How skillful are satellite and reanalysis precipitation products at reproducing hydrological processes in semi-arid catchments?

Reanalysis and satellite-based precipitation products have undergone an unprecedented and increasing availability, which has triggered numerous hydrological modelling studies using them as input. In this context, a key open question concerns the ability of these products to reproduce plausible patterns of hydrological processes. <P />This work assesses the skill of 10 satellite-based precipitation products (TAMSAT, CHIRPS, ARC, RFE, MSWEP, GSMAP, PERSIANN-CDR, CMORPH-CRT, TRMM3B42, TRMM3B42-RT), and of 7 reanalysis precipitation products (JRA55, EWEMBI, WFDEI-GPCC, WFDEI-CRU, MERRA2, PGF, ERA5) if used as forcing data in hydrological modelling, and there subsequent ability to reproduce the temporal and spatial dynamics of several hydrological flux and state variables (i.e. streamflow, evaporation, soil moisture and terrestrial water storage). In total, 102 combinations are tested using 6 different temperature datasets (JRA55, EWEMBI, WFDEI, MERRA2, PGF, ERA5) for the calculation of potential evaporation. The model outputs are evaluated with in-situ streamflow data, soil moisture from ESA CCI, evaporation from GLEAM, land surface temperature from MODIS, and terrestrial water storage from GRACE. The study is carried out in the predominantly semi- arid Volta River basin in West Africa, using the fully distributed mesoscale Hydrologic Model (mHM) over a ten-year period (2003-2012). <P />The results highlight the contrasting performances of the rainfall- temperature product combinations with respect to pattern reproduction of hydrological state and flux variables. The best performing rainfall products are CHIRPS, TAMSAT, ARC, MSWEP and GSMAP for streamflow simulation; ARC, TAMSAT, CHIRPS, TRMMB42, and EWEMBI for evaporation; EWEMBI, TRMM3B42RT, TRMMB42, CMORPH-CRT and ARC for soil moisture; and finally CMORPH-CRT, ERA5, MERRA2, PGF and GSMAP for terrestrial water storage.