Spatial Temperature-Precipitation Compound Events in Europe: Weather Generator vs. Regional Climate Models, Future vs. Present climate

We define spatial compound events characteristics and use them to validate the spatial weather generator (WG) and compare its performance with Regional Climate Models.

WGs are often used to produce input weather data for climate change impact studies. To justify their use, WGs are validated for their ability to represent various features of statistical structure of the real-world weather regime, especially those which may significantly affect outputs from the impact models. The validation indices may include characteristics of (a) probability distribution functions of weather variables (mean, variability, extremes), (b) temporal structure (persistence, spells of specific weather), (c) spatial structure, and (d) relationships between variables. Having been calibrated with observational data, the WG may be used to produce weather series representing the baseline climate. To produce future climate series, the WG parameters are typically modified by climate change scenarios derived from GCM or RCM simulations.

We present results obtained by the parametric spatial weather generator SPAGETTA and compare its performance for selected European regions with results based on the ensemble of RCM simulations available from the CORDEX database. In the first part, the WG (calibrated with E-OBS data) and RCMs (EUR11 and EUR44 simulations) are validated for their ability to reproduce selected temperature-precipitation compound events: spells of spatially extensive hot-dry, hot-wet, cold-dry and cold-wet weather; the use of these compound spells was motivated by the fact, that they are affected by multiple aspects of the statistical structure of weather series: spatial and temporal structure of weather data, and correlation between individual weather variables. In the second part, the future-climate spells are analysed from both WG-produced synthetic series (WG parameters being modified by RCM-based climate change scenarios, which also include changes in spatial and temporal correlations) and RCM future climate simulations. In comparing the WG-based and RCMs-based results, a special attention is given to inter-model variability of results in the RCM ensemble.

Acknowledgements: GRIMASA project (Czech Science Foundation, no. 18-15958S), SustES project (European Structural and Investment Funds).