Soil erodibility and its influencing factors on the Loess Plateau of China: a case study in the Ansai watershed

Abstract. The objectives of this work were to identify the best possible method to estimate soil erodibility (K) and understand the influencing factors of soil erodibility. In this study, 151 soil samples were collected during soil surveys in the Ansai watershed of the Loess Plateau of China. The K values were estimated by five methods: erosion-productivity impact model (EPIC), nomograph equation (NOMO), modified nomograph equation (M-NOMO), Torri model and Shirazi model. The main conclusions of this paper are (1) K values in the Ansai watershed ranged between 0.009 and 0.092 t  ⋅  hm²  ⋅  h/(MJ  ⋅  mm  ⋅  hm²), and the maximum values were 1.9–7.3 times larger than the corresponding minimum values, and the Shirazi and Torri models were considered the optimal models for the Ansai watershed. (2) Different land use types had different levels of importance; the principal components (PCs) accounted for 100 % (native grassland), 48.88 % (sea buckthorn), 62.05 % (Caragana korshinskii), and 53.61 % (pasture grassland) of the variance in soil erodibility. (3) The correlations between soil erodibility and the selected environmental variables differed among different vegetation types. For native grasslands, soil erodibility had significant correlations with terrain factors. For the most artificially managed vegetation types (e.g., apple orchards) and artificially restored vegetation types (e.g., sea buckthorn), soil erodibility had significant correlations with the growing conditions of vegetation. Soil erodibility had indirect relationships with both environmental factors (e.g., elevation and slope) and human activities, which potentially altered soil erodibility.