Inverse problem for the Navier-Stokes equations and identification of immersed obstacles in the Mediterranean Sea

Abstract

This paper presents a theoretical and numerical investigation of object detection in a fluid governed by the three-dimensional evolutionary Navier--Stokes equations. To solve this inverse problem, we assume that interior velocity measurements are available only within a localized subregion of the fluid domain. First, we present an identifiability result. We then formulate the problem as a shape optimization task: to identify the obstacle, we minimize a nonlinear least-squares criterion with a regularization term that penalizes the perimeter of the obstacle to be identified. We prove the existence and stability of a minimizer of the least-squares functional. To recover the unknown obstacle, we present a non-iterative identification method based on the topological derivative. The corresponding asymptotic expansion of the least-squares functional is computed in a straightforward manner using a penalty method. Finally, as a realistic application, we demonstrate the robustness and effectiveness of the proposed non-iterative procedure through numerical experiments using the INSTMCOTRHD ocean model, which incorporates realistic Mediterranean bathymetry, stratification, and forcing conditions.