Application of Inverse Method to Calibrate Estuarine Eutrophication Model

Abstract

A procedure to calibrate an estuarine eutrophication model of the tidal Rappahannock River, a western tributary of the Chesapeake Bay, Va., using an inverse model is described. The inverse model using variational technique in conjunction with a real-time vertical two-dimensional eutrophication model consists of a forward model and an adjoint model. To relax the ill-conditioned Hessian matrix and speed up the model convergence, the parameter transformation method is introduced in the inverse model to construct a preconditioner. The condition of the Hessian matrix with respect to the new parameters is improved and the rate of convergence of the inverse model is increased. Thirteen unknown parameters were calibrated with a set of slack water survey data collected on July 5, 1990, in a 32-day model simulation. The model was verified using data collected on August 7, 1990. The results of the eutrophication model calibration and verification presented in this paper show that the inverse model is a good tool to aid model calibration. With the use of the inverse model, the unknown parameters can be estimated satisfactorily.