Revising One Time Lag of Water Level Forecasting with Neural Fuzzy System

The rainfall intensity of annual typhoon is high in Taiwan. It often result in the downstream flooding and losses of economics. Thus, simulation model of rainfall-runoff is an very important subject . In past research, time series model often forecast water level . But time series often have an influence on one time lag for forecasting of data.When reach to peak water level ,the error occurs in the model simulation. So this research combines with Adaptive-Network Based Fuzzy Inference System (ANFIS) to revise time series model. In ANFIS, it will utilize rainfall and water level to set up the model. Estimated the water level trends by the rainfall. To revise one time lag of time series . Expect to improve the error of time series with this method. In addition set up one group ANFIS model. The model utilizes rainfall to estimate water level . According to the result, it is not well that ANFIS simulate water level with rainfall but it can obtain water level trends with rainfall trends . To revise the error of time series about reach to peak water level and it can obtain accurate and steady result.