The objective of this study was to evaluate the performance of USLE, RUSLE, and WEPP soil loss prediction models for different Brazilian soil and climatic conditions. To achieve this objective, soil losses estimated by the mentioned models in nine locations in Brazil were compared with the losses measured in the field by the following statistical parameters: correlation coefficient, root mean square error, Wilmott concordance index, confidence index, Nash-Sutcliffe coefficient, and efficiency coefficient models. The experimental plots adopted in studies of soil losses used in this study had varied dimensions (38 to 1875 m²) and monitoring times and periods (01 to 07 years). There was a significant difference between the overall average of soil loss estimated by the models and the corresponding values measured in the field. Soil losses estimated by the WEPP model were more accurate in 46% of the cases, followed by RUSLE and USLE models with 42 and 12%, respectively. The statistical indicators showed that the WEPP model had better performance when compared with the other models. Despite the difference between the observed and the measured data for the three models, the WEPP model, which presented highest performance, angular coefficient closer to the unity in the relation between soil losses measured in the field and estimated by the model, and because it is based on physical processes, demonstrates to have higher potential to predict erosion for different Brazilian edaphoclimatic conditions.
erosion; soil conservation; hydrological modeling
