Introduction

Soil erosion, a serious environmental problem of global concern, leads to land degradation, which seriously affects sustainable agricultural development. The total area of the Loess Plateau suffers from serious soil erosion, with about 472,000 km2 of its 640,000 km2 erosion-prone and about 80% of the cropland area suffers from moderate to severe erosion (Meliho et al., 2019). The Loess Plateau is also a key area for soil and water conservation and ecological security construction in China (Huang, 2021; Zhang, 2016). Soil erosion restricts the sustainable development of the regional economy and society and determines the evolution of the Yellow River and its downstream safety (Fan et al., 2012). Soil erosion causes irreversible soil degradation, leading to topsoil losses and the migration of sediments (and associated nutrients) to water bodies, decreasing the available farmland area (Molina et al., 2012). Excessive sediments and pollutants lead to river siltation and ecosystem degradation, endangering regional food security and ecological environment quality. Thus, there is a need to understand the soil erosion status of the watershed through soil erosion assessments and mapping and determine the key areas for controlling soil erosion on the Loess Plateau. The general loss equation of soil erosion can calculate and analyze the distribution of soil erosion in watersheds. Various studies have used more than 30 soil erosion models to assess soil erosion rates on the Loess Plateau, with the common models usually divided into empirical models [MUSLE (Luo et al., 2015), RUSLE (Feng et al., 2010)] and physical models [WEPP (Han et al., 2016; Shen et al., 2009), SWAT (Shen et al., 2009), WATEM/SEDEM (Feng et al., 2010), Yang’s model (Yang et al., 2011)].
The soil erosion models mentioned above are generally not suitable for the Loess Plateau because: (1) some are based on the revised general soil loss equation, RUSLE, which is not suitable for the unique terrain of the Loess Plateau with vertical and horizontal ravines and ignores the impact of important soil and water conservation engineering measures on erosion (MUSLE); (2) the function model obtained from experimental data in a certain area can not be used in other areas (Yang’s model); (3) physical models require high-quality parameters and verification data, which is difficult to obtain in areas where little or poor quality data is available (WEPP, SWAT). Liu et al. (2002) considered the characteristics of soil erosion terrain on the Loess Plateau and proposed an LS(Slope length and slope factor) algorithm suitable for the region. Combined with the influence of human activities, they proposed incorporating biological measures (B factor), soil and water conservation engineering measures (E factor), and soil and water conservation tillage measures (T factor) to improve the applicability of the CSLE model on the Loess Plateau. This model is used widely in the region due to its relatively simple structure, high accuracy, and easy access to parameter factors, enabling rapid and comprehensive soil erosion assessments at different temporal and spatial scales. Shi et al. (2018) modified the CSLE model using runoff data from three different watersheds on the Loess Plateau over three different periods (1956–1959, 1973–1980, 2010–2013) to predict soil losses accurately at the farmland scale. Duan et al. (2020) used the CSLE model and 0.5 m Worldview satellite images, combined with soil and water conservation engineering practices to estimate soil looses in China and the erosion modulus and soil erosion intensity in Yunnan Province, a mountainous area in southwest China. Liu and Liu (2020) used satellite remote sensing image interpretation to understand land use changes in a small watershed in southern China from 1989 to 2015 and the CSLE model to calculate the contribution rate of soil erosion change and gully erosion to sediment yield.
Since the 1970s, numerous soil and water conservation projects have been undertaken on the Loess Plateau, such as the Converting Farmland to Forest Project (1999–2010), the construction of check dams, and improvements to sloping land (Fang et al., 1993). The continuous development of soil and water conservation and governance has significantly improved soil erosion and the ecological environment on the Loess Plateau, changing the spatial pattern of land use (Zhao et al., 2022). Land use and management affect runoff and sediment transport by changing the surface morphology, the dominant factor associated with soil erosion (Fu et al., 2020). The hilly and gully area of the Loess Plateau is the most severely eroded area. Extreme rainfall events and changes in land use patterns in recent years have setback the soil erosion improvements (Li et al., 2022). Using soil erosion models to assess soil erosion changes under the LULCC will help understand the governance effects of previous engineering measures and the evolution process of soil erosion to formulate future spatial pertinence and sustainable land management and restoration strategies. However, few studies have assessed soil erosion in the background of LULCC using CSLE in the hilly and gully area of the Loess Plateau. Therefore, considering the superiority of the CSLE model and changes in land use in the recent ten years (2010–2020), we evaluated: (1) temporal and spatial changes in soil erosion factors in the CSLE model for the Jiuyuangou watershed, a typical watershed in the hilly and gully region of the Loess Plateau, from 2010 to 2020; (2) temporal and spatial dynamic evolution of soil erosion under land use change in the study area from 2010 to 2020; (3) the spatial correlations between the soil erosion change and LULCC. our study will provide a scientific reference for land use management and ecological restoration projects in the basin.