Using of Adaptive Neuro-Fuzzy Inference System for Rainfall-Runoff Modeling

The correct and accurate prediction of runoff can be a reliable way to overcome on huge risk of flood disaster. The complexity of natural systems and of hydrological processes makes physically modeling very difficult. In the last decade the intelligence models such as fuzzy in rainfall– runoff have been developed. Fuzzy based model can be used to model the process behaviors even with incomplete and imprecise or ambiguous information. The study presents three intelligence methods (artificial neural network, fuzzy inference system and neural-fuzzy inference system) for prediction daily and monthly runoff of Ligvanchai watershed then are compared these models with together and olden methods such as time series (ARIMA)
and regression models. The most advantage of fuzzy modeling is conceptually easy to understand. The results showed that the fuzzymodel performed better than other models and leaded to a high Nash-Sutcliffe criterion because of its ability in catching uncertainty of the phenomenon.