Parameter estimation using optimization methods in land subsidence models

Mathematical models can be used to predict the magnitude and the rate of land subsidence in alluvial aquifer systems. In general, mathematical models simplify the variability of geologic and hydrologic systems. The accuracy of subsidence estimates from such models is dependent, among other things, on realistic values for geologic, hydraulic, and deformation parameters used in the models. These parameters have some degree of uncertainty; moreover, parameters determined from laboratory tests usually cannot reflect the complicated structure of deforming layers. These factors affect the accuracy and reliability of land subsidence predictions. Optimization methods were used to determine the parameters necessary for simulating ground-water flow and land subsidence in the Saga Plain in coastal Japan. Using optimal parameters, which account for variability in geologic and hydrologic systems, model results compare favorably with field observations. The model has become an useful tool for government agencies that must mitigate subsidence.