Exploring Application of Satellite-based Soil Moisture in Streamflow Predictions

In this study, we examine the impact of satellite-based soil moisture estimates on streamflow predictions using a distributed hydrologic model. We use SMAP (Soil Moisture Active Passive) and SMOS (Soil Moisture Ocean Salinity) data in an agricultural region of the state of Iowa in central US. We use three different strategies for updating model soil moisture states by satellite-based products. Firstly, we use a "hard update" method that is equivalent to replacing the model soil moisture by satellite observed soil moisture. Second, we use Ensemble Kalman Filter (EnKF) to update the model soil moisture, accounting for the modeling and observational errors. Finally, we incorporate the time-dependent satellite soil moisture error variance in perturbation of observations in EnKF. Additionally, we use streamflow assimilation at upstream inlet to isolate the effect of satellite-based soil moisture assimilation in a distributed hydrologic model. We compare model's streamflow predictions with streamflow observations from USGS gauges for four years (2015-2018). Our results indicate assimilating satellite-based soil moisture along with streamflow observations, reduces predicted peak difference compared to predictions from open loop and streamflow only assimilation. Further inclusion of time-dependent errors in EnKF improves streamflow prediction performance. Implications of our study can be useful for application of satellite soil moisture in real-time streamflow forecasting.