Parameters identification in a seawater intrusion model using adjoint sensitive method

A numerical method is presented for simultaneous identification of spatially distributed hydraulic conductivity and storativity in a seawater intrusion model. The technique is based on a nonlinear optimization procedure that minimizes the squared error between observed and computed water head values. Gradient information is efficiently computed by the adjoint method, so that the same effort is required regardless of the number of parameters involved. The cost function is minimized via a limited memory variable metric algorithm. Numerical experiments are presented to illustrate the technique and demonstrate its reliability and efficiency in determining the unknown parameters.