A nonlinear perturbation model (NLPM) based on Artificial Neural Network (ANN) and considering the antecedent precipitation index (API) is proposed and developed. The model structure is similar to the NLPM-API model. The difference is that the ANN is adopted to simulate the relationship between the input perturbing terms and the output perturbing terms. The daily rainfall-runoff data from the Mumahe and Nianyushan reservoir basins is selected to test the model. The proposed model is compared with the LPM, NLPM-AMN and NLPM-API models, the model efficiencies in these two basins are increased 10.84%, 1.54%, 10.6% and 21.59%, 0.67%,10.11% during calibration period; 5.56%, 0.97%, 4.41% and 11.86%, 1.76%, 7.97% during verification period, respectively. All other assessment indexes are also superior to other models.
