Regional Water Resource Routing Using Optimization

The South Florida Water Management District is responsible for managing and protecting water resources in a 46,439-square kilometer (17,930-square mile) region of South Florida. This area extends from Orlando in the north to Key West in the south and from the Gulf Coast in the west to the Atlantic Ocean in the east and includes Lake Okeechobee, the country's second largest freshwater lake. The District operates approximately 3,000 kilometers (1,800 miles) of canals and more than 200 primary water control structures across 16 counties to serve a population of over seven million people. To manage the water in this region so as to satisfy human as well as ecological needs, models that predict the economic and ecological impacts of water management decisions are essential. Over the past decade the District has been developing improved hydrologic simulation models to meet increasing demands for higher spatial resolution and realism in surface — groundwater interactions. These simulation models provide estimates of what will happen if water is managed in any particular way. Now the question being addressed is just what way should it be managed so as to best meet desired hydrological and ecological targets over space and time. One approach to answering the management question is to develop optimization models that can consider the state of the entire system — i.e., the water available in various basins throughout the region and the capacities of control structures to move that water around the region so as to best meet desired stages, flows, hydroperiods, and ecological habitats now and into the future. This is a multiobjective problem requiring judgments about the priorities of meeting all these targets over space and time. The challenge facing those involved in this model development effort is the linking of these relatively simple regional optimization models with the more complex, and accurate, hydrologic simulation models. Large regional optimization models are purposely simplified representations of reality, whereas the hydrologic simulation models are much more detailed. To capture the advantages of both, they are linked together in a daily time step. The optimization model `suggests' management actions, the outcomes of which will be determined by the hydrologic simulation model. This in turn will result in final system states that differ from what the optimization model predicted. The optimization model then needs to be modified so that it can make improved management decisions based on a more accurate assessment of the impacts of those decisions. So, within a day some iterations of optimization and simulation models are needed before moving on to the next day. Just how optimization models are linked to the simulation models and how the iterations between optimization and simulation models take place and how their results can be rapidly evaluated within the `plug-and-play' modeling architecture will be the subject of this third presentation and paper in this series on regional hydrological modeling.