Background Rainfall erosivity reflects the impact of rainfall on soil erosion, which is an important factor in USLE (Universal Soil Loss Equation), RUSLE (Revised Universal Soil Loss Equation) and CSLE (Chinese Soil Loss Equation). In-depth study of the characteristics of intra-annual distribution and inter-annual variation of rainfall erosivity is of great significance to improve the accuracy of matching between vegetation cover factor and rainfall erosivity in the corresponding time period, especially for the loess hilly and gully region of Ningxia, where the ecological environment is inherently sensitive and fragile. The few detailed rainfall data of long time series in the loess hilly and gully region of Ningxia, restricted revelation of the characteristics of rainfall erosivity in 24 half-months, and the soil and water control in the loess hilly and gully region of Ningxia demands applicability of models based on daily rainfall data.

Methods Based on the detailed rainfall data of 1 207 rainfall events in Wangwa small watershed in the second subarea of the loess hilly and gully region of Ningxia, the EI₃₀ of each rainfall event and the 24 half-month rainfall erosivities were calculated using the method of EI₃₀ accumulation, and the characteristics of 24 half-months rainfall erosivities were clarified. Furthermore, three models of YIN Shuiqing cold and warm season model, ZHANG Wenbo model and CREAMS model, were used to validate the performance of three models on predicting rainfall erosivity at semi-month and annual scale, respectively.

Results 1) The overall rainfall in Wangwa small watershed was dominated by light rainfall events, with < 10 mm individual rainfall events accounting for 70.51%, and ≥50 mm individual rainfall events accounting for only 1.49%. 2) The annual erosivity was mainly contributed by heavy rainfall events. The erosivity with a secondary rainfall of ≥25 mm accounted for 67.00% of the total erosivity. 3) The erosivity had obvious characteristics of centralized distribution within the year. The erosivity in the 12^th to 16^th half-months accounted for 83.51% of the annual erosivity value, and the erosivity in the 15^th half month was the highest in the year, accounting for 31.68% of the annual erosivity, which was a period of great potential water and soil loss. 4) In the calculation of EI₃₀, detailed rainfall data was recommended. The YIN shuiqing cold and warm season model was more effective in predicting annual erosivity and semi-month erosivity when the rainfall process data was lack.

Conculsions The results revealed the centralized distribution of rainfall erosivity in 24 half-months, and suggested the soil and water conservation measures against surface disturbances in the 12^th to 16^th half-months. Compared with ZHANG Wenbo model and CREAMS model, YIN Shuiqing cold and warm season model is recommended in predicting annual erosivity and semi-month erosivity.