Mohammad J. Tourian & Nico Sneeuw

Institute of Geodesy, University of Stuttgart

The satellite missions GRACE and GRACE FO provided a fundamentally new observation-type for a wide spectrum of Earth science applications. They have fostered a number of novel approaches in oceanography, geophysics, hydrology and hydrometeorology. Despite all the revolutionary findings, the utility of GRACE data was mainly narrowed down on large catchments due to their poor spatial resolution. We propose a method to downscale GRACE data by incorporating the data and the distribution of available high-resolution hydrological data and model. For this purpose, we propose a Bayesian framework that facilitates the estimation of GRACE downscaled data in the face of auxiliary data and their distribution. While it is common within the Bayesian problems to consider a predefined model for the Bayesian ingredients (likelihood function and the prior distribution), we rely on copula to obtain nonparametric distributions from the data itself. We employed our method over the Amazon Basin and assessed the plausibility of our results by comparing them against spaceborne surface soil moisture data. The results show that the proposed methodology can successfully estimate downscaled GRACE terrestrial water storage changes and its uncertainty.