Author(s)

Susana Gomez, Angel Perez & Rosa M. Alvarez

Abstract

In this paper we use multiscale optimization to find the transmissivities of a confined aquifer in the presence of errors in the data. We show that this method is able to find the true solution, and is faster and more robust than direct or Tikhonov optimization. Introduction The automatic identification of the parameters of a dynamical phe- nomenon, in this case describing the flux of water in a porous medium, is an inverse problem, and as such is ill-posed in general. This may be expressed in non-uniqueness of the solution and/or continuous dependence of the solution on the data. This is handled by using a regularization technique to approxi- mate the solution and is especially necessary in the presence of errors in the measured data used to estimate the parameters, otherwise oscillatory so

Keywords