Trans-dimensional Markov chain Monte Carlo inversion of sound speed and temperature: Application to Yellow Sea multichannel seismic data

Abstract

Understanding the oceanographic features of the sea water is important for ecosystem studies. In seismic oceanography, structures are imaged and physical properties, such as the sound speed, temperature, or salinity, are calculated using multichannel seismic data. These data provide high lateral resolution information at the full depth of the ocean. However, when the sea water depth is shallow, such as in shallow basins, conventional seismic oceanographic data processing techniques might not provide accurate inversion results for oceanographic properties or accurate images of the sea water structures. In this study, we use the trans-dimensional Markov chain Monte Carlo inversion technique, which assumes both the dimension and properties of the model as unknowns in inversion problems, to estimate the sound speed and define the locations of layer interfaces of the Yellow Sea, which is a semi-enclosed shallow basin. The ocean temperature is calculated using the estimated sound speed and the sound speed-temperature relationship. The estimated sound speed and temperature are compared with the true sound speed and temperature obtained from an expendable bathythermograph. The result shows that the proposed algorithm correctly estimates the sound speed and temperature and accurately images the oceanic structure. As a result, the trans-dimensional Markov chain Monte Carlo inversion can accurately identify the distribution of the Yellow Sea bottom cold water.