Hydrological performance of ERA5 and MERRA-2 precipitation products over the Great Lakes Basin

The Laurentian Great Lakes Basin (GLB) of North America. Precipitation is often the paramount driver in hydrological systems and dynamics. Gridded precipitation data from the fifth generation of European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA5) and the second Modern-Era Retrospective analysis for Research and Applications (MERRA-2) are playing a key role in hydrological modelling and hydro-climatological analysis activities. This study explores the discrepancies between ERA5 total precipitation (TP) and MERRA-2 bias-corrected total precipitation (PRECTOTCORR) products and their impacts on streamflow simulation over the GLB. ERA5 TP and MERRA-2 PRECTOTCORR perform differently, with substantial discrepancies in the eastern and mid-southern regions of the GLB. MERRA-2 PRECTOTCORR severely underestimates precipitation along the eastern shores of the Great Lakes within the eastern GLB, where lake-effect snowfall is common. Accordingly, the MERRA-2 PRECTOTCORR-driven hydrological modelling typically significantly underpredicts streamflow in those areas. In contrast, MERRA-2 PRECTOTCORR performs very well for precipitation estimation and stream discharge simulations in the mid-southern portion of the basin (i.e., the southern portion of the U.S. State of Michigan). However, the northernmost region of the GLB exhibits streamflow overprediction from the PRECTOTCORR-driven hydrological modelling. On average, ERA5 TP overestimates precipitation, especially in winter and spring. Accordingly, ERA5 TP typically results in higher simulated streamflow when compared to PRECTOTCORR. The quantified hydrological performance for the two state-of-the art precipitation products provides critical guidance for hydrological modelling applications in the GLB or other similar regions.