Loss potentials associated with European windstorms under future climate conditions

Abstract

Possible changes in the frequency and intensity of windstorms under future climate conditions during the 21st century are investigated based on an ECHAM5 GCM multi-scenario ensemble. The intensity of a storm is quantified by the associated estimated loss derived with using an empirical model. The geographical focus is 'Core Europe', which comprises countries of Western Europe. Possible changes of losses are analysed by comparing ECHAM5 GCM data for recent (20C, 1960 to 2000) and future climate conditions (B1, A1B, A2; 2060 to 2100), each with 3 ensemble members. Changes are quantified using both rank statistics and return periods (RP) estimated by fitting an extreme value distribution using the peak over threshold method to potential storm losses. The estimated losses for ECHAM5 20C and reanalysis events show similar statistical features in terms of return periods. Under future climate conditions, all climate scenarios show an increase in both frequency and magnitude of potential losses caused by windstorms for Core Europe. Future losses that are double the highest ECHAM5 20C loss are identified for some countries. While positive changes of ranking are significant for many countries and multiple scenarios, significantly shorter RPs are mostly found under the A2 scenario for return levels correspondent to 20 yr losses or less. The emergence time of the statistically significant changes in loss varies from 2027 to 2100. These results imply an increased risk of occurrence of windstorm-associated losses, which can be largely attributed to changes in the meteorological severity of the events. Additionally, factors such as changes in the cyclone paths and in the location of the wind signatures relative to highly populated areas are also important to explain the changes in estimated losses.

Mistral and Tramontane wind systems in climate simulations from 1950 to 2100

“…Numerous studies deal with near-surface wind speeds over Europe in climate projections (see Pryor et al 2006Pryor et al , 2012Rockel and Woth 2007), as well as changes in wind energy potential (e.g., Hueging et al 2013) and loss potential due to windstorms (Pinto et al 2012). Rockel and Woth (2007) found that the number of storm peaks (gusts greater then 8 Bft) increase over Western and Central Europe when applying the Special Report on Emissions Scenarios (SRES, Nakicenovic and Swart 2000) A2 scenario, while their number decreases over the Western Mediterranean Sea.…”

The 21st century decline in damaging European windstorms

“…The NAO is the dominant mode of lower-to midtropospheric pressure variability over the North Atlantic (Pinto et al, 2012). For high values of the NAO index, which often occur in winter, pressure differences between the Icelandic Low and Azores High increase, hence the frequency of low-pressure systems increases, leading to increased storm genesis.…”

“…In subsequent studies such as , Pinto et al (2007Pinto et al ( , 2012 and Donat et al (2010Donat et al ( , 2011, SSIs had similar composition to Klawa and Ulbrich (2003). In addition, Cusack (2013) used a variation on the Klawa and Ulbrich (2003) SSI, based on the cube root of L 98 , divided by the number of locations in which damaging winds occur.…”

“…Klawa and Ulbrich, 2003;Haylock, 2011;Bonazzi et al, 2012;Cusack, 2013;Roberts et al, 2014) or under future climate change conditions (e.g. Pinto et al, 2007Pinto et al, , 2012Leckebusch et al, 2008;Donat et al, 2011). In each study the severity of an event is measured using a storm severity index (SSI), i.e.…”

“…The NAO is the dominant mode of lower-to mid-tropospheric pressure variability over the North Atlantic (Hurrell et al, 2003). It, therefore, has a large influence on weather conditions over the North Atlantic basin and western Europe, hence the occurrence and intensity of European windstorms (Pinto et al, 2012;Economou et al, 2014). Cusack (2013) speculated that a different mix of climate-forcing mechanisms were influencing storms in the two identified low periods in the footprint loss function, but did not explore this relationship explicitly.…”