The Use of Statistical Weather Generator, Hybrid Data Driven and System Dynamics Models for Water Resources Management under Climate Change

Nowadays, the assumption of stationary patterns in hydrologic time series is being challenged, mainly because of climate change and the uncertainty brought about by it. In this paper, climate change and impacts on water resources of Zolachay in Urmia Lake basin in northwestern Iran have been studied comprehensively. Expected precipitation and temperature changes are obtained from the results of general circulation models (GCMs) approved by IPCC AR4 in three emission scenarios of A1B, A2, and B1. To simulate climate change conditions for horizon 2020, LARS-WG, as a stochastic weather generator, has been employed. Analyzing results by a Kernel density estimator indicates a decrease in annual precipitation and a tendency to a warmer climate. Then different data-driven models such as artificial neural network and M5 model tree, in conjunction with wavelet transform, have been used to develop a rainfall runoff model of the basin on a monthly time scale. Results show that a warmer and drier climate in the future will cause the hydrograph to have temporal and quantitative changes. Operation of the multipurpose Zola Reservoir (located on the main stream) is simulated using the system dynamics approach. In addition to changes in the runoff of the basin, development scenario is also considered. Results demonstrate considerable changes in the reliability and deficiency measures in the operation of Zola Reservoir under climate change condition and development scenario. These results indicate that a revision in the rule curve of the reservoir is needed. Finally, this study predicts that more groundwater could be extracted to supply demands in the basin.