This paper investigates a comparative analysis of erosion modeling techniques based on deterministic models, both empirically-based and process-based models. The empirical modeling is based on statistical and artificial intelligence techniques. In the first, two types of statistical regression model structures are investigated, a linear multiple regression model structure and a nonlinear multiple regression model structure. In the second, tools such as artificial neural networks (ANN) and fuzzy inference system (FIS) are used. The physical process-based modeling involves the calibration, validation and testing of the models components: WEPP, EUROSEM and CIHAM-UC. The input and output variables of models were collected during rainy and dry (irrigation) seasons in Chirgua river basin, Venezuela for two years (2008–2009). Ninety-seven rainfall storms and 300 irrigation events were measured. Satisfactory fit was found in the techniques investigated, R² close to 0.7.
