Satellite soil moisture data assimilation for improved operational continental water balance prediction

Abstract

Abstract. A simple and effective two-step data assimilation framework was developed to improve soil moisture representation in an operational large-scale water balance model. The first step is the sequential state updating process that exploits temporal covariance statistics between modelled and satellite-derived soil moisture to produce analysed estimates. The second step is to use analysed surface moisture estimates to impart mass conservation constraints (mass redistribution) on related states and fluxes of the model in a post-analysis adjustment after the state updating at each time step. In this study, we apply the data assimilation framework to the Australian Water Resources Assessment Landscape model (AWRA-L) and evaluate its impact on the model's accuracy against in-situ observations across water balance components. We show that the correlation between simulated surface soil moisture and in-situ observation increases from 0.54 (open-loop) to 0.77 (data assimilation). Furthermore, indirect verification of root-zone soil moisture using remotely sensed vegetation time series across cropland areas results in significant improvements of 0.11 correlation units. The improvements gained from data assimilation can persist for more than one week in surface soil moisture estimates and one month in root-zone soil moisture estimates, thus demonstrating the efficacy of this data assimilation framework.

An Extended Kalman Filtering Technique for Geodetic First Order Data Assimilation

“…The first uses measurements to update the state vector while the second update enacts the balance constraint between uncertainty in measurement and estimate via the gain matrix. Data merging algorithms along with the EKF has been applied in recent studies (see, e.g., Koch et al, 2020;Tian et al, 2020) for a more stable composition and balance between measured and estimated data. Even though the enhanced datasets have consistently resulted in better estimates over different time varying signals and analysis, the subject of assimilating a perfect set of measurements despite the resulting robust constraint still poses a major setback in the system state (e.g., Khaki et al, 2018).…”