Keywords

MOEAs, water allocation, climate uncertainties, robust decision making, Pearl River basin

Location

Session D5: Advancing in Environmental Decision Making Under Deep Uncertainty: Emerging Tools and Challenges

Start Date

12-7-2016 4:50 PM

End Date

12-7-2016 5:10 PM

Abstract

Water allocation is facing profound challenges due to increasing climate uncertainties. To deal with these uncertainties, we developed a modelling framework linking decision making on water allocation and multi-objective evolutionary search. The framework aims to assess how nine different multi-objective evolutionary algorithms (MOEAs) perform in identifying robust water allocation plans for large rivers. It was applied to the Pearl River basin in China to allocate sufficient flow into the delta to reduce salt intrusion in the dry season under current and future climate conditions. Results show all the MOEAs perform well in sustaining required discharge at two gauge stations from 1979 to 2010. However, the MOEAs differ greatly in computational time, percent of the reference Pareto approximate set captured, and hypervolume indicator. Overall, the epsilon multi-objective evolutionary algorithm (eMOEA) and non-dominated sorting genetic algorithm III (NSGAIII) are the two best algorithms. Performance of the MOEAs was assessed under different climate change scenario (2079-2099). Also under future flow conditions, eMOEA and NSGAIII have the best overall performance. Even the best performing MOEAs cannot maintain the required minimum discharge for all the possible futures. Historically, additional releases mainly came from upstream reservoirs in Hongshuihe and Nanpanjiang River basin. In the future, large increases in average water releases are projected at Baise reservoir in the Yujiang River basin to maintain minimum flows under a future climate. More reservoirs, especially in the Yujiang River basin and middle and lower reaches of the Xijiang River could potentially improve the future low flows into the Delta. This study shows that optimising water allocation using carefully selected state-of-the-art MOEAs can help limit water shortages and salt intrusion in the Pearl River Delta.

Share

COinS
 
Jul 12th, 4:50 PM Jul 12th, 5:10 PM

Comparing State-of-the-art Multi-objective Evolutionary Algorithms for Water Allocation under Different Climate Scenarios

Session D5: Advancing in Environmental Decision Making Under Deep Uncertainty: Emerging Tools and Challenges

Water allocation is facing profound challenges due to increasing climate uncertainties. To deal with these uncertainties, we developed a modelling framework linking decision making on water allocation and multi-objective evolutionary search. The framework aims to assess how nine different multi-objective evolutionary algorithms (MOEAs) perform in identifying robust water allocation plans for large rivers. It was applied to the Pearl River basin in China to allocate sufficient flow into the delta to reduce salt intrusion in the dry season under current and future climate conditions. Results show all the MOEAs perform well in sustaining required discharge at two gauge stations from 1979 to 2010. However, the MOEAs differ greatly in computational time, percent of the reference Pareto approximate set captured, and hypervolume indicator. Overall, the epsilon multi-objective evolutionary algorithm (eMOEA) and non-dominated sorting genetic algorithm III (NSGAIII) are the two best algorithms. Performance of the MOEAs was assessed under different climate change scenario (2079-2099). Also under future flow conditions, eMOEA and NSGAIII have the best overall performance. Even the best performing MOEAs cannot maintain the required minimum discharge for all the possible futures. Historically, additional releases mainly came from upstream reservoirs in Hongshuihe and Nanpanjiang River basin. In the future, large increases in average water releases are projected at Baise reservoir in the Yujiang River basin to maintain minimum flows under a future climate. More reservoirs, especially in the Yujiang River basin and middle and lower reaches of the Xijiang River could potentially improve the future low flows into the Delta. This study shows that optimising water allocation using carefully selected state-of-the-art MOEAs can help limit water shortages and salt intrusion in the Pearl River Delta.