The Weather Generator and The Regional Climate Model: The Rivals or The Allies?

In climate change (CC) impact studies, various weather-dependent models (e.g. crop growth models or rainfall-runoff models) are employed to assess possible CC impacts on crop yields, river flows, as well as on various climatological and other indices. To perform such experiments, one has to decide what meteorological data (representing present & future climate) will be used as inputs. Two most commonly used approaches at hand are Regional Climate Models (RCMs) and Stochastic Weather Generators (WGs). While one might say, that these two approaches are the rivals which compete for getting the main role in providing the weather data for CC impact studies, we want to demonstrate that these two approaches could be rather allies  which could be very effective while being used together.

In our presentation, we show three ways of such co-operation: (1) To produce weather series representing the future climate by the WG, WG parameters must be modified by CC scenarios, which may be derived by comparing RCM simulations of present vs. future climates. (2) To assess separate effects of changes (projected by RCMs) in individual weather variables (e.g. temperature or precipitation) and their statistical characteristics (means, variabilities, spatial and/or temporal correlations), WG may be used: only selected WG parameters representing chosen variable and its statistical characteristic may be modified before producing the synthetic series. (3) To assess statistical significance of the changes derived from a given RCM simulation, WG may be also used: the significancy of projected changes may be based on analyzing the spread of the changes derived by comparing multiple pairs of synthetic time series produced by WG calibrated with RCM simulations for future vs. present time slices.

In the experiment, we use: (a) our multi-site multi-variate parametric weather generator SPAGETTA, (b) E-OBS data to calibrate the generator for the present climate conditions in 8 European regions, and (c) outputs from ensemble of 19 RCM simulations for present and future climate (CORDEX database) in these regions.

Acknowledgement: The experiment was made within the frame of the PERUN project funded by the Technological Agency of the Czech Republic (project no. SS0203004000).