Data Assimilation for Improving Soil Moisture Estimation at Hillslope Scales: Experiments with Synthetic SMAP Radar Data

In a series of synthetic experiments we test the hypothesis that data assimilation algorithms can be employed to improve soil moisture estimation at spatial scales of hillslopes (e.g.10 ⁰ –10 ² m). We use the Ensemble Kalman Filter (EnKF) to update an ensemble of hillslope-scale soil moisture fields simulated by a physically-based ecohydrology model with synthetic SMAP radar observations. For sparse vegetation, assimilation of the synthetic observations substantially reduces estimation error in near-surface soil moisture (e.g. top 5 cm), relative to the synthetic true soil moisture conditions. Key components of our data assimilation system are: (1) explicit representation of the impact of hillslope-scale topography on microwave observation, and (2) a Latin Hypercube-based soil parameter generator that preserves the correlation between soil properties and improves the reproducibility of soil moisture ensemble statistics.