A new global method of satellite dataset merging and quality characterization constrained by the terrestrial water budget

During the last decades, satellite observations have increasingly been used to study the global water cycle over land. Although their value is now appreciated by the hydrological community, they are still limited by their uncertainties and their inability to close the water budget. In a previous study, we optimally integrated several datasets for each component (precipitation, evapotranspiration, storage change and discharge) to close this budget at a basin scale. Furthermore, an independent and simple calibration of each satellite dataset was designed to reduce the budget residual. In this paper, we extend the calibration procedure to the global scale. Pixels are first classified into surface types characterized by their NDVI and net precipitation values. We show that the global calibration transforms the original datasets towards a consensus that is hydrologically more coherent, with a budget residual reduced by 26%. The calibrated datasets are compared to ground-based observations, showing an improvement for more than 65% of the sites tested. This opens new perspectives to generate long-term datasets at global scale based purely on all available satellites observations, which describe all the terrestrial water components useful for climate purposes. Beyond the simple calibration presented here, inconsistencies among the various satellite datasets can be used as a proxy for satellite observation uncertainties. The quality of our calibration procedure is constrained by the availability of discharge measurements, and could therefore be improved in the future, as discharge measurement networks become more extensive.