Inverse groundwater modeling with emphasis on model parameterization

摘要: [1] This study develops an inverse method aiming to circumvent the subjective decision regarding model parameterization and complexity in groundwater modeling. The number of parameters is included as a variable along with parameter values. A based on B-spline surfaces (BSS) selected approximate transmissivity, genetic algorithms were perform error minimization. transform linear least squares (LLS) developed, so that different parameterizations may be combined by standard algorithm operators. First, three applications, isotropic, anisotropic, zoned aquifer parameters, are examined single objective optimization problem estimated transmissivity found near true one. Interestingly, anisotropic case, converged solution distribution control points. Next, regularization, penalizing complex models, considered, last, expressed multiobjective framework (MOO), where goals simultaneous minimization calibration complexity. result MOO Pareto set potential solutions user can examine tradeoffs between select most suitable model. By comparing prediction errors, it appears, promising models ones region rate decrease increases drops (bend curve). This useful practical interest real modeling applications.