Direct Temperature and Salinity Acoustic Full Waveform Inversion

Abstract

Acoustic reflectivity provides insight into the fine structure of the ocean by using the multichannel seismic reflection method. Retrieving physical properties of the ocean from acoustic reflectivity has been one of the main topics of this subject. Until now, the main inversion strategy has been to invert sound speed from reflectivity. Afterwards, a combination of equations of state and a local T-S relationship is required to derive subsequently the rest of parameters (temperature, salinity and density). We have developed an alternative approach based on the direct inversion of temperature and salinity from the reflected waveform. We have implemented a time domain, multi scale, acoustic FWI algorithm with a gradient-based optimization scheme that solves for T and S. The sensitivity kernels of those parameters are calculated via the adjoint method, and we use the iterative non-linear conjugate gradients (NL-CG) search method to minimize a misfit function. We present in this work the T and S inversions obtained from acoustic synthetic data propagated through the thermohaline finestructure of the Gulf of Cadiz, North Atlantic Ocean