Continental freshwater discharge estimates from satellite observations and terrestrial water balance
Résumé
Freshwater discharge is a key component of the global water cycle. During the last few decades, remotely sensed hydro-climatic data has given the opportunity to overcome the numerous technical, political and economical challenges raised by ground-based observation networks. In the present study, we used satellite observations combined with atmospheric reanalysis in a coupled ocean-land-atmosphere water mass balance to estimate freshwater discharge over the last few years at basin, continental and global scales. Variations in terrestrial water storage (TWS) are provided by the Gravity Recovery and Climate Experiment (GRACE) mission, whereas precipitation minus evapotranspiration (P-E) is computed from atmospheric reanalyses (NCEP-NCAR and ECMWF datasets). In order to assess the accuracy of freshwater discharge estimates, uncertainties in each terms of the water balance are explored. GRACE-derived TWS is compared to outputs of Land Surface Models (GLDAS, WGHM and ISBA), keeping in mind that LSMs generally do not account for human-induced effects on the hydrological cycle. Independently, reanalysis-based P-E is compared to remotely sensed and/or in situ observations of precipitation (GPCC, GPCP, CMAP), evaporation over oceans (OAFlux, HOAPS) and modeled evapotranspiration over land (Penman-Monteith and Priestley-Taylor). Finally, freshwater discharge estimates are compared to in situ discharge observations from GRDC and to estimates from previous global studies. Discrepancies between the different datasets, in terms of annual mean as well as seasonal and inter-annual variability, allowed us to draw uncertainty maps, then highlighting regions where freshwater discharge estimates are the most uncertain. Such results may help to focus future developments on hydrological modeling for further improvement in freshwater discharge estimates.