EGU General Assembly 2023
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

­­Joint assimilation of ASCAT backscatter and slope into the ISBA land surface model at ISMN stations over Western Europe

Xu Shan1,2, Susan Steele-Dunne1,2, Sebastian Hahn3, Wolfgang Wagner3, Bertrand Bonan4, Clement Albergel4,5, Jean-Christophe Calvet4, and Ou Ku6
Xu Shan et al.
  • 1Department of Geoscience and Remote Sensing, Faculty of Civil Engineering and Geosciences, TU Delft, Delft, the Netherlands (
  • 2Department of Water Management, Faculty of Civil Engineering and Geosciences, TU Delft, Delft, the Netherlands
  • 3Department of Geodesy and Geoinformation (GEO), Vienna University of Technology, Vienna, Austria
  • 4CNRM, Université de Toulouse, Météo-France, CNRS, Toulouse, France
  • 5now at European Space Agency Climate Office, ECSAT, Harwell Campus, Didcot, Oxfordshire, UK
  • 6Netherlands eScience Center, Amsterdam, the Netherlands

ASCAT normalized backscatter and slope are jointly assimilating into the ISBA-A-gs land surface model (LSM) to constrain plant water dynamics processes. An Extended Kalman filter is used as the data assimilation (DA) algorithm with a trained Deep Neural Network as the observation operator to link the states and the observations (Shan et al., 2022). DA and model open loop (OL) runs are performed on ASCAT grid points (GPIs) containing ISMN stations in Europe and validated using data from 2017 to 2019. Performances of DA and OL are evaluated against ISMN in-situ soil moisture observations in different layers and satellite-based LAI observations from the 1km v2 Copernicus Global Land Service project (CGLS) product.

Analysis of DA diagnostics suggests that our DA system is free of bias. Domain median values of innovations, residuals and increments are all around zero. The reduction of standard deviation of residuals compared to innovations shows that DA is effective in reducing uncertainties. Median values of O-A/O-F are close to unity, suggesting the weight given to the ASCAT observables ensures that they provide valuable information to constrain the model. Time series standard deviation of normalized innovations are shown to be around 1 which means our DA system satisfies the Gaussian hypothesis. Regional variations in the mean standard deviation suggests that the performance of the assimilation framework varies somewhat across different land covers.

Aggregated across space and time, the improvement in domain median values of ubRMSE and KGE are not statistically significant. However, improvement is observed in some land cover types, and at specific times of year. For example, analysis of the monthly performances in Agricultural grid points shows that DA corrects deeper soil moisture in spring. Results from our previous studies suggest that this may be due to the indirect link between deeper soil water availability and vegetation water status revealed by ASCAT slope. There are also improvements in LAI in fall and winter, suggesting potential values of ASCAT observables in crop senescence. This is consistent with results from Bonan et al. (2014), who found that assimilation of LAI with SSM  could shift the delayed plant phenological cycle simulated by ISBA compared towards real observations (Bonan et al., 2014). In addition, it is important to note that assimilation is performed at the ASCAT resolution scale, while the ISMN provides point-scale soil moisture.

Analysis of DA diagnostics as well as performance statistics suggest that the efficacy of ASCAT assimilation is sensitive to the prescribed model and observation errors. Ongoing research is focused on providing realistic quantitative values of both to ensure that the information contained in the ASCAT backscatter and slope can be optimally used.


How to cite: Shan, X., Steele-Dunne, S., Hahn, S., Wagner, W., Bonan, B., Albergel, C., Calvet, J.-C., and Ku, O.: ­­Joint assimilation of ASCAT backscatter and slope into the ISBA land surface model at ISMN stations over Western Europe, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-8489,, 2023.