Optimal estimation of parameters for a estuarine eutrophication model

The problem of eutrophication model calibration is formulated as a coupled, multi-objective optimization. An optimal algorithm is integrated into a vertically averaged, two-dimensional eutrophication model for the model calibration. The sensitivity coefficient matrix is used to estimate the Hessian matrix. The modified Gauss-Newton method is used to estimate the unknown parameters. The approach is straightforward and requires less effort in code modification for an existing model. A series of model experiments with a real-time eutrophication model of the tidal Rappahannock River estuary is conducted. Thirteen unknown kinetic parameters are estimated with hypothetical observations generated by the model using 'true' model parameters. Model experiments are conducted using both observations with and without observation errors. The experiments demonstrate the efficiency and accuracy of the optimal methods and suggest that the method is feasible for estimating eutrophication model parameters within acceptable error ranges.