Exploring the utility of the downscaled SMAP soil moisture products in improving streamflow simulation

Abstract

Study regionThe Susquehanna and upper Susquehanna watersheds in the Northeastern of the United States of America (USA) Study focusThis study explored the utility of the Soil Moisture Active Passive (SMAP) soil moisture downscaled to a range of spatial resolutions for improving ensemble streamflow simulations. The SMAP level 3 soil moisture product with spatial resolution of roughly 40 km was downscaled to a range of spatial resolutions including 1, 3 and 9 km over the Susquehanna and upper Susquehanna watersheds. A set of experiments was conducted through direct insertion of the downscaled SMAP soil moisture into a physically-based distributed hydrological model. New hydrological insights for the regionThe updating of the model with the original and downscaled SMAP surface soil moisture markedly improved the accuracy of the ensemble streamflow simulations with the CRPSS and NRMSE values in the range of 0.10–0.17 and 0.79–0.85, respectively when compared to the non-updated model for the Susquehanna watershed. In addition, the ensemble spread was reduced, and the ensemble mean compares well with the observed streamflow. The 1 km downscaled SMAP soil moisture showed the highest accuracy in improving streamflow simulation with the CRPSS and NRMSE value of 0.21 and 0.72, respectively for the Upper Susquehanna watershed, whereas for the Susquehanna watershed downscaled SMAP at 9 km adequately improved the accuracy of the ensemble streamflow simulations with the CRPSS and NRMSE value of 0.17 and 0.80, respectively. Besides the top layer of the model, updating the second layer of the model with the vertically extrapolated SMAP soil moisture only slightly further improved the accuracy of the model.