An ANN-based approach to modelling sediment yield: a case study in a semi-arid area of Brazil

This paper describes an Artificial Neural Network (ANN) model for estimating sediment yield based on runoff and climatological data. The model has been applied to an erosion plot inside the Sao Joao do Cariri experimental basin, which is located in the semi-arid portion of Paraiba State, Brazil. Large quantities of sediment tend to be generated only periodically in semi-arid regions, thus accurate estimations of when sediment yields are likely to be high are needed to improve erosion management in such areas. A total of 61 rainfall events, which occurred between 1999 and 2002, were utilized to calibrate and test the model. Another model, based on multiple linear regression (MLR) was used for comparison. The results produced by the ANN model appear to be superior to those generated by the MLR model. The results also indicate that the ANN model is suitable for identifying and extracting nonlinear trends for significant variables.