Abstract:
The Pasak Jolasid Dam is located in the Pasak River Basin. The flood magnitude change highly in the flood season (Aug.-Sep.). The storage capacity of the reservoir has only one-third of average annual runoff. The appropriate outflow release method can reduce downstream flood problem. In the past, the pattern of release from the reservoir used the operation rule curve with the flood events of all magnitudes. This operation might not fit with some flood events. Therefore this study was to develop the outflow release method in flood season to cope with the occurring flood event. The forecasted inflow results were applied jointly with outflow release method. In addition, a number of operation rule curves were developed and the capability of the reservoir to reduce flood was studied. The ANN model based on BPNN algorithm, time series model (AR model) and various correlation graphs were applied to develop the forecasting method. There were forecasting runoff in 5 subcatchments. The input variables were compared in 2 cases, one case with daily discharge and average daily rainfall and another case with discharge only. The forecasting periods were 1-13 days ahead. The efficiency of forecasting was found in training and testing statges higher than 84%. The operation rule curves were developed using the remaining time release equation and water balance equation. Then the capability of the reservoir to reduce flood in the last real flood 30 events and the design floods was studied. The appropriate operation rule curves were found as the water release from reservoir between 1 Aug.-10 Sep. used RC0300 or RC2546, between 11 Sep.-15 Oct. used RC0400 and then to the end of flood season used RC0500. The development of outflow release method considered the remaining inflow, reservoir storage and time with the water balance equation. It was found that the outflow hydrographs could be changed to distribute more uniformly, the flood peak could be reduce as average as 47.2% and could raising store the flood water at the end of floodseason to fill up the reservoir storage capacity. When the forecasted inflows to the reservoir 1-13 days ahead were applied along the outflow release method with developing the water release method, the flood peak could be reduced from the case without forecasting in the large flood events about 2.39-29.54%.