Uncertainty quantification during seismic oceanography inversion with start Temperature-Salinity models of different lateral resolutions

Seismic oceanography (SO) has been widely used on the inversion of physical oceanographic properties due to its higher lateral resolution up to 10m, compared to conventional oceanographic measurement methods. Normally, the inversion process requires seismic data and in-situ hydrographic data, and the latter is acquired by deploying XBTs/XCTDs. Recently, due to the advantage of providing quantifiable uncertainties of the inverted parameters, a Markov chain Monte Carlo (MCMC) algorithm has been used for the temperature and salinity inversion from SO data. Based on the MCMC inversion method, this study investigates the effect of the lateral density of XBT deployments on the resultant uncertainties of inverted temperature and salinity. We analysed the seismic data acquired in the Gulf of Cadiz (SW Iberia) in 2007 in the framework of the Geophysical Oceanography project. A nonlinear Temperature-Salinity relation is modelled using a Genetic Algorithm from CTD casts collected in the research area. Combining the temperature data from XBTs with the T-S relation, smoothed temperature and salinity prior distributions are derived. Then the posterior distributions of temperature and salinity are estimated using the prior information and the field reflectivity data. In this study, priors are changed by controlling the amount of XBTs used, after which the corresponding uncertainties of the inverted temperature and salinity are calculated. The result quantifies the impact of the prior models with different XBT deployment densities on the uncertainties of inverted results. It is proposed that the acquisition of a reasonable temperature starting model is the prior consideration when deciding the XBT deployment strategy along the seismic oceanography survey.