Stochastic Optimization of the Highland Lakes System in Texas

Abstract

A multistage stochastic optimization model using linear programming was developed to provide planning tools for Lower Colorado River Authority in the operation of the Highland Lakes system, as well as a framework for examining the operation of four irrigation districts. Three primary objectives were maximized in the model: (1) Revenues from rice production, (2) recreation benefits associated with lake use, and (3) revenues from hydropower generation. The model includes stochastic inflows, weather-dependent irrigation demands, an interruptible contract decision function, a reservoir space rule to balance storage volumes between reservoirs, hydropower production, municipal and irrigation return flows, and bay and estuary inflow requirements. Model weights and coefficients were calibrated to reflect actual market prices and economic constraints, or to represent water management priorities. Stochastic optimization is used to account for meteorological uncertainty using representative inflow scenarios. Considering the uncertainty in the inflows, the model predicts optimum acreage levels, reservoir storage levels, reservoir releases, and other decisions, to maximize total expected benefits in the system.