The impact of soil initialization on regional decadal climate predictions in Europe

Abstract

Inaccuracies in soil initialization, caused by unbalanced soil conditions, are one possible reason for biases in regional climate predictions. Therefore, in the present study, stand-alone simulations with a land surface model are performed in order to create balanced soil conditions. The impact of different atmospheric forcing data on the quality of the stand-alone simulation is evaluated by comparing the results to observations. The results reveal that stand-alone simulations driven by the atmospheric reanalysis ERA-Interim give the best results in terms of the correlation to observed soil temperatures and soil water contents. Based on these results, a regional climate simulation is initialized with these balanced soil conditions and compared to a reference run with unbalanced soil. The results show that the soil initialization of decadal climate simulations generally affects the spatial variability of summer precipitation in Europe. However, no systematic added value of the initialization with balanced soil conditions is evident for precipitation. Still, a clear added value can be observed for the simulated summer temperatures, especially in warm and semi-arid regions like the Iberian Peninsula. Since the balanced soil initialization contains less soil water than the unbalanced one, this simulation is able to reproduce the characteristic drought stress in summer in a more realistic way. We conclude that proper soil initialization can lead to potentially better decadal predictions, particularly for warm and semi-arid regions. An accurate soil initialization may therefore become essential, especially in European areas that are expected to become warmer and drier in summer due to climate change, e.g. Central and Eastern Europe.