Seasonal sediment yield estimation on the basis of a revised RUSLE for a typical loess watershed

Background The Revised Universal Soil Loss Equation (RUSLE) model is widely employed for simulation of soil erosion. However, RUSLE commonly overlooks seasonal dynamics of rainfall and vegetation in its estimation of sediment production, and numerous studies usually neglected sediment transport processes and the relevant mechanisms, leading to substantial errors commonly observed in model simulation results, especially in well-vegetated area.

Methods In order to investigate whether it is possible to well estimate sediment yield by mainly using RUSLE accounting for seasonal information with respect to both rainfall and vegetation dynamic, SSY-RUSLE model incorporating monthly-scale rainfall erosivity and vegetation cover factors was developed, and an empirical factor representative of the processes of sediment transport and connectivity was involved. The SSY-RUSLE model was applied to the Xinshui River Basin in Shanxi province, China. The time period of 1995-1999 and 2010-2014 were used for model calibration, whilst the period of 2000-2009 used for validation. The only parameter of a was calibrated by comparing the annual sediment yield estimations derived from SSY-RUSLE against the measurements of the outlet of the watershed. Model performances were evaluated at either annual scale and monthly scaly. And spatial patterns of sediment yield estimation of SSY-RUSLE were assessed as well.

Results The SSY-RUSLE performed well in estimating sediment yield at either monthly scale or annual scale. Compared with RUSLE, the RMSE of SSY-RUSLE was enhanced by 5% to 10% for the calibration and validation periods, and NS by 13% to 16%. There was no statistically significant difference (P > 0.05) between the average monthly sediment yield estimations and the observations. The erosion rates from SSY-RUSLE for forest, grassland and cropland were less than those from RUSLE, with the values of 54.8%, 61.6% and 66.5%, respectively. And the erosion rate estimated from SSY-RUSLE for the mainly erosion-producing section of 15° to 25° was less than that of RUSLE (decreased by 25.1% around).

Conclusions It is concluded that, when using the models such like RUSLE, accounting for seasonal dynamic information of vegetation growth may significantly improve performance of soil erosion and sediment yield estimations. With the proceeding of regional ecological restoration, regional vegetation is deemed to be gradually improved. To more accurately projecting soil erosion and sediment yield estimation under improved vegetation condition, it is necessity to account for seasonal vegetation dynamic, which would be helpful for effectively implementing the measures of erosion controls and watershed management in future conditions.