Assessment of Efficiency Increase of Reservoir Operation by the Use of Inflow Forecasts: A Case Study of The Mae Klong River Basin in Thailand.

Abstract

Inflow forecast is one of the most useful information for efficient reservoir operation. This paper presents the method of assessing the benefit of the use of the inflow forecasts. Once the forecasts are available, the information is used in the following way: First, deterministic dynamic programming (DDP) is employed in the period when the forecasted inflows are available. Secondly, stochastic dynamic programming (SDP) under periodical first order Markov inflow assumption is employed in the period beyond that. The advantage of this proposed method is that it can directly incorporate the inflow forecasts obtained from any forecasting technique for real-time reservoir operation provided that the forecasts can be used as deterministic input. The Mae Klong River basin in Thailand is applied to assess the efficiency increase of reservoir operation under different forecast accuracy and lead time. Substantial improvements were found in the aspects of higher average annual energy, smaller standard deviation, and higher reliability for irrigation water and water supply in comparison with the case that employs no forecasts. For this river basin, it is recommended to utilize the inflow forecasts for real-time operation even though the forecast accuracy is as poor as R2=0.6 where R2 is the ratio of the explained variance to the original variance relative to the mean of the whole observations.