Evaluation the performances of stochastic streamflow models for the multi reservoirs

Abstract

Pedu-Muda reservoirs responsible to supply sufficient water capacity during paddy cultivation period twice a year. Thus, improper management and operation of the reservoirs creating the scarcity issue of water availability especially during dry season. Synthetic streamflow being as a main role in predicting the capability and sustainability of these reservoirs to cope the demand. This study evaluated the performances of stochastic streamflow model to produce the synthetic streamflow generation. Two comparable models, Valencia Schaake (VS) and Thomas Fiering (TF) represented for disaggregation and aggregation models, respectively. Each model was analyzed for 100 times of simulation to generate the long-term synthetic streamflow. There were 3 basis of the statistical analyses consist of lag one correlation, mean, mean absolute error (MAE), and standard deviation (St.D) of annual and monthly levels for both models were evaluated to compare the model performances. The results revealed the generated streamflow series by VS models had better performances to the historical streamflow record than the TF model in term of annual and monthly excepted on Feb, Aug, Sept, and Oct with less correlation values. The errors of these months between historical and generated correlation values are in the range of 0.14 to 0.20. However, both models can preserve a good agreement to the mean even the range of monthly streamflow were overestimated/underestimated by VS and TF models respectively. The average annual generated streamflow is predicted to reduce 0.7% (by VS) and 2.4% (by TF) from the historical record.