Uncertainty of gridded precipitation and temperature reference datasets in climate change impact studies

Abstract. Climate change impact studies require a reference climatological dataset providing baseline period to assess future changes and post-process climate model biases. High-resolution gridded precipitation temperature datasets interpolated from weather stations are available in regions of high-density networks stations, as is the case most parts Europe United States. In many world's regions, however, low density observational renders gauge-based highly uncertain. Satellite, reanalysis merged product have been used overcome this deficiency. However, it not known how much uncertainty choice may bring studies. To tackle issue, study compares nine two over 1145 African catchments evaluate contribution results These deterministic all cover common 30-year needed define climate. The include gauge-only products (GPCC CPC Unified), satellite (CHIRPS PERSIANN-CDR) corrected using ground-based observations, four (JRA55, NCEP-CFSR, ERA-I ERA5) one gauged, (MSWEP). gauged-only (CPC Unified) (ERA5) product. All combinations these were streamflows. against that other sources uncertainty, top-down hydroclimatic modeling chain 10 CMIP5 (fifth Coupled Model Intercomparison Project) general circulation models (GCMs) under RCP8.5 lumped hydrological (HMETS GR4J) generate streamflows 2071–2100 period. Variance decomposition was performed compare different contribute actual uncertainty. Results show provide good streamflow simulations period, but outperformed others for catchments. They are, order, MSWEP, CHIRPS, PERSIANN ERA5. For present study, two-member ensemble provided negligible levels equal or larger than related GCMs metrics A selection best-performing (credibility ensemble) significantly reduced attributed still remained main source some metrics. can therefore be critical apparently small differences between propagate large amounts