C1-WP2 - Exploring the predictive potential of land surfaces SVATs

Project aims

The goal of the project is to investigate the predictive potential of land surface processes on the European Climate, by performing simulations with COSMO-CLM (CCLM) coupled to different soil-vegetation-atmosphere transfer schemes (SVATs). In the first phase of MiKlip it could be shown that the SVAT as well as the initialisation used, have a considerable impact on quantities like near surface temperature and precipitation in Europe and in Africa. But the resulting potential for improvement of decadal predictions of near surface variables like temperature needs to be explored further. Therefore, the alternative SVAT VEG3D coupled successfully with CCLM via the OASIS3-MCT coupler during MiKlip I was developed further and made operational. Simulations with this model system were performed with a horizontal grid spacing of 0.22° and 0.0025°. The results of these simulations were assessed in Breil et al. (2017), Breil & Schädler (2017) and Breil et al. (2018).

Project structure

In this work package the different CCLM-SVAT simulations were performed and analysed by the Karlsruhe Institute of Technology (KIT).

Tasks of the project

  • Create a reference data set for soil initialisation in Europe.

  • Optimise VEG3D model code and settings.

  • Investigate the uncertainties of soil and vegetation processes by stochastic parametrisations.

  • Perform decadal hindcasts with CCLM coupled to different SVATs.

  • Assess the implemented SVATs and give recommendations for optimal model settings.

  • Integrate the CCLM-SVAT runs in the MiKlip ensemble.


D1: Provide soil water and temperature fields to initialize CCLM hindcasts in Europe.
D2: Quantification of uncertainties in soil and vegetation processes.
D3: Decadal hindcasts with CCLM coupled to different SVATs.
D4: Recommendations for optimal model settings and quantification of the impact of soil-vegetation-atmosphere interactions on the decadal climate variability in Europe.

Progress so far

  • A transient stand-alone simulation with VEG3D, driven by ERA-20C reanalysis for the period 1955-2010, was performed, to provide soil water and temperature fields for the initialisation of CCLM hindcasts for Europe.
  • The VEG3D source code and the model settings were further optimised to improve the robustness of the simulation results.

  • The impact of soil and vegetation processes on the uncertainties in regional climate predictions was investigated, by using stochastic soil and vegetation parameterisations. It could be demonstrated that stochastic soil and vegetation parameterisations in coupled CCLM-VEG3D runs improve the simulation of monthly rainfall sums all over Europe, in particular the simulation of summer precipitation inside the continent (Fig. 1).

  • Regional decadal hindcasts with CCLM-VEG3D for the decades 1980-1989 were performed, driven by the MPI-ESM-HR realisations 1-5 and compared to the standard CCLM-TERRA runs. For the monthly precipitation sums, an added value could be recognised in Central Europe, but for a slight majority of the grid cells no added value could be observed (Fig 2a). Additionally, an added value occured in Central and Eastern Europe and the British Isles for the predicted 2m temperatures. In the Iberian Peninsula and North Scandinavia no added value could be recognised (Fig. 2b).

Fig. 1: Mean Square Error Skill Score (MSESS) of the stochastic ensemble mean compared to the unperturbed CCLM-VEG3D reference run for the monthly precipitation sums in summer (a) and in winter (b) between 2001-2010.
Fig. 2: Mean Squared Error Skill Score (MSESS) of CCLM-VEG3D compared to CCLM-TERRA for the monthly precipitation sums (a) and the monthly 2m temperatures (b) in the decades 1980-1989 for the realisations 1-5.