Assessing impacts of floods disaster on soil erosion risk based on the RUSLE-GloSEM approach in western Iran

Floods cause great damage to ecosystems and are among the main agents of soil erosion. Given the importance of soils for the functioning of ecosystems and development and improvement of bio-economic conditions, the risk and rate of soil erosion was assessed using the RUSLE model in Iran’s Lorestan province before and after a period of major floods in late 2018 and early 2019. Furthermore, soil erosion was calculated for current and future conditions based on the Global Soil Erosion Modeling Database (GloSEM). Through the analysis of rainfall events, as the most important agent of soil erosion, the average R-factors for the period before and after flooding were 58.87 and 157.6 MJ mm ha− 1 h− 1 y− 1, respectively. The results showed that agricultural development and land use change are the main causes of land degradation in the southern and central parts of the study area. The impact of floods was also significant since our evaluations showed that soil erosion increased from 4.12 t ha− 1 yr− 1 before the floods to 10.93 t ha− 1 yr− 1 afterwards. Field surveys using 64 ground control points determined that erodibility varies from 0.17 to 0.49% in the study area. Orchards, farms, rangelands, and forests with moderate or low vegetation cover were the most vulnerable land uses to soil erosion. The results of GloSEM modeling revealed that climate change is the main cause of change in the rate of soil erosion. The results also established that when the combined effects of land use change and climate change are taken into account, soil erosion has increased under SSP1-RCP2.6, SSP2-RCP4.5, and SSP5-RCP8.5 scenarios, so that about 80% of the region has experienced moderate to very high erosion. Therefore, both natural factors (e.g. climate change) and human factors (e.g. agricultural development, population growth, and overgrazing) are among the drivers of soil erosion in the study area.


Introduction
Environmental change influences ecosystems through changes in ecosystem components. These changes can in turn impact natural disasters (Sholihah et al. 2020). While the damage caused by most natural disasters is temporally immediate, floods often have more lasting consequences such as property damage and reduced crop yields. Additionally, floodwaters can become contaminated and consequently cause diseases (Arnell and Gosling 2016;Adikari and Yoshitani 2009). Based on the reports by the Emergency Events Database (EM-DAT), Asia is particularly vulnerable to natural disasters such as floods; in the period from 2000 to 2019, 7,348 major disaster events were recorded in Asia, claiming 1.23 million lives, affecting 4.2 billion people, and costing approximately $2.97 trillion dollars (UNDRR 2020; Adeel et al. 2020). More recently, climate change has become a major factor affecting floods by concentrating total annual rainfall in increasingly sporadic and intense rainstorm events (Cramer et al. 2014). Floods can also cause damage to the environment through affecting ecosystem components such as soil through soil erosion and degradation (Nepal et al. 2021).
Although soil erosion is only one of the manifestations and aspects of land degradation, it can be considered the most critical (Lal 2008). Soil erosion is a complex process with direct and indirect socio-economic and environmental impacts. The type and severity of soil erosion is determined by factors such as landform and geomorphology, physical and chemical properties of soil, hydrography, rainfall regime, land use, land cover, and land management practices (Salvati et al. 2013;Masroor et al. 2022). By reducing soil nutrients, soil erosion increases food production costs (Quinton et al. 2010;Brevik et al. 2020) and, as a result, causes economic and environmental damage (Telles et al. 2011).
Modern societies increasingly depend on cultivated lands, so that more than a third of earth's land area is currently covered by agricultural lands (Foley et al. 2011). 95% of global food production relies on the cultivation of plants and livestock, which is itself dependent on the capacity of soils to sustain living organisms (FAO 2015). In recent decades, the unbalanced development and expansion of agricultural lands have led to a significant increase in soil erosion, especially in developing countries (FAO 2015). Inattention to topography, texture and structure of the soil, and a mismatch between soil and climatic conditions (Pal and Chakrabortty 2019a) and the ecological needs of cultivated crops are among the most important causes of severe soil erosion in agricultural lands (Chandra pal et al. 2021). Therefore, attention should be paid to two key issues in agricultural science to reduce soil erosion: land suitability for crop selection, and optimal cropping patterns (Derakhshan-Babaei et al. 2021;Akbari et al. 2019;Neamatollahi et al. 2017).
Human activities such as agricultural development have been significant drivers of soil and environmental degradation in recent years (Montanarella 2015;Mueller et al. 2013;Zhang et al. 2016;Nga et al. 2018). Therefore, soil erosion assessment has been proposed as a method to respond to these concerns, especially in industrialized countries (Nachtergaele 2004). The latest report by the Food and Agriculture Organization of the United Nations (FAO 2015) on global soil resources highlights the worrying condition of the world's soil resources and emphasizes the threat of soil erosion to the environment on a global scale. The most significant anthropogenic drivers of soil erosion are land use change and climate change (O'Neal et al. 2005).
Past works on soil erosion have shown more interest in soil erosion in arable lands (Boardman and Poesen 2006). At the same time, assessment of soil erosion in natural lands has been limited due to the lack of climatic and soil data on local and global scales (Bosco 1 3 et al. 2015;Akbari et al. 2016b) as long-term assessment of soil erosion requires reliable long-term data (Salvati et al. 2013;Cox et al. 2018). In response, semi-empirical methods have been developed to incorporate existing knowledge to generate actionable results (Borrelli et al. 2020). To estimate soil erosion in agricultural lands, models such as CREAMS, SLEMSA, GUESS, USLE, RMMF, and MUSLE have been developed, while others such as EUROSEM, RUSLE, WEPP, and CORINE have been created for rangelands and watersheds (Panagos et al. 2014;Silva et al. 2012;Tsara et al. 2006;Aiello et al. 2015;Chandra Pal and Shit 2017).
Among these models, the RUSLE model has been applied to estimate present and future soil erosion in many locations around the world due to its simplicity (e.g. Pal and Chakrabortty 2019b;Pal et al. 2021;Aiello et al. 2015;Bhandari et al. 2015;Das et al. 2018;Jiang et al. 2015;Kouli et al. 2009;Samanta et al. 2016). Recently, integrative approaches such as Soil and Water Assessment Tool-Evidential Belief Function-Logistic Regression (SWAT-EBF-LR) (Chakrabortty et al. 2020a) and Geographically Weighted Regression-Artificial Neural Network (GWR-ANN) (Chakrabortty et al. 2020b) have also been employed to estimate and identify soil erosion potential hotspots. Integrating the output of such models into risk-based management can reduce the impact of uncertainties and losses (Sarbazi et al. 2021;Akbari et al. 2016a). In this context, risk assessment allows the preparation of maps of potential and actual risks (Akbari et al. 2022) using a method of risk assessment selected based on the type of risk being studied (i.e. direct or indirect, tangible or intangible) (Merz et al. 2010).
Recent advances in remote sensing (RS) have made earth observation data more available and more accurate, and have expanded our data processing capabilities, ushering in an array of new indicators to capture important variables in erosion risk assessment such as vegetation (Liu et al. 2021). In this regard, models that consider the impact of climate change and land use parameters through incorporating RS products can assist us with evaluating the risk of soil erosion and therefore contribute to more accurate and more reliable risk assessment. The Global Soil Erosion Modeling Database (GloSEM) utilizes the RUSLE model and offers a more comprehensive modeling framework for estimating future global soil erosion scenarios (Borrelli et al. 2020).
Although there are examples of risk assessment for flooding risk in Iran, the literature on the topic is limited in terms of scope (both temporal and geographical) and the methods utilized to perform the analysis. In this study we address this gap by performing risk assessment using a recent case of flooding in western Iran. Iran was affected by three strong weather systems from November 27, 2018 to April 12, 2019. During this period, western and southwestern Iran experienced unprecedented flooding, leading to immense costs in terms of human lives and economic damage. We used these events to assess soil erosion in one of the most-severely affected regions in the country and performed soil erosion assessment under current and projected conditions using the Global Soil Erosion Modeling Database (GloSEM).

Study area
Lorestan province is located in the west of Iran and has an area of 28064 km 2 (8.1% of the total area of Iran). The province is located between 46° 51′ and 50° 01′ E, and 32° 37′ and 37° 34′ N. Based on the latest national census, the population of the province was 1,760,649 people in 2015, with the largest population belonging to Khorramabad county, and the smallest to Romeshkan county. Natural and agricultural land uses in the province include forests (43.34%), rangelands (30.12%), deserts (0.41%), agricultural lands (18.57%), and orchards (7.55%) (Nabati et al. 2020). In the past decade, the area of agricultural lands in Lorestan province has increased remarkably, increasing from 3.2% of total national agricultural land in 2011 to 4.7% in 2020 (Iranian Ministry of Agriculture statistical yearbook 2020). The province overlaps with the watersheds of two major rivers in Iran (Karkheh and Dez), and receives more precipitation than its neighbors due to its location and elevation. The province is susceptible to floods, especially since warm rainfall in April can accelerate snowmelt and flooding (Sharafi and Noorollahi 2020). Data shows that 436 spots in the province are vulnerable to floods (Hydrological Report, Regional Water Company 2019). Figure 1 shows the geographical location of the study area and its flood-prone regions. Tables 1 and 2 summarize the topographical, meteorological, and hydrological characteristics of the province.
Lorestan province experienced serious instances of flooding on 27 November 2018, 4-5 December 2018, and 7-8 February 2019, 4-5 April 2019, and 11-12 April 2019. The high intensity and large volume (up to 170 mm) of rainfall caused significant damage across the province, especially in Khorramabad and Poldakhtar counties, which are more vulnerable to flooding (Hydrological Report, Regional Water Company 2019). The extreme precipitation events flooded most of Poldakhtar county and parts of Khorramabad, Delfan, and Kuhdasht and damaged many urban and rural residential and commercial units, as well as impacting infrastructure including main roads, interurban and rural road structures, water supply lines, power lines, the telecommunication network, and oil and petrochemical facilities (Hydrological Report, Regional Water Company 2019). A warming climate and changing precipitation patterns in the study area have contributed to more severe flooding Fig. 1 The geographic location of the study area and flood-prone regions as warm rainfall in spring leads to the melting of the snow accumulated in mountainous areas. The role of snow as a contributing factor to extreme hydrological events and natural hazards has been established in past research (Akbari et al. 2017). Furthermore, vegetation cover in the province has been under increasing pressure as a result of human exploitation (e.g. wood harvesting and grazing). Finally, the soils in the study area include inceptisols, vertisol, and entisols (Hydrological Report, Regional Water Company 2019), all of which are very sensitive to erosion (Soil Taxonomy 1975).

Database
Meteorological data were accessed from meteorological stations in the study area for the period before the floods (11/01/2017 to 30/04/2018) and the flooding period (11/01/2018 to 30/04/2019). The time intervals were selected to include the rainy season in the study area. The soil map of the study area (at 1:100,000 scale) was initially used to investigate soil characteristics. A field survey using 64 ground control points (GCPs) in different land units was then conducted to update the soil map. A digital elevation model (DEM) provided by Iran's National Cartographic Center was used to calculate topographic variables. The elevation model had a cell size of 30 × 30 m and was prepared based on 1:50,000 topographic maps and 1:50,000 contour maps with a 20-m elevation difference between contour lines.
Sentinel-2 images and a mosaic of 9 frames for the period before and after the floods (11/01/2017 to 30/04/2018 and 11/01/2018 to 30/04/2019, respectively) were used to evaluate vegetation cover. Sentinel-2 is an earth observation project developed by the European Space Agency as part of the Copernicus program and includes two identical satellites (Sentinel-2E and Sentinel-2B). Sentinel 2B is capable of multispectral imaging in 13 bands including the visible, near-infrared, and short-wave infrared spectra. Images are recorded at 5-day intervals and are available to the public for free. In this research, images captured by Sentinel 2B on April 30th, 2018, and April 30th, 2019 were used. The list of images is presented in Table 3. Future changes in soil erosion were predicted using the Global Soil Erosion Modelling Database (GloSEM). GloSEM was developed based on the RUSLE model to provide a more comprehensive framework for global assessment of soil erosion under different scenarios (Borrelli et al. 2020). This database uses the latest climate change and land use change scenarios (Shared Socio-economic Pathway-Representative Concentration Pathway, SSP-RCP) to simulate erosion in the future. GloSEM takes advantage of the land use change scenarios for 2070 extracted from the Land Use Harmonization project and the climate change scenarios developed by the European Union's Joint Research Centre. In this database, uncertainties in predictions are assessed and validated using Monte Carlo-Markov Chain (MCMC) methods.

RUSLE model
In this study, soil erosion following flooding in Lorestan province was evaluated using the Revised Universal Soil Loss Equation (RUSLE) model in ArcGIS 10.8. Integration of the RUSLE model in GIS allows for the estimation of sediment yields and identification of vulnerable areas (Pal and Chakrabortty 2019b). RUSLE has shown strong performance in various settings ranging from mountainous terrain to tropical forests (Samanta et al. 2016;Sadeghi et al. 2014). The model takes advantage of a rainfall and runoff erosivity factor, which improves model implementation compared to its predecessor (USLE). The revised version of the global soil erosion equation is used to evaluate and predict erosion, and to develop conservation plans based on six factors (Eq. 1).
(1) SE = R * K * L * S * C * P where SE is average annual soil erosion per unit of area (t ha − 1 y − 1 ), R is the erosivity factor (MJ mm ha − 1 h − 1 y − 1 ), K is the erodibility factor (t ha MJ − 1 mm − 1 ), L is the slope length factor, S is the slope steepness factor, C is the cover management factor, and P is the support practices factor (note: L, S, C, P are dimensionless) . The flow diagram for calculating soil erosion before flooding (01/11/2017 to 30/04/2018) and after flooding (01/11/2018 to 30/04/2019) is depicted in Fig. 2.
Information layers which were subject to change before and after flooding such as NDVI and erosivity (R) were calculated separately for each period. Other factors such as L, P and K were considered unchanged. Figure 3 shows the map of variables used in the RUSLE model.

Rainfall and runoff erosivity (R)
Rainfall is a major driving force of soil erosion since raindrops (through splash erosion), and runoff flow erode the surface (Pal and Chakrabortty 2019b). The annual R-factor is computed based on complete long-term records of storm kinetic energy and maximum thirty-minute intensity, known as EI 30 (Morgan 2005;Renard et al. 1997). R was obtained based on annual average total erosion of rainfall events using Eq. 2 (Arnoldus 1980).
where R is rainfall erosivity (MJ mm ha − 1 h − 1 year − 1 ), P i is monthly rainfall (mm), and P is annual rainfall (mm). Rainfall erosivity values were interpolated in ArcGIS 10.8 using kriging/cokriging.

Soil erodibility (K)
Soil erodibility (K) reflects the inherent vulnerability of soils to water erosion (Renard et al. 1997). K is defined as the "average annual soil loss per unit of rainfall erosivity under  Renard et al. 1997) have provided a formula for calculating K based on soil texture and structure, organic matter content, and profile permeability (Pal and Chakrabortty 2019a). Thus, soil erodibility was determined based on sand, clay, silt, and organic matter content according to Table 4. To convert soil erodibility to metric units, the figures in Table 4 were multiplied by 0.1317 to give K in terms of t h MJ − 1 mm − 1 .

Fig. 3
The data used in the RUSLE model

Slope length (L)
Slope length (L) affects sheet, rill, and inter-rill erosion (Wischmeier and Smith 1978). L is the horizontal distance from the upstream source of flow to the point where the slope changes significantly. Soil erosion increases with increasing slope length (Renard et al. 1997;Krishna Bahadur 2009). More advanced methods for topographically complex units employ approaches which incorporate contributing area and flow accumulation (Desmet and Govers 1996). The slope map was obtained based on the DEM raster using the spatial analyst toolbox in ArcGIS (Table 5). To calculate L, Eq. 3 was applied to the DEM (Moore and Burch 1986): where L is slope length, Flow accumulation is the accumulated slope area contributing to flow in a given cell, and SineSlope is slope in degrees.

Cover management (C)
The cover management factor (C) is the ratio of the amount of soil lost from cultivated land to the amount of soil lost from the same land under continuous fallow without vegetation (El Jazouli et al. 2017;Alkharabsheh et al. 2013). An increase in C leads to an increase in exposed soil, and thus an increase in potential soil loss (El Garouani et al. 2008). C is calculated using empirical equations based on field measurements of ground cover (Wischmeier and Smith 1978). The most appropriate method for determining vegetation cover is to employ the normalized vegetation difference index (NDVI), which is obtained using satellite imagery (Renard et al. 1997). NDVI is an indicator of the energy reflected by the earth (Kouli et al. 2009), and ranges between − 1 and + 1. In this study, NDVI was calculated based on satellite images according to Eq. 4: where NIR is the near-infrared band, and IR is the red optical band. Due to the Mediterranean climate of Iran, most of the rainfall occurs in autumn and winter; therefore, the formula presented by Lin et al. (2002) provides a more appropriate   Wischmeier and Smith (1978) Slope (%) P < 3 0.6 3-6 0.5 6-9 0.5 9-12 0.6 12-15 0.7 15-20 0.8 20-25 0.9 > 25 1 estimate of the status of vegetation and the cover management factor (C). In this method, C is obtained based on Eq. 5: where α and β are dimensionless parameters, and NDVI is the normalized difference vegetation index (Chakrabortty et al. 2020c).

Support practice (P)
The support practices or conservation practices factor (P) is calculated as the ratio of soil loss after implementation of support practices to soil loss in the absence of support practices (Renard et al. 1997;Pal and Chakrabortty 2019a). In large areas such as our study area, P is estimated based on slope since areas with similar slopes often experience the same level of protection from erosion. Accordingly, we used the table provided by Wischmeier and Smith (1978) to derive P.

Prediction of the combined impacts of climate and land use changes on soil erosion
We compared the predictions made by GloSEM with our regional evaluation of soil erosion in Lorestan province. The impacts of climate change on rainfall erosivity (R) were considered under RCP2.6, RCP4.5, and RCP8.5, and the impacts of land use change on support practices (P) and cover management (C) were considered under SSP1-RCP2.6, SSP2-RCP4.5, and SSP5-RCP8.5. The combined effects of these changes were investigated in the framework of the RUSLE model. SSP1 represents the most sustainable trajectory, with minimal pollution or environmental degradation. In this scenario, land use is highly regulated and controlled, food is traded on an international level, and international collaboration against climate change begins early (after 2020). SSP2 assumes a continuation of historic social, economic, and technological trends in the future. Land use change is not regulated and an international collaboration to control climate change is delayed until 2040. In SSP5, the world seeks sustainable development through competitive markets, innovation, and collaboration for rapid development of technology and human capital under the guidance of emerging industrial economies. Land use change is unregulated and unplanned although deforestation might slow down in the later years (Memarian and Akbari 2021).

Results
RUSLE applications usually estimate soil loss at the annual timescale (Renard et al. 1997;Sadeghi et al. 2014), as a result, our findings are reported on a per year basis despite the selected periods being shorter. The data layers (maps) extracted for R, K, L, C, and P were integrated using the raster calculator in ArcGIS 10.8 to quantify, evaluate, and generate the maps of soil erosion risk and severity for Lorestan province. The values of R, K, L, C, and P are shown in Table 6.

Impact of rainfall erosivity on soil erosion
Meteorological data for the region showed that the average R-factor for the period before and after flooding was 58.87 and 157.6 MJ mm ha − 1 h − 1 y − 1 , respectively. Rainfall intensity, duration, and frequency are the main determinants of runoff generation. All three factors were extremely high for the precipitation that occurred in Lorestan province, especially on April 12, 2019 (Table 7). The value of R more than doubled at most stations following the floods, with the exception of Aligudarz. The largest absolute increase in R occurred in Khoramabad (from 40.63 to 158.8 MJ mm ha − 1 h − 1 y − 1 ) and the smallest in Aligudarz (from 71.56 to 139.6 MJ mm ha − 1 h − 1 y − 1 ). In terms of relative increases in R, Khoramabad ranked first (290% increase), followed by Kuhdasht (218%), and Borujerd (205%). Maps obtained from the analysis of meteorological data showed that the highest value of R belonged to the southern parts of the province such as Poldakhtar county and the northern and western parts such as Dorud, Khorramabad, and Delfan counties (see Fig. 2). Official reports by the local departments and organizations also confirm this finding (Hydrological Report, Regional Water Company 2019). These areas are high-risk flood zones because of severe land use changes (such as Poldokhtar) or due to being located in mountainous terrain (such as Dorud, Khorramabad, and Delfan).

Impact of soil erodibility on soil erosion
Soil erodibility was obtained based on the percentage of clay, silt, sand, and organic matter in the soil profile according to the soil map and the data collected at 64 ground control points ( Table 4). The results show that erodibility varies from 0.17 to 0.49. The highest figures were observed in orchards, agricultural lands (i.e. rainfed agricultural lands in the southern parts of the region), and rangelands and forests with moderate and low vegetation cover (i.e. piedmont plains of the central region). Khorramdareh county, with the largest area of rainfed and irrigated arable lands, and Poldakhtar county with the largest area of rangelands and forests with moderate and low vegetation cover, face the highest rates of erosion. The areas between Borujerd and Dorud and between Azna and Aligudarz had relatively small K values.

Impact of topography on soil erosion
The value of the slope length factor (L) varied from 0 to 0.27. About 71% of the study area is mountainous and has a slope greater than 25%. This steep mountainous topography has played an important role in shortening the time of concentration of runoff (Tc) and flood generation. In addition, the prevalence of fine-grained and impermeable formations in the region produces large amounts of runoff after each rainfall (Hydrological Report, Regional Water Company 2019). High and mountainous areas are generally located in the northern and eastern parts and piedmont plains and lowland areas are found in the southern and western parts. The slope map also showed variations in slope ranging from 1 to 48%. The average slope of the region is about 19%, which indicates that the region is mountainous. Under rainy conditions, steep slopes are prone to landslides and increased soil erosion.

Impact of land use on erosion
NDVI was calculated based on the analysis of Sentinel-2B satellite images for the period before and after the floods. The results showed that vegetation cover was moderate in the eastern and northern parts (mountainous areas with moderate density pastures and forests) and low in the western and southern parts (low-density pastures with seasonal agricultural lands). The large value of the cover management factor (C) in the study area indicates a high risk of soil erosion. The map obtained for the support practices factor (P) showed that the highest values belong to areas with steep slopes. In the study area, P varied from 0.5 to 1, with the lower values associated with the southern parts such as Poldakhtar, Dasht mountain, Delfan, and Selseleh, which are mostly located on plains and foothills.

Spatial distribution of soil erosion
Analysis of the spatial distribution of soil erosion showed that the southern parts (such as Poldakhtar) and central and eastern parts (such as Khorramabad and Aligudarz) experienced significant soil erosion. The map obtained from the combination of different soil erosion factors in the RUSLE model (Fig. 4) showed that the study area was affected by different factors before and after the floods, but changes in climatic parameters played an important role in both periods. We also found that the floods occuring between 01/11/2018 and 30/04/2019 substantially increased soil erosion. Soil erosion rate in the pre-flood period was 4.12 t ha − 1 yr − 1 , but increased to 10.93 t ha − 1 yr − 1 after the floods (Table 8). While the majority of the study area experienced low to moderate erosion before the floods, more than a third (about 36%) of the study area faced moderate to high erosion post-flood.

Predicting future soil erosion
The projections provided by GloSEM were used to predict future soil erosion in the study area using the RUSLE model. The predictions showed minor deviations from the figures published by the European Union and USDA, indicating the accuracy of our predictions. This agreement also shows that the global estimations available from GloSEM are reliable and valid predictions on the regional scale. Figure 5 presents the area of different soil erosion risk classes under different land use change-climate change scenarios. Our results showed that land use changes in the southern and eastern parts of the study area will lead to reduced organic matter content, reduced soil quality, and changes in the physical and chemical properties of the soil. Along with increased erosivity of rainfall, these factors will increase soil erosion in the area. There was also a marked difference between different scenarios in terms of erosion risk in the easternmost part of the study area, near Aligudarz. The models created for future conditions show that climatic changes are the main drivers of the increase in soil erosion. Combined land use change-climate change simulation indicates significant soil erosion under SSP1-RCP 2.6, SSP2-RCP4.5, and SSP5-RCP8.5 scenarios ( Table 9). The results showed that when the combined effects of land use change and climate change are taken into account, soil 1 3 erosion will increase under SSP1-RCP2.6, SSP2-RCP4.5 and SSP5-RCP8.5 in about 80% of the region. Under SSP1-RCP 2.6, more than 60% of the study area will face high to very high risk of erosion. However, the same two categories will cover more than 70% of the region under SSP5-RCP 8.5. A comparison between SSP1-RCP2.6 and SSP5-RCP8.5 also shows an increase of more than 10% in the areas facing high erosion risk by 2070, and a decline of almost 9% in the share of areas facing moderate risk of erosion.

Discussion
This study evaluated soil erosion in a particularly flood-prone region in western Iran before and after extreme weather events in 2018 and 2019. The results indicate that the unprecedented amount of rainfall increased erosion risk and exposed vast parts of the study area to increased erosion. Past research confirms this observation; although various factors are involved in increasing the intensity of soil erosion, the variation in climatic parameters due to climate change has come to play a very important role in recent years (Borrelli et al. 2020). Before the recent episode of flooding, snow had covered the mountainous areas of the region. Following an increase in air temperature and rainfall, the accumulated snow melted, which intensified the floods. In addition, due to the large amount of rainfall on April 12th, 2019, soils were almost saturated, which reduced the infiltration rate and increased runoff generation (Hydrological Report, Regional Water Company 2019). These findings indicate that the large differences in erosion before and after floods are attributable to the erosivity factor in the RUSLE model. Yao et al. (2016) also report that soil erosion was driven mainly by R and was significantly impacted by K, particularly during rainy periods. Similarly, Polykretis et al. (2020) indicate that R was a more dominant factor in determining the intensity of monthly and seasonal fluctuations in water erosion than the other factors considered in the RUSLE model. The results of a historical study for the period between 1938 and 1988 by Kakembo and Rowntree (2003) also support our findings. The researchers found that gully initiation and intensification coincided with extreme rainfall events in Eastern Cape, South Africa. According to the available official reports, soil salinity is more severe in Poldakhtar county, compared to other counties; the droughts in recent decades have also increased salinity (Akbari et al. 2016a). Increased soil salinity has reduced permeability and increased erosion during floods, which partly explains the severity of flooding and erosion in Poldokhtar county. The cities of Khorramabad, Delfan, and Selseleh in the northern and central parts of the study area have highly erodible soils due to extensive dryland agriculture. This result is supported by findings by Polykretis et al. (2020), reporting that olive groves faced a higher risk of water erosion than other land uses. In contrast, mountainous lands in the eastern and northern parts of the study area were less erodible due to their dense forest cover. Overall, the results showed that human activities and natural factors such as extensive and continuous drought along with erosion-sensitive soils have increased erodibility. Long-term drought has also increased rainfall erosivity in the area. According to a report by the Hydrological Report, Regional Water Company 2019 using standardized precipitation-evapotranspiration index (SPEI), about 78% of the province has experienced moderate to severe drought; while central counties have been less affected (e.g. Selseleh county with only 5% of its area being affected), those in the southern, northern, and eastern parts of the region have been severely impacted (e.g. Poldakhtar county, with 99% of its area affected). Masroor et al. (2022) report a positive relationship between drought indices and soil erosion values. An important part of this correlation was due to the effect of drought on the quantity and quality of vegetation, which itself had an effective role in reducing the energy of raindrops and consequently reducing the amount of soil erosion. The role of the slope length factor (L) was consistent with the changes in the elevation and slope of the area. It should be noted that despite their gentle slope, mildly sloped and flat regions experience more erosion because soils are deeper in such areas, and therefore there is a larger amount of soils that can be eroded. On steep slopes, however, what little soil exists is quickly eroded, i.e. the amount of soil is the limiting factor for how much erosion occurs, although potential erosion might be higher compared to flatter areas. The results obtained by Salvati et al. (2013) indicate the importance of topography in the severity of soil erosion. They specify that increasing the slope in lands with low vegetation density increased the severity of soil erosion. Meshesha et al. (2012) also report a positive correlation between slope and soil loss in the Central Rift Valley in Ethiopia, stating that areas with slopes > 10° made up only 12% of their study area, but contributed disproportionately to soil erosion (31.7% of soil loss). They report that areas with steep slopes experienced erosion rates as high as 130 t ha − 1 yr − 1 . Nabati et al. (2020) utilized climatic, soil, and topographic variables to determine suitable areas for agriculture in Lorestan province. They found that only 19.88% of the province is suitable for irrigated agriculture and 54.2% is suitable for rainfed agriculture. Vegetation cover, along with slope angle and length, are major determinants of soil erosion and sediment production (FAO 2015;El Jazouli et al. 2017). A study by Khademalrasoul and Amerikhah (2021) in the neighboring provinces of Khuzestan and Chaharmahal Bakhtiari similarly highlights the importance of the vegetation cover factor. The authors report that the lowest C factor and water erosion potential were observed in areas with the highest NDVI. A study by Prasannakumar et al. (2011) in the Siruvani watershed in southern India similarly attributes the increase in soil erosion to shifting cultivation patterns, and forest degradation. Moreover, Chinnasamy et al. (2020) indicated that floods and the associated erosion were due to a combination of rainfall events and changes in land use and land cover. The changes in land use also affect river morphology and behavior during floods, which in turn affects soil erosion. The results of field surveys and satellite imagery show that the main river beds have been widened and displaced and that the direction of flow has been altered. The reason for the displacements is the intensity of floods, loose and erodible alluvial sediments, and the interventions of rural communities in the form of expansion of agricultural lands on river banks. Shifting and widening of river beds has also caused the erosion of agricultural lands.
According to the available reports and field studies, agricultural development and land use changes are the main causes of land degradation (Memarian and Akbari 2021;Derakhshan-Babaei et al. 2021;Nga et al. 2018;Zhang et al. 2016) in the southern and central parts of the study area. These results are consistent with that of Memarian and Akbari (2021) on the impacts of climate change on soil erosion. According to a FAO report (2015), both natural factors such as climate change, and anthropogenic factors such as agricultural development, population growth, and overgrazing, intensify erosion. The pre-flood and post-flood periods in our study area clearly show the effects of both categories of factors on causing and intensifying erosion. Alkharabsheh et al. (2013) evaluated the impact of land cover change on soil erosion hazard in northern Jordan during a 17-year interval and found that the main driver of soil erosion in the region was land use change. A study by Haregeweyn et al. (2017) similarly implicates a combination of natural and anthropogenic factors in soil erosion. The researchers state that variations in agricultural practices, slope, and population density were strongly linked with variations in soil erosion.
Although future changes in land use can impact soil erosion by changing the area of agricultural lands in either direction, the fact that the global climate is moving towards more extreme events suggests that soil erosion will become more severe in the future. A comparison of the scenarios showed that although land use change can impact global soil erosion (especially through increasing or decreasing the area of agricultural lands), the shift in global climate toward more extreme hydrological cycles will act as the main driver of the increase in soil erosion. Similar to our findings, Pal and Chakrabortty (2019b) predicted increases in soil erosion in the future (from 5 to 15 year return periods) in the Dwarkeswar River basin in India. An increasing trend in average annual soil erosion was observed under RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5, with the lowest and highest predictions belonging to RCP 2.6 and RCP 8.5, respectively. A comparable result is reported by O'Neal et al. (2005) in their study on the impact of climate change in the Midwestern United States. They predicted significant increases in runoff (10-310%) and soil loss (33-274%) in 2040-2059 relative to 1990-1999. However, the decisions made by land managers and environmental policymakers will be highly impactful on soil erosion. Despite the considerable soil erosion during floods, these conditions could evolve into an environmental catastrophe by 2100 according to land use change and climate change scenarios if adequate measures are not adopted by the responsible entities (Chandra Pal et al. 2021;Memarian and Akbari 2021).

Conclusion
This study assessed the impacts of flood disasters and determined areas facing possible or potential soil erosion risk based on the RUSLE-GloSEM approach in western Iran in a GIS environment. Although various factors are involved in increasing the intensity of soil erosion, changes in climatic parameters, mainly due to climate change, play a very important role. In the study area, rainfall erosivity caused significant damage both before and after the floods. The highest amount of rainfall erosivity belonged to the southern parts of the study area such as Poldakhtar county (mainly due to land-use change, as most of the population is directly and indirectly involved with agricultural production) and the northern and western parts such as Dorud, Khorramabad, and Delfan counties (due to mountainous terrain). These areas constitute the high-risk flood zones in Lorestan province. Before the recent episode of flooding, snow had covered the mountains in the region. Following an increase in air temperature and rainfall, the accumulated snow melted, which intensified the floods. In addition, due to the large amount of rainfall on April 12th, 2019, soils were almost saturated, which reduced the infiltration rate and increased runoff generation. Additionally, continuous and long-term droughts in the study area have contributed to erosion. About 78% of the province has experienced moderate to severe droughts although central areas have suffered less severe erosion compared to the rest of the study area. Overall, orchards, farms, rangelands and forests with moderate or low vegetation cover were the most vulnerable land uses to soil erosion. Khorramdareh county, with the largest area of rainfed and irrigated farms, and Poldakhtar county with the largest area of rangelands and forests, faced the highest rates of erosion. Due to the continuous drought in the study area during the past years, vegetation cover has declined across the province, depriving soils of protection against water erosion. Droughts in recent decades have increased soil salinity as well. Soil salinity is more severe in Poldakhtar county (in the southern part of the study area with extensive areas devoted to agriculture). Moreover, droughts have subsequently reduced permeability and increased soil erosion during floods. Analysis of the spatial distribution of soil erosion showed that the southern parts (such as Poldakhtar) and central and eastern parts (such as Khorramabad and Aligudarz) have experienced significant erosion before and after floods.
The results indicated that both natural factors such as climate change, and anthropogenic factors such as agricultural development, population growth, and overgrazing have intensified erosion. Studies conducted in different regions of the world have shown that variations in agricultural practices, slope, and population density are strongly linked with variations in soil erosion. According to a FAO report (2015) both natural factors such as climate change, and anthropogenic factors such as agricultural development, population growth, and overgrazing intensify erosion. The study area clearly shows the effects of both categories of factors (natural and anthropogenic) on causing and intensifying erosion. The results indicate the importance of topography for the severity of soil erosion. Increasing slope in lands with low vegetation density has increased the severity of soil erosion. Our results also showed that land use changes in the southern and eastern parts of the study area will lead to reduced organic matter content, reduced soil quality, and changes in the physical and chemical properties of the soil. Along with increased erosivity of rainfall, these factors will increase soil erosion in the area.
Climate studies indicated that total rainfall in Lorestan province has decreased, similar to many other regions in Iran. On the other hand, the frequency of maximal rainfall events has increased. These conditions can indicate an increase in heavy and short-term rain storms and shortening of the rainy season in the region, which consequently escalate soil erosion potential. Modeling of future conditions also confirmed that climatic changes along with land use change are the main drivers of the increase in soil erosion. Combined land use change-climate change simulation indicates significant increases in soil erosion under SSP1-RCP2.6, SSP2-RCP4.5, and SSP5-RCP8.5 scenarios. The percentage of areas falling into the high and very high classes of ersosion is expected to increase in step with worsening climate change-land use change scenarios; while 64% of the study area falls into these two classes under SSP1-RCP2.6, SSP5-RCP8.5 predicts that 71% of the study area will face high or very hight erosion by 2070.
What is clear from the outcomes of this research is that floods, due to their role in the destruction of vegetation and soil and as important factors in changing river morphology and slope topography, can more than double the amount of soil erosion in the studied watershed. A wide variety of indices and factors are included in the RUSLE model. Therefore, future research should focus on fine-tuning the model to better represent real-world conditions. Future works should augment previous studies by including other types of erosion, estimating erosion at shorter time scales, and improving the accuracy of the equations used to estimate erosion.
Funding This study was supported by the Faculty of Natural Resources and Environment at Ferdowsi University of Mashhad under Grant No. 50654. Therefore, we thank all those who have helped us in the process. The authors are grateful to the anonymous reviewers for their insightful and helpful comments to improve the manuscript.