CLIMATE CHANGE IMPACT ON OPERATION POLICY AND PERFORMANCE INDICES OF A RESERVOIR USING MACHINE LEARNING TECHNIQUES

Abstract

Water resource systems play a major role in human wellbeing. The hydrologic changes resulting from climate change will affect the planning, design, and operation of water resource systems. Developing water schemes based on present conditions without considering possible future changes could increase water resource pressure in the future period. Therefore, it is of utmost importance to consider these changes in the future design and management of water supply systems. To comprehend the impact of climate change on water resource management, a comprehensive modelling framework that considered both hydrological responses and reservoir operation was indispensable. In this regard, Malampuzha reservoir system in Palakkad district of Kerala was selected to evaluate the impact of climate change on reservoir and its performance. To select the suitable climate model for the study, the performance of 15 CMIP6 GCMs in precipitation, maximum and minimum temperature was compared to the observed data of Malampuzha for the period 1990-2014 with the help of Compromise Programming (CP) that involves metrics such as R2, PBIAS, NSE and NRMSE. The results of the CP analyses of the statistical metrics suggests that CNRM-CN6-1 model for precipitation and MRI-ESM2-0 model for maximum and minimum temperature are the suitable models for Malampuzha region. Different bias correction techniques were applied to improve the raw predictions of GCMs. Power Transformation (PT) for precipitation and Variance Scaling (VS) technique for temperature has shown superiority over other techniques. Three future scenarios were considered in this study from CMIP6 Shared Socioeconomic Pathways (SSP126, SSP245 and SSP585). Selected bias correction techniques were applied to the future period to get bias corrected future climate variables. Rainfall-runoff modelling was chosen to predict reservoir inflow. Three hydrological models (IHACRES, SWAT and HECHMS) and Machine Learning (ML) models such as ANN, SVM, RF, and Wavelet coupled models were compared and Wavelet coupled RF (WRF) model was selected because of its greater accuracy and was used to simulate future reservoir inflow under different SSPs. CROPWAT model was used to estimate the irrigation water requirement for baseline and future periods. Land use change analysis of Malampuzha reservoir command area was done with the 192help of MOLUSE plugin of QGIS. Optimization program was developed by considering all the necessary constraints with the objective of minimizing squared relative deficiency in water allocation. Optimal water allocation was derived from the developed optimization technique using genetic algorithm for baseline and future periods. Reservoir performance indices such as Reliability, Vulnerability, resiliency and sustainability were calculated and compared for both timelines. Climate change impact on reservoir performance is evaluated. Selected GCM models predicted an increase in average annual maximum temperature (from 0.23oC in near future to 3.26oC in far future), an increase in average annual minimum temperature (from 0.62oC in near future to 3.12oC in far future) and decrease in average annual precipitation (2.73% in near future to 10.89% in far future) in the future compared with the base period. The LULC changes indicate a shift towards urbanization and plantation expansion, with a concurrent decline in agricultural lands (2425 ha (14%) reduction by the end of the century) and water bodies. Because of an increase in crop water demand of 11.7% and decrease in reservoir inflow of 27.3%, the amount of water allocation under optimal reservoir management conditions was less than the demand. The command area water demand will not be met by the reservoir for far futures of SSP245 and SSP585 scenarios because of more increase in temperature (3.26oC) and erratic behavior of precipitation which indicates the impact of climate change. A suitable optimization technique using genetic algorithm was developed which can be used for deriving the best operation policy for the Malampuzha reservoir in future. The observed decline in reliability and resiliency along with a notable increase in vulnerability from 0.028% to 9.47%, emphasizes the substantial challenges that climate change imposes on reservoir operation. These findings highlight the urgent requirement of climate-resilient management strategies to ensure the long-term sustainability of reservoir in the far future.