Dynamic Models for Floodplain Management

A dynamic model of floodplain management is shown to address nonstationary conditions, including land-use changes, channel modifications, economic development, and climate change and variability. The dynamic approach permits zoning, levee construction, and other decisions to be made sequentially, rather than only at the present. The dynamic model is formulated as a Markov decision process. A single-floodplain, single-objective, stationary model is extended to include multiple floodplains, nonstationarity, and multiple objectives. Linear programming is used for solution, though the problem may be large. The model is applied to a problem at Chester Creek, Pa. The optimal policy for levee building or replacement is found to depend on if a flood has just occurred and on the costs of buying out property owners and rebuilding homes and levees. Two cases of nonstationarity are examined, future bridge construction and future hydrologic changes. With nonstationarity, buyout of property owners following levee overtopping is an optimal policy since increased future flooding reduces the expected benefits of structural flood measures. When economic development is included with management costs and flood damages in a multiobjective formulation, the optimal policies include building larger levees and increasing floodplain development.