Stochastic optimization to maximize water supply index in conjunctive use of surface and groundwater

Abstract

Systematic operation of water resources systems requires rules that consider the uncertainties affecting system performance. Here, stochastic dynamic programming (SDP) is used to structure and solve a full-scale surface and groundwater conjunctive use model to derive a multiparameter operational policy (rule) that maximizes the water supply index ($\mathrm{WSI}$). The model uses a Markovian representation of inflows to account for hydrologic uncertainties. Both the reliability of allocated water and its vulnerability are employed to define the $\mathrm{WSI}$. To overcome the computational burden inherent with SDP, infeasible solutions are identified and removed from the model solution process. Optimal expected values of $\mathrm{WSI}$ and surface and groundwater uses are assessed. The system simulation model with embedded rules is executed to assess the performance of the derived rules. The derived rules employ water rationing and account for long-term benefits during periods when the available surface and groundwater resources may suffice to meet demand. The simulation results demonstrate that the derived operational rules produce high $\mathrm{WSI}$ values for the long-term operation of the system and ensure sustainable groundwater use.