Satellite-derived potential evapotranspiration for distributed hydrologic runoff modeling in Midwestern basins

The historical lack of distributed input data has been a key factor in hindering the use of distributed hydrologic models for operational streamflow prediction by the National Weather Service (NWS), which currently relies primarily on lumped models. Satellite remote sensing has held the promise of providing the needed spatial variables for hydrologic applications for some time, and currently sufficient data from sensors such as the Moderate Resolution Imaging Spectro-radiometer (MODIS) have amassed such that robust modeling applications testing is now possible.;The goal of this study is to test the use of satellite-derived potential evapotranspiration (PET) estimates, computed using 13 MODIS observations and the Priestly Taylor formula (MODIS-PET), as input in the NWS Hydrology Laboratory Research Distributed Hydrologic Model (HL-RDHM). Daily PET grids at 4km resolution are generated for 13 watersheds in the upper Mississippi River basin. Precipitation data are obtained from the Climate Prediction Center's (CPC) Climatology-Calibrated Precipitation Analysis (CCPA). Application of the MODIS-PET is compared to model results using the PET grids that are provided as a default in the HL-RDHM. The default PET grids are based on historical ground-based evaporation measurements and are spatially and temporally less variable than the MODIS-PET. Model results are evaluated for the May 1 through September 30 period for eight years using observed evapotranspiration (ET) adjacent to two watersheds and daily discharge observations for all watersheds.;Results indicate that even with the more physically realistic MODIS-PET input, simulated basin discharge at the outlet shows little to no improvement compared to the default PET simulations. The simulated basin mean ET results exhibit mixed results when analyzing the MODIS-PET simulations against the default PET simulations. Calibrating several model parameters substantially improves simulated discharge for both MODIS-PET and default PET simulations; however, the range of improvement for simulated streamflow among individual basins varies between the two different PET data sources, and simulated discharge errors can often be directly related to simulated ET errors.