Assimilation of SCATSAR Soil Wetness Index in SURFEX 8.0 to improve weather forecasts

Abstract. To date, in numerical weather prediction models it has only been possible to assimilate surface soil moisture data. Due to the structure of this surface soil layer in soil models, the effect of the assimilation vanishes fast. This results in a small impact on model performance. Here we present a combination of two new developments to overcome this problem. On the one hand, a new satellite based soil moisture data set is assimilated that combines the advantages of two different sensors (MetOp ASCAT and Sentinel-1 SAR) and the so-called T-value approach to estimate the soil moisture content of deeper soil layers. On the other hand, an advanced version of the well-established SURFEX soil model data assimilation software is used to ingest this new data source for improved short-range weather forecasts. Comparisons of the two data sets from satellite and soil model indicate that the estimation of deep soil moisture from superficial measurements produces reasonable estimates down to 0.5 m. Assimilation experiments with a simplified Extended Kalman Filter for a model domain covering Austria shows the benefit of this new combination with improved verification scores for temperature and relative humidity forecasts at 2 m above ground.