Soil erosion is one of the most severe problems facing environments and has increased throughout the 20th century. Soil erodibility (K-factor) is one of the important indicators of land degradation, and many models have been used to estimate K values. Although soil erodibility has been estimated, the comparison of different models and their usage at a regional scale and, in particular, for different land use types, need more research. Four of the most widely distributed land use types were selected to analyze, including introduced and natural grassland, as well as introduced and natural shrubland. Soil particle size, soil organic matter and other relevant soil properties were measured to estimate soil erodibility in the Loess Plateau. The results show that: (1) the erosion productivity impact calculator (EPIC) model and SHIRAZI model are both suitable for the Loess Plateau, while the SHIRAZI model has the advantage of fewer parameters; (2) introduced grassland has better ability to protect both the 0⁻5 cm soils and 5⁻20 cm soils, while the differences between introduced and natural shrubland are not obvious at a catchment scale; (3) the K values of introduced grassland, natural grassland, introduced shrubland and natural shrubland in the 0⁻5 cm layer vary from 0.008 to 0.037, 0.031 to 0.046, 0.012 to 0.041 and 0.008 to 0.045 (t·hm²·h/(MJ·mm·hm²)), while the values vary from 0.009 to 0.039, 0.032 to 0.046, 0.012 to 0.042 and 0.008 to 0.048 (t·hm²·h/(MJ·mm·hm²)) in the 5⁻20 cm layer. The areas with a mean multiyear precipitation of 370⁻440 mm are the most important places for vegetation restoration construction management at a regional scale. A comprehensive balance between water conservation and soil conservation is needed and important when selecting the species used to vegetation restoration. This study provides suggestions for ecological restoration and provides a case study for the estimate of soil erodibility in arid and semiarid areas.
Keywords: Loess Plateau; grassland; models; shrubland; soil erodibility.