Wind speeds at a specific location depend not only on the large-scale atmospheric circulation, but also on the local environmental conditions. For example, over the Brocken (Harz Mountains) wind speeds in excess of 120 km/h (Hurricane force) are registered several times a year, whereas such a wind  speed occurs only very rarely over the Rhineland. The reason for these differences is that wind speed increases with altitude and decreases with higher roughness of the terrain (rougher orography), such as over densely built-up areas or forests.

Short-term fluctuations of the wind, so called “gusts”, are most decisive for the storm damage, which is approximately proportional to the gusts in the third power  (Klawa, 2001). Therefore, small changes in gust speeds have a significant impact on the amount and pattern of the damage.

The storm hazard is defined as the the probability of a certain wind speed (or vice versa). To consider the large spatial variability of the gust wind speed, either very dense measurement networks or high-resolution model simulations are necessary. Based on extreme value theory of storm fields over a climatological period (e.g. 30 years), wind speeds for a certain return period can be estimated.

Extreme value statistics in combination with numerical modeling were applied to create the CEDIM’s storm hazard maps as part of the project „Risk Map Germany” of the Center for Disaster Management and Risk Reduction Technology (CEDIM) at the Karlsruher Institute of Technology (KIT; Heneka et al., 2006; Hofherr und Kunz, 2010). As an example, Figure 1 shows maximum wind speeds that are expected or exceeded once in 10 years. The map shows a decrease in wind speeds from North to South as well as several maxima over more exposed areas such as mountain tops or mountain ridges.

According to the study of Heneka and Hofherr (2011), the loss related to winter storm Lothar (1999) had a return period of 8 years for the whole area of Germany, whereas for storm Kyrill (2007) this value was 7 years. Concerning the maximum wind gust speeds, however, storm Lothar had a much higher return period, for example for Switzerland of about 40 to 50 years (on average; Ceppi et al., 2008).

Historical storm events

Based on different reports recorded in more than 3000 different historical documents, an interactive database of different storm events in Baden-Württemberg (79 in total) during the last 200 years was constructed. As shown in Figure 2, the number of events has a very high temporal variability with several clustering of the events over the entire period. In fact, periods with high storm activity, such as at the end of the 19th century, take turns with periods of low storm activity, such as the first half of the 20th century. Overall, the storminess in the second half of the 19th century was much higher than during the second half of the 20th century. When considering only the nine most severe storms, it is conspicuous that alone six of these storms occurred in the second half of the 20th century. However, due to the small sample of events and the high temporal variability a clear pattern cannot be deduced for the entire time period. Neither it is possible to establish a connection to the observed temperature increase over the last 50 years due to climate change. This analysis also highlight the need for long time series to distinguish among the effect of natural climate variability and climate change (Kottmeier et al., 2004).

Storm hazard and climate change

Based on our current state of knowledge, it can be assumed that the frequency and/or intensity of winter storms will change in the future. The expected changes in the regional storm hazard greatly depend on how the large-scale weather systems will react to the expected temperature rise, especially to the different heating between high and low latitudes and the sea surface temperature increase.

Changes in the regional storm climate can be estimated from projections of climate models. Current global climate models, however, provide due to their low horizontal resolution (e.g., ECHAM5 ~ 200 km) only strongly smoothed and too low wind speeds (Fig. 3a). Extreme events such as severe winter storms are reproduced only by regional climate models (RCM) with a spatial resolution of less than 20 km (Fig. 3b).

Most of the scientific studies estimate that the number of extratropical low pressure systems in the northern hemisphere will decrease in the future due to a reduced meridional temperature gradient, which provides the energy of the storms (Ulbrich et al., 2009). Only in a few regions such as the North-East Atlantic region or over the British Isles may expect an increase in storminess in the future. Overall, the projected changes in the storm climate depend on the applied methods, the considered physical quantities, their thresholds and the specific region.

The expected changes in the local storm climate in Germany in the future show distinct spatial differences that partially depend on the RCM run. To account for the large uncertainty in the future climate (due to natural climate variability, unknown development and emissions, and limitations of the climate models), it is necessary to consider not only one RCM, but an ensemble of possible realizations.

Such an ensemble enables to determine both the robustness of the change signals as well as their probability. While, for example, the ensemble mean from eleven RCM simulations (Fig. 4) provides evidence for an increase in the10-year gust speeds of up to 5% , the results for Central and South Germany are not consistent and, thus, not reliable (Rauthe et al., 2010).
Note, however, that the considered control period from 1971 to 2000 included an unusual large number of events compared with other periods of the 19th and 20th centuries. Thus, the results of the climate models suggest that a high number of severe storms will also occur over future decades in Germany.

Text and Data: CEDIM, an interdisciplinary research institution of the German GeoForschungsZentrums and the Karlsruhe Institute of Technology.