Hydrologic evaluation of satellite and reanalysis precipitation datasets over a mid-latitude basin

Using precipitation data from satellite or global reanalysis products to force hydrologic models exhibits complex rainfall error and resolution effects in the simulation of streamflows. This study assesses the error propagation of two global (or near-global) precipitation datasets in terms of flood modeling for a range of basin scales (300–70,000km2) focusing on multi-year (2002–2011) simulations over a mid-latitude basin (Susquehanna River Basin) in the Northeastern United States. These datasets are the TRMM Multi-satellite Precipitation Analysis 3B42V7 (TRMM3B42V7) research product and the Global Land Data Assimilation (GLDAS) reanalysis system precipitation dataset, which represent 3-hourly rainfall time series at 25-km and hourly time series at 100-km spatial grid resolutions, respectively. The precipitation products, aggregated to a common 3-hourly time resolution, are used to force a distributed hydrologic model (Hillslope River Routing – HRR) for moderate and heavy precipitation events over the basin. The NCEP multi-sensor precipitation analysis (Stage IV) is used as the reference rainfall field for the evaluation of the precipitation and hydrologic simulation errors. Results show that the satellite product exhibits significantly better error statistics compared to the GLDAS. Particularly for the simulated streamflow, GLDAS is shown to have up to 7 (3) times higher mean relative error compared to the corresponding TRMM3B42V7 error metric for moderate (extreme) streamflow values. This significant divergence in the runoff simulation error statistics is attributed to differences between the two precipitation products in terms of the propagation of their error properties from precipitation to simulated streamflow. Significant improvement of the statistical scores (up to 50%) with increasing basin size is shown for the satellite product; this basin scale effect is marginal for the GLDAS product. •Significant uncertainty associated with coarse resolution reanalysis products (GLDAS int this study) when used for hydrological simulation purposes.•TRMM3B42V7 exhibits significantly improved error statistics than the GLDAS reanalysis product in terms of basin average rainfall and simulated runoff values.•The differences in the error statistics are more pronounced in the basin runoff simulations than basin average precipitation.•The error propagation analysis showed that the GLDAS product exhibits significant increase of the mean relative error and random error