Systems modeling to integrate river valley water supply and irrigation decision making under uncertainty

Most water resources management and planning models typically fail to fully integrate the management of water supply and demand into a total system within stochastic environments. Previous work involving the authors has achieved this for river reservoir/irrigation systems. System management decisions were optimized by stochastic dynamic programming supported by computer simulation, both simplified by using very few state and decision variables. Definition of crop status through time was especially scant. Since computational burdens increase rapidly as state and decision variables increase, which variables to include becomes a critical choice. After refining the previous simulation models, this paper describes and tests the inclusion of extra variables which better define the crop's condition in the optimization model to help decide which variables to include in a specific environment. Testing the resulting dynamic programming strategies in a system simulation model shows that inclusion of these extra variables makes very little difference to the results. This means that the costs of deleting them in favor of other system management variables, such as on-farm storage contents, would be very low.