The Downscaling of the SMOS Global Sea Surface Salinity Product Based on MODIS Data Using a Deep Convolution Network Approach

Downscaling is a very important process to convert a coarse domain satellite product to a finer spatial resolution. In this paper, a deep learning based downscaling method was designed to improve the spatial resolution of the global sea surface salinity (SSS) products of Soil Moisture and Ocean Salinity (SMOS) satellite. The proposed algorithm is able to efficiently and effectively use high spatial-resolution Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data to improve the spatial resolution of SMOS SSS products.