The reduction of the systematic RCM summertime warm bias in South-Eastern Europe by stochastic root depth variation

The intensity of latent heat fluxes strongly depends on the available amount of soil water for evapotranspiration, i.e. the water amount stored within the rooted soil depth. But the determination of the root depth itself, and consequently also of the water supply for evapotranspiration, is associated with large uncertainties. The latent heat fluxes are therefore in many cases spuriously simulated, leading to temperature biases especially in soil-moisture limited evapotranspiration regimes like Southern Europe.

To take these uncertainties into account, a new method is introduced, in which the root depths within the Regional Climate Model (RCM) COSMO-CLM are stochastically varied. For this, the root depths in each model grid box are perturbed by uniformly distributed random numbers. The results of this stochastic simulation are compared to the results of an unperturbed reference run. The study reveals that during the winter season, evapotranspiration is virtually not affected by the stochastic root depth perturbation. Changes in the simulated near-surface temperatures are caused by indirectly induced chaotic changes in the atmospheric circulation. But in summer, the latent heat fluxes are considerably increased all over Southern Europe, due to a stochastic increase of the available soil water amount for evapotranspiration. Soil warming is consequently reduced and lower near-surface temperatures are simulated for the whole Mediterranean region. These lower temperatures reduce the strong warm bias in South-Eastern Europe in the reference run, which is also consistently simulated in several other RCMs and General Circulation Models (GCMs). Therefore, stochastic root depth modelling constitutes a simple method that can be implemented in every modelling system, to mitigate systematically simulated warm biases and thus, potentially improving model performances in semi-arid regions.