Figure 8. Statistical comparison of soil erosion by slope between previous study(Huang,2020) and our study in 2015(a) Statistical correlation analysis of soil erosion modulus by slope between the previous study(Huang,2020) and our study in 2015(b).

Relationship between land use and soil erosion

Our results showed that for the 2010–2020 LULCC, the multi-year average soil erosion rate for different land uses significantly differed, with cropland (55.1 t ha–1 yr–1) > grassland (8.93 t ha–1yr–1) > building (1.6 t ha–1 yr–1) > forest land (1.65 t ha–1 yr–1), similar to the findings of Zhao et al. (2022) and Han et al. (2020). The ‘Returning Farmland to Forest Project’ program implemented in 1999 significantly increased the vegetation coverage rate, increasing the grassland and forestland area from 83.1% in 2010 to 90.7% in 2020. The Loess Plateau has recently experienced large-scale cropland abandonment due to rural population movements, cropland quality, and economic benefits. At the end of 2016, a survey of 235 villages in China revealed that 78.3% had abandoned up to 14.32% of their cultivated land. Soils on plots with lower vegetation cover were being disturbed more frequently and intensely by tillage, accelerating erosion (García-Ruiz, 2010; Zuazo and Pleguezuelo, 2009). The abandoned cultivated land resulted in increased biocrust and vegetation cover in interplant patches, increasing soil infiltration and reducing the generation of erosive runoff (Chirino et al., 2006; Yuan et al., 2020), and the vegetation root systems reduced soil erodibility, protecting the soil from raindrop erosion and runoff transport (Chen et al., 2021; Chuenchum et al., 2020). Therefore, soil erosion was curbed in the Jiuyuangou watershed from 2010 to 2015.
However, the increase in forest/grassland area and vegetation coverage from 2015 to 2020 increased the soil erosion rate in the watershed from 2015 to 2020, primarily due to more frequent extreme rainfall events after 2015. The Loess Plateau has an extremely uneven distribution of annual rainfall. The rainfall erosivity in the basin from June to September in 2010, 2015, and 2020 accounted for 85%, 50%, and 88% of annual rainfall erosivity, respectively. At the same time, frequent heavy rainfall events occurred in the basin after 2015 (e.g., July 26, 2017 (212.2 mm), August 7, 2018 (151.3 mm), August 7, 2019 (150 mm), and August 4–6, 2020 (80.3 mm)). Flooding season precipitation in the Jiuyuangou watershed has become the main precipitation for the year, with the erosion and sediment production caused by rainstorm erosion an important feature of this area. It is far sufficient to rely on surface micromorphological changes from vegetation restoration to manage rainstorm erosion (Han et al., 2020; Yang et al., 2011). The current land use distribution pattern in the hilly and gully region of the Loess Plateau in northern Shaanxi cannot prevent soil erosion under extreme rainstorm conditions. Optimizing the distribution of land use types in the catchment area should be the focus of soil erosion control (Wang et al., 2020). The frequent occurrence of extreme rainfall has affected the consistency of soil erosion changes and LULCC, evidenced by large-scale soil erosion losses in areas with no change in land use patterns. In the northern part of the basin, far from human settlements and construction land, land use and soil erosion followed a common ‘loss’ trend, ‘no change’ occurred in settlements and construction land. The frequent occurrence of extreme rainfall has affected the consistency of soil erosion changes and LULCC, evidenced by large-scale soil erosion losses in areas with no change in land use patterns. In the northern part of the basin, far from human settlements and construction land, land use and soil erosion followed a common ‘loss’ trend, ‘no change’ occurred in settlements and construction land. Therefore, the LULCC and soil erosion changes are spatially correlated in areas near the human activities or settlements, roads, cropland, and valleys with good soil and water conservation.