated by using the following equation (Prasannakumar et al. 2012; Pradeep et al. 2015; Demirci and Karaburun 2012; Lee 2004):

$$ {LS} = ({\text{Fac}}*{\text{cell}}\;{\text{resolution}}/2.13)^{0.4} *({ \sin }\;{\text{slope}}/0.0896)^{1.3} $$
(3)

where Fac (flow accumulation) is the number of cells contributing to flow into a given cell and derived from the ASTER GDEM after conducting fill, flow direction and flow accumulation processes, cell resolution is the size of the grid cells (for this study 30 m), and sin slope is the sin of slope angle in degree. The slope map of study area (in degree) and the LS factor grid map are shown in Fig. 4c, d, respectively.

Crop and soil management factor (C)

The C factor reflects the effect of cropping and management practices on the soil erosion rate (Wischmeier and Smith 1978). It is the effect of soil-disturbing activities, vegetation, crop sequence, soil cover and subsurface bio-mass on soil erosion (Prasannakumar et al. 2012). Vegetation cover protects the soil by dissipating the raindrop energy before it reaches the soil surface (Demirci and Karaburun 2012; Bhandari et al. 2015). The C factor for the present study was computed from LULC map of the watershed. The LULC map was generated from Landsat satellite image through supervised image classification technique. Finally, reclassification technique was applied to identify six LULC classes and to assign the corresponding C factor values (Table 3). The suggestions of different studies (Bewket and Teferi 2009; Amsalu and Mengaw 2014; Hurni 1985; Ganasri and Ramesh 2016; Balasubramani et al. 2015) were considered to assign the C factor values. The C factor maps of 1986 and 2016 are shown in Fig. 5a, b, respectively.

Table 3 C factor values of different land use/cover classes
Full size table
Fig. 5
figure 5

Spatial distribution of RUSLE factors: C factor for 1986 (a), C factor for 2016 (b), P factor for 1986 (c), and P factor for 2016 (d)

Full size image

Conservation practices factor (P)

The P factor is the soil loss ratio with a specific support practice to the corresponding soil loss with up and down slope tillage (Renard et al. 1997). The type of soil management practice, such as terracing, contour tillage, and permanent barriers or strips, reduces the overall amount of erosion (Demirci and Karaburun 2012). The value of P ranges from 0 to 1, in which the highest value is assigned to areas with no conservation practices. A detail field survey was conducted, and there were no organized data available for identifying soil and water conservation practices in the study area. Only small areas have been treated with poorly designed stone terraces, bunds and check dams. The existing conservation measures were also not well maintained. Therefore, the P factor values of the present study were computed by combining the LULC classes with slope categories, as suggested by Wischmeier and Smith (1978). The entire watershed was classified into two broad classes: agricultural land (including both crop and pasture land) and other land (non-agricultural land). The agricultural land was further subdivided into six slope classes by superimposing land use and slope maps using ArcGIS 10.3 software. Finally, the P factor value for each class was assigned (Table 4). The P factor maps of 1986 and 2016 are shown in Fig. 5c, d, respectively.

Table 4 Adopted values of P factor in different land use/cover classes
Full size table

The average annual soil loss was calculated using a cell-by-cell multiplication of these five parameter layers (R, K, LS, P, and C) in ArcGIS 10.3 software. To evaluate the effect of LULC change on soil erosion, the RUSLE was run for 1986 and 2016 separately. While running the model, the three inputs (R, K and LS) remained constant and the two inputs (C and P) were changed to the respective year, due to LULC change. The rainfall interpolation in this study was done by using the 30 years precipitation data (1986–2016). As a result, the same R factor values was used to calculate the annual soil loss of initial (1986) and final (2016) years. As well, the changes in soil erodibility and topography in the study area were insignificant over the period 1986–2016, and the same K and LS factor values were used to calculate the soil loss of both years.

Results and discussion

Land use/cover change

The major LULC classes identified in the Rib watershed include cropland, grassland, shrubland, forestland, built-up land and waterbody. Cropland was the dominant LULC type in the watershed, covering 77.6 and 88% of the total area in 1986 and 2016, respectively. Table 5 indicates the statistical summary of different LULC classes in the study area in 1986 and 2016. The spatial distribution of LULC types in 1986 and 2016 are indicated in Fig. 6a, b, respectively. The LULC change analysis result revealed that grassland and shrubland were significantly converted into cropland and built-up land during the study period (Table 5). This might be attributed to rapid population growth and advancement in living standard of the society, which has created a high competition on land for agriculture, residence, infrastructure, and urban expansion.

Table 5 Land use and land cover change analysis (1986–2016)
Full size table
Fig. 6
figure 6

Spatial distribution of land use/cover for 1986 (a), land use/cover for 2016 (b), soil erosion for 1986 (c), and soil erosion for 2016 (d)

Full size image

Slight increment in forestland (1.11% per year) was observed in the watershed over the period 1986–2016. This change was basically due to the expansion of eucalyptus plantation as a means of livelihood of the community. Waterbody also considerably increased from 0.7 km2 in 1986 to 6.5 km2 in 2016 (27.62% increase per year). The construction of Rib dam (in the eastern part of the watershed, Fig. 6b) has played a significant role for this drastic increase of waterbody during the study period. Moreover, the regeneration of upslope vegetation might be another reason for expansion of water at downstream areas. This result is in agreement with the findings of Nyssen et al. (2009) and Descheemaeker et al. (2006), which indicated the increasing availability of water at downstream area when the upslope landscape becomes revegetated.

Soil erosion change

The input layers of RUSLE model were converted into grid cells of 30 x 30 m in a uniform coordinate system and multiplied to estimate the annual soil erosion and to recognize its spatial distribution in the watershed. The estimated annual soil erosion (A) was finally classified into six severity categories by using soil erosion rate standards suggested by Morgan (2005) and Gemechu (2016). These categories include very slight erosion for A < 3 t ha−1 year−1, slight for 3 ≤ A < 10 t ha−1 year−1, moderate for 10 ≤ A<25 t ha−1 year−1, high for 25 ≤ A<50 t ha−1 year−1, very high for 50 ≤ A<75 t ha−1 year−1, severe for A ≥ 75 t ha−1 year−1. Table 6 and Fig. 6 illustrate the statistical summary of soil erosion and its spatial distribution in the study area in 1986 and 2016.

Table 6 Summary of soil erosion categories and their changes (1986–2016)
Full size table

The annual soil loss in the watershed ranged from 0 to 236.5 t ha−1 year−1 in 1986 and 0 to 807 t ha−1 year−1 in 2016. The average annual soil loss for the entire watershed was estimated at 40 and 68 t ha−1 year−1 in 1986 and 2016, respectively, a substantial increase. As shown in Table 6, the areas under very slight, slight, and moderate erosion categories diminished from ~ 93% in 1986 to 58% in 2016. On the other hand, the areas under high, very high and severe erosion categories increased from 7% in 1986 to 42% in 2016. The implication of this result is that the parts of watershed with minimum soil erosion intensity in 1986, which were mostly found in the western and central part of the watershed (Fig. 6c), were changed into the next higher erosion categories in 2016 (Fig. 6d). This shift in erosion intensity was basically due to human intervention on natural environment. The areas with soil erosion categories from high to severe were mostly located in the eastern, south-eastern, and some northern parts of the watershed where frequent cultivation of steep and marginal lands has been a common practice.

In Rib watershed, soil erosion potential that was estimated to exceed the average soil loss tolerance level of 11 t ha−1 year−1 (Bewket and Teferi 2009; Hudson 1981) increased from 34.5% in 1986 to 66.8% in 2016. Soil loss tolerance level is the maximum rate of soil erosion that may occur and still permits a high productivity in a given land (Wischmeier and Smith 1978). For proper decision and effective conservation, tolerance level should be determined scientifically and rationally based on real situation of a given area, like soil depth and topography. Bewket and Teferi (2009), furthermore, suggested the concept of critical soil loss and its importance in conservation and land use planning decisions. Critical soil loss is the level where reducing soil erosion is no more possible regardless of the effectiveness of conservation measures used. Therefore, these two points, i.e., soil loss tolerance and critical soil loss level, can be used as indispensable indicators to recognise the degree of soil erosion and decide on erosion intervention issues.

Areas with critical soil erosion level in the study area were observed in the eastern part of the watershed (Fig. 6d) where the maximum amount of soil loss (807 t ha−1 year−1) was detected. Based on the above suggestion, in such critical soil loss areas, the natural resource conservationists and local communities should focus on changing or modifying the current LULC type to another ecologically friendly type, rather than investing on soil and water conservation programs.

Impact of land use/cover change on soil erosion

The results of present study showed that there was complex change in soil erosion associated with dynamic LULC changes over the study period. As depicted in Fig. 6a, the western, northern, and south-eastern parts of the watershed were densely covered with grassland and shrubland in 1986. However, this LULC classes were significantly converted into cropland and built-up land in 2016. This conversion, in turn, has directly caused the shift in soil loss rate from very slight, slight, and moderate levels in 1986 (Fig. 6c) to the next higher erosion levels in 2016 (Fig. 6d). The detail cause–effect relationship between LULC dynamics and subsequent changes in soil erosion potential was analysed by superimposing soil erosion intensity maps with LULC change maps. Table 7 provides the detail of change in area of different soil erosion intensity classes due to LULC change over the study period.

Table 7 Areas of different soil erosion categories for each LULC types (km2) (1986–2016)
Full size table

As seen in Table 7, the cropland under very slight and slight erosion categories decreased from 953 km2 in 1986 to 495 km2 in 2016, while the areas of other erosion categories from moderate to severe showed significant increase from 578 km2 in 1986 to 1241 km2 in 2016. The implication is that large part of cropland which was under minimum state of soil erosion in 1986 has been changed into the next higher erosion categories in 2016. The main factors for this increase in soil erosion severity might be highly attributed to clearing of vegetation, frequent cultivation, overgrazing, cultivation of marginal lands, and removal of crop residues from the farm lands. As indicated in Table 4 and confirmed during filed observation, more than 65% of cropland was located in areas where slope is exceeding 10 percent, and natural resource conservation practices in these areas were also not satisfactory. Therefore, cultivation of steep slopes with poor conservation practices might be another contributing factor for higher potential of soil erosion in croplands. Waterbody and built-up land also showed obvious increase at all erosion intensity classes due to extensive increase in land use area.

The effect of LULC change on soil erosion can be also analysed by calculating mean annual soil loss amount of different LULC types (Table 8). The most noticeable change in soil erosion intensity was observed from cropland with mean annual soil loss amount increased to 41.38 t ha−1 year−1 in 2016 from 26.60 in 1986. The second largest change in soil erosion amount was seen in grassland, with mean annual soil loss amount increased from 13.18 t ha−1 year−1 in 1986 to 33.87 t ha−1 year−1 in 2016. The dramatic increase in soil losses from cropland and grassland could be highly attributed to increasing human activities to meet the demand of ever-growing population in the study area.

Table 8 Statistical results of soil loss intensity in 1986 and 2016 per various LULC types
Full size table

Although forestland was expected to have low soil erosion intensity, the result showed that the mean annual soil erosion potential had undergone a great degree of change from 17.27 t ha−1 year−1 in 1986 to 33 t ha−1 year−1 2016. The highest potential of soil loss in forestland might be due to poor coverage of trees in the forest and persistent removal of trees (most likely eucalyptus trees planted adjacent to natural forests) for different livelihood purposes. Significant part of forestland was confined to hillside areas of the watershed. So, highly slopping areas are most likely to facilitate runoff and reduce infiltration and finally increase soil erosion in forestland.

Validation of the model used to predict soil erosion is essential to ascertain the quality of results and test the usefulness of the model. However, the absence of measured sediment yield data in the study area constrained comparison of predicted erosion values with field derived values. Field observation and image analysis were used to validate only the presence and absence of soil erosion in the study area. Previous studies in the north-western highland Ethiopia (Nurelegn and Amare 2014; Estifanos 2014; Sewnet 2015) were used to crosscheck the consistency of predicted results and effectiveness of the model. Therefore, the estimated soil loss rate and spatial distribution of erosion classes in our watershed are more realistic when compared with results of field observation and previous studies.

Conclusion

This study successfully employed a GIS-based RUSLE model to evaluate the soil erosion induced by LULC change in Rib watershed, north-western highland Ethiopia. The spatio-temporal variation and distribution of LULC change and soil erosion were predicted over the period 1986–2016. The highest mean annual soil loss was observed in croplands followed by grasslands, which are highly subjected to frequent human intervention. The annual soil loss for the entire watershed ranged from 0 to 236.5 t ha−1 year−1 in 1986, with an average annual loss estimated at 40 t ha−1 year−1 and 0–807 t ha−1 year−1 in 2016, with an average annual loss estimated at 68 t ha−1 year−1. About 70% of the watershed suffered from moderate to severe soil loss rates in 2016, and only the remaining 30% of the watershed had undergone slight and very slight soil erosion categories. At present, the eastern, south-eastern and some northern parts of the watershed are the areas with the most serious soil erosion problem. Therefore, these areas can be considered as high-priority areas for soil and water conservation activities to reduce soil losses. Since croplands were the dominant LULC type with severe erosion problem, plot level agronomic and biological soil conservation practices could help minimize the problem. Further studies by using the same approach but high-resolution satellite data, process-based models, and sediment yield could also suggest more reliable management methods in the watershed.