| Abstract | Tha Tapao River Basin, located in the South of Thailand, is a vulnerable basin to high risks of flooding. It is also the gateway to other southern provinces. Therefore, there is only way for land transportation to the south, the way to bring the economic growth to the south of Thailand and also the other countries connected to Thailand. Therefore, when flood occur, the damages caused both short term and long term impacts to socio-economic and environment conditions of the basin. Since the telemetering system has been installed by RID for forecasting and warning system. The results of prediction were satisfied but there will be better if the forecasting model was consistent to rainfall prediction. Moreover, the accuracy of flood forecasting depends on rainfall forecasting, and the performance of stream flow forecasts is strongly dependent on the quality of quantitative precipitation forecasts used. Thus, this study aims to develop an hourly precipitation forecasting model by using Artificial Neuron Network (ANN) and develop rainfall runoff model for the study basin. Tha Ta Pao River Basin can be divided into 3 sub-basins; Tha Sae, Rub-ro and Tha Ta Pao. There are 12 rainfall and water level automatic gauging stations in Tha Ta Pao River Basin. When rainfall occurs, the discharge flow through Tha Sae and Rub-ro River from the upper part of basin, and then flow together to Tha Ta Pao River. Therefore, The water level in Tha Ta Pao River would be high enough to flood Muang Chumphon which located in the south of basin. Artificial Neuron Network (ANN) is applied for simulating future rainfall. The input parameters for rainfall forecasting are rainfall at time of forecast, rainfall 3 nearby stations, mean of rainfall from the 3 correlated stations, rainfall classification code, relative humidity, cloudiness, wind speed, wet-bulb temperature and pressure above sea level. The appropriate ANN architecture is 11-33-28-11-1 which uses Generalize Feedforward type and TanhAxon transfer function. Since the results of forecasting rainfall are satisfied, they are input into NAM model with specific NAM parameters. The EI and r of CHP04 outlet discharge are 0.845 and 0.916. The EI and r of CHP07 outlet discharge are 0.796 and 0.897. The EI and r of CHP08 outlet discharge are 0.715 and 0.736, respectively. The accuracy of stream flow forecasting may depend on the error of rainfall forecasting results and NAM parameters calibration. However, the ANN architecture and NAM parameters in this study should be re-calibrated time to time for the better results because there will be change for rainfall, runoff and meteorological condition in the future. |