Windstorms in Sweden - variations and impacts

Abstract: Windstorms have caused severe damage on several occasions in Sweden during the last century. A general concern about future projections of climate extremes has raised questions on the variations and impacts of windstorms. The main objective with this work was to reach a deeper understanding of the variability and possible trends in ­windstorms, and related storm impacts in forests. This was achieved by combining different methods and data sets to describe the historic wind climate and specific storm events. The overall research questions were: How has the wind climate varied in the past? How does storm damage in forests vary in relation to the windstorm climate? How does the forest storm damage vary in relation to the wind field during a storm event? The wind climate has been studied from long time series of pressure observations from Lund and Stockholm (1823 to 2006), the geostrophic wind based on pressures from Göteborg-Visby-Lund/Falsterbo (1881 to 2001). From the Lund and Stockholm ­pressures the east-west component of the geostrophic wind, as a proxy for the NW and SE winds, was derived. Further, a regionally resolved storm damage time series for Sweden was compiled. It was found that the increased volume of storm damage in Swedish forests, with very severe storm damage in 1954, 1969, 1999, 2005 and 2007, has not been mirrored by an increase in windstorms. Rather results indicate a weak negative trend of the 99th ­percentile of NW winds (from 1850 to 2006) and of geostrophic winds (from 1881 to 2001) in Southern Sweden. Periods of higher windstorm frequency and wind speed were found in the 1860s to 1900s and, almost as high, in the 1970s to 1990s. The south Swedish wind climate on a millennium scale, based on simulated pressures from a regional climate model, was compared with aeolian sediment influx data from two bogs. Findings from the simulations indicate a stormier early period in the last ­millennium, around 1100s, and less windy time period in the 1600s. This was partly in tune with the aeolian sediment peak during the early time period, but less clear, due to model uncertainty and larger error margins in the proxy data, for the less windy period. In addition the wind field of two storm events, the storm Anatol, on the 3rd of ­December 1999, and the storm Gudrun, on the 8th of January 2005, were simulated with a regional climate model (CRCM) and a weather prediction model (AROME) respectively. The ­distribution of the forest damage coincided with the simulated maximum wind field during the Anatol storm, whereas the relation was less clear between the simulated wind field characteristics during the Gudrun storm, and the distribution of forest damage.