Daily operation of Bukit Merah reservoir with stochastic dynamic programming under climate change impact

In recent decades, growing populations and economic development in urban regions have resulted in severe water shortage in many countries, whereas around 70% of the total global water is used in agriculture. Anthropogenic climate change is another serious concern, potentially causing water shortages over different spatial and temporal scales. Due to increases in the global mean temperature, changes in the frequency and intensity of precipitation, and rising sea levels. These changes will be having adverse effects on water resources management. Bukit Merah Reservoir (BMR) located in Perak, Malaysia is chosen as a study site to examine future optimal release policies to supply paddy irrigation water to the Kerian Irrigation Scheme (KIS), as well as meeting the domestic and industrial water demand. Kurau River Basin (KRB), where 4 weather stations are located, is considered as the main source of water supply to BMR. This study attempted to optimal reservoir operation with the adaptive future strategies under the new realities of climate change on the hydrological regimes at a tropical agro-hydrological watershed. Many studies have been conducted on the future change in the hydrological cycle at the global and continental levels over the coming decades using General Circulation Models (GCMs) under different greenhouse gas emission scenarios of Representative Concentration Pathways (RCPs). Climate projections from GCMs require downscaling for use in regional water resources management applications, to convert the variables from coarse resolutions to local or regional scales. In this study, future climate variables are generated through statistical downscaling, stochastic Weather GENerator (WGEN) method used to downscaling current and future rainfall and temperature from 10 GCMs output for the 3 future periods: 2010-2039, 2040-2069 and 2070-2099. The GCMs are driven by 3 of the recent updated RCPs scenarios,namely, RCP4.5, RCP6.0 and RCP8.5. The Richardson-type model was discussed to clarify trends and variations in each GCM and ensembles of the variables in the context of the different RCPs and future periods. The Soil and Water Assessment Tool (SWAT) hydrologic model is applied to KRB to predict streamflow for both historical (1976-2006) and future (2010-2099) periods by following a rigorous calibration and validation analysis using the Sequential Uncertainty Fitting (SUFI-2) technique. SUFI-2 procedures gave good results in minimizing the differences between observed and predicted flows at the outlet of the KRB. The objective functions, viz. coefficient of determination, (R2), Nash-Sutcliff, (NSE) and Percent Bias, (PBIAS), have been tested and show better correlation and agreement between the observed and predicted streamflows on monthly scale. The impact of climate change on future flows of the KRB is evaluated in the validated SWAT model. There is projected streamflow reduction during the off-season months and increasing trend is projected in the main cropping season, with the exception of June and July months where the streamflow remains low, which could be due to high surface warming in future. The response characteristics of the runoff process in KRB identified by SWAT is used for setting model structures for operation of BMR. Stochastic Dynamic Programming (SDP) is applied to determine the optimal policies for release discharges from the BMR, under current and future conditions (25 scenarios combinations of 10 GCMs, 3 RCPs and 3 future time periods). In particular, 6 sets of projections representing the upper and lower limits of changes in rainfall are considered. The penalty function is prescribed to minimize the difference between the actual release and demand while avoiding overflows from the irrigation canals and maintaining the BMR water level as close to the normal level as possible. Discounting the penalty is necessary to obtain a one-year periodic optimal policy as the limit of the terminal time T goes to infinity. Two extreme projections of rainfall change with opposite signs have been chosen. Using a realistic value of 0.950 for the discount rate, the optimal policies impose restrictions on the discharge rates to meet domestic demand more often for these two extreme cases than under current conditions. The results show that a striking consequence of the optimal policies for the two precipitation extremes both impose restrictions on supplying the irrigation water, resulting in similar increases in the maximum of the value function. This demonstrates that even if operators follow the optimal operation policy, mitigation measures against climate change and increasing water demands are necessary. As the development of alternative water sources currently seems to be inefficient. However, max penalty function mitigates the maximum deficit (MRI) from 23.4% to 11.6%, SDP is very powerful in suppressing the impact of climate change in term of vulnerability. The promotion of water saving technologies for water users is highly recommended.

Preview Text FK 2018 106 IR.pdf Download (1MB) | Preview

Actions (login required)

View Item