Comparison and evaluation of RUSLE and RUSLE2 models in soil loss estimation at rangeland hillslopes of Khamesan watershed, Kurdistan

The recent study was carried out to evaluate the RUSLE and RUSLE2 models for predicting soil loss in stony rangeland hillslopes of Khamesan watershed using event data obtained from natural rainfall. For this purpose, total runoff and soil loss were measured during three years (2011, 2012 and 2013) in eighteen plots with 24 m long by 1.8 m wide (six hillslopes with three replicates). The input data for both models were measured and monitored including soil characteristics, plant cover, ground cover etc. Twenty-four rainfall events were examined during the study period. Two models, particularly RUSLE2, estimated average annual soil loss relatively well, despite the overestimation. The results showed that both models, especially RUSLE2, was able to estimate single events soil loss relatively acceptable, especially in rainfalls with lower erosivity index (R2(RUSLE)=0.25, R2(RUSLE2)=0.53; EF(RUSLE)=0.23, EF(RUSLE2)=0.53). Calibration of the erodibility factor in two models led to considerable improvement of model efficiency (R2(RUSLE)=0.69, R2(RUSLE2)=0.82; EF(RUSLE)=0.63, EF(RUSLE2)=0.80). Comparison of estimations of two models showed that in all studied situations, RUSLE2 model had higher efficiency than RUSLE. It seems that with respect to characteristics of studied hillslopes on one hand and improvements of RUSLE2 model relative to RUSLE on the other hand, it has been led better estimates of RUSLE2. Generally, the results showed that in hillslopes with higher rock fragment cover and lime content, efficiency of non-calibrated two models was lower than other hillslopes. This result shows that more research is needed to examine the effect of these parameters on estimation of soil loss in two models, especially for the RUSLE