Soil erosion, a widespread form of soil degradation, is one of the most severe threats to the terrestrial ecosystems in the world (Pimental, Harvey, & Resosudarmo, 1995). It is directly related to decreased agriculture productivity and water pollution, and it has many negative effects on nature, such as degradation of soil structure, depletion of soil fertility, reducing the effective rooting depth, and ruining the most fundamental of all natural resources (Fitton, Saffouri, & Blair, 1995; Lal & Bruce, 1999; Nearing, 2005). The Chinese Loess Plateau is the most heavily eroded area in the world (Fu, 1989), and the soil erosion modulus with 5000– ۱۰,۰۰۰ mg km
۲ per year were larger than other areas (Chen, Wang, Fu, & Qiu, 2001). Since the 1950s, to control severe soil erosion, improve agriculture production and reduce sediment loads in the Yellow River, there has been implemented a lot of soil and water conservation projects in the Loess Plateau catchments. In order to promote sustainable development in the Loess Plateau, the Grain for Green project was implemented in 1999. These conservation efforts, including terracing, afforestation and vegetation restoration, reduced the sediment yield from hill slopes and sediment delivery to rivers by increasing hydrologic surface roughness, and the vegetation coverage gradually increased (Chen, Ma, & Zhang, 2016; Yan, Zhang, Yan, & Zhao, 2016). Generally, the more the vegetation restoration implemented in the region, and the lower the soil erosion modulus (Ritsema, 2003). Soil erosion caused by water has been assessed by some developed models, including the Water Erosion Prediction Project (WEPP), Universal Soil Loss Equation (USLE), the Revised Universal Soil Loss Equation (RUSLE), and the Erosion Productivity Impact Calculator (EPIC) (Flanagan & Laflen, 1997; Renard, Foster, & Weesies, 1997; Williams, Renard, & Dyke, 1983; Wischmeier & Smith, 1978). The USLE model considers most of the factors to assess long term soil erosion from interrill and rill areas (Wischmeier & Smith, 1978). Compared to the USLE, the RUSLE has become available during the last 40 years with the characteristics of easily to parameterize and requiring less data (Lee, 2004). The combination of erosion models with GIS was an effective method to assess the temporal and spatial distribution of erosion (Mitasova, Hofierka, Zlocha, & Iverson, 1996; Qin & Zhu, 2009). Ferro and Porto (2000) used the USLE combined with the travel time concept and assessed the erosion in a watershed. There have many researches to investigate soil erosion either in plot or watershed scale in the Loess Plateau of China (Fu, Wang, & Lu, 2009; Sun, Shao, & Liu, 2013; Zhao, Kondolf, Mu, & Han, 2016). Fu, Zhao, and Chen (2005), based on the RUSLE to assess soil loss in the Yanhe watershed of the Loess Plateau, found that the annual average soil erosion was 14,458 mg km
۲ per year. Sun, Shao, and Liu (2014) used the RUSLE to analyze the influence of land cover change on soil erosion in the Loess Plateau from 2000 to 2010. The results found that the steadily increased vegetation cover lead to gradually decreased soil erosion rates from 2000 to 2010. Zhao, Kondolf, Mu, and Han (2017) found that the sediment yield also decreased in the Huangfuchuan catchment of the Loess Plateau from 1990 to 2006. A series of previous studies demonstrated that these measures have obtained success to some extent. The Beiluo River, located in the center of the Loess Plateau, is representative of the changes in vegetation coverage resulting from conservation projects.