Progress so far

Within the work package a down-scaling with CCLM of the three-member MiKlip I DROUGHTCLIP ensemble of assimilation simulations has been performed over the period 1900 – 2009 (figure 2). The DROUGHTCLIP simulations were driven by the NCEP 20CR reanalysis (CCLM EUR-20C). In addition, coupled regional simulations with CCLM-NEMO have been generated for this period with both Northern and Baltic seas as well as Mediterranean Sea coupled to the atmospheric model.

A regional centennial hindcast ensemble with 3 members has been established for the starting years 1910 – 2009, based on data from the module E DROUGHTCLIP global hindcast simulations (Müller et al., 2014). The simulations show a significant correlation to observed near surface temperature data for most of Europe (figure 1). The skill is lower for the first half of the period (starting years 1910 – 1959) than for the second half (1960 – 2009, overlapping with the main MiKlip period). This can be attributed to the lower availability of reliable observation data in the first half of the 20^th century needed for initialisation and verification. But it also points to a lower impact of the climate trend during this period. The spatial skill pattern is similar to those of the regional MiKlip core ensembles of C3-WP3, with higher skill in Southern and Western Europe and lower values in Northern and Eastern Europe. This indicates that the distribution is quite robust over the time.

In collaboration with C2-WP2, the multi-decadal variability of extreme values over Europe and their links to large-scale patterns has been analysed. Figure 2 shows the long-term variability (5-year running means) of the number of heavy precipitation days (> 20mm/day) in Europe derived from the CCLM EUR-20C reference simulations. This extreme value index is highly correlated with the index for the detrended North Atlantic sea surface temperature (AMO index). A high sea surface temperature leads to a moister atmosphere over the Atlantic. Downstream in Europe this causes an increase of heavy precipitation. In cold phases of the AMO the likelihood of extremes is lowered.

The coupled regional simulation shows improvements over islands (e.g. British Isles, Corsica and Sardinia) and in some coastal regions, which is expected due to the better representation of atmosphere-see interactions. There are also differences between the coupled and uncoupled simulation over the continental regions as well. For example, over a part of Central Europe the precipitation values correlate better with observation than the uncoupled results. Nevertheless, there are also regions where the uncoupled simulation fit better to the observations. The origin and further characteristics of these differences are still under investigation.