Assimilation of Soil Moisture Data from a Regional Network of Cosmic Ray Probes in the Land Surface-Subsurface Model TerrSysMP (Invited Presentation)

A limitation of traditional land surface models is the representation of hydrology. Lateral flow is not considered and aquifers are not represented or represented in a strongly simplified manner. However, we expect that soil moisture content is influenced by shallow aquifers and lateral flow of groundwater in aquifers. Some new generation land surface models like TerrSysMP (TSMP, consisting of the subsurface model ParFlow, the land surface model CLM and the atmospheric model COSMO) include lateral flow and aquifers in a physical manner. We coupled the data assimilation framework PDAF to TSMP, and were able to generate a code (TSMP-PDAF) which is very efficient for parallel computation and allows data assimilation studies at a high spatial resolution. The framework was tested for the Rur catchment in Germany, a simulation period > 1 year, and the assimilation of in-situ measured soil moisture data with cosmic ray probes. The simulation experiments focused, amongst others, on the following questions: (i) value of soil moisture data measured by cosmic ray probe for improving predictions, evaluated with independent verification data like EC data; (ii) impact of soil map quality (ranging from coarse scale FAO-soil map to high resolution regional soil map) and modelling of uncertainty of soil parameters on data assimilation performance; (iii) difference in prediction quality between state updating only and joint state-parameter updating; (iv) difference in performance between the land surface models CLM (land surface only) and TSMP (coupled subsurface-land surface) for data assimilation; (v) difference in data assimilation performance between real-world case and synthetic tests which mimicked the Rur catchment. Conclusions will be drawn on the potential of networks of cosmic ray probes for improving predictions with land surface models.