Figure 2.35 99th percentiles of geostrophic wind speeds for winter (DJF). Triangles show regions where geostrophic wind speeds have been calculated from in situ surface pressure observations. Within each pressure triangle, Gaussian low-pass filtered curves and estimated linear trends of the 99th percentile of these geostrophic wind speeds for winter are shown. The ticks of the time (horizontal) axis range from 1875 to 2005, with an interval of 10 years. Disconnections in lines show periods of missing data. Red (blue) trend lines indicate upward (downward) significant trends (i.e., a trend of zero lies outside the 95% confidence interval). (From Wang et al., 2011.)
AR4 noted a likely net increase in frequency/intensity of NH extreme extratropical cyclones and a poleward shift in storm tracks since the 1950s. SREX further consolidated the AR4 assessment of poleward shifting storm tracks, but revised the assessment of the confidence levels associated with regional trends in the intensity of extreme extratropical cyclones.
Studies using reanalyses continue to support a northward and eastward shift in the Atlantic cyclone activity during the last 60 years with both more frequent and more intense wintertime cyclones in the high-latitude Atlantic (Schneidereit et al., 2007; Raible et al., 2008; Vilibic and Sepic, 2010) and fewer in the mid-latitude Atlantic (Wang et al., 2006b; Raible et al., 2008). Some studies show an increase in intensity and number of extreme Atlantic cyclones (Paciorek et al., 2002; Lehmann et al., 2011) while others show opposite trends in eastern Pacific and North America (Gulev et al., 2001). Comparisons between studies are hampered because of the sensitivities in identification schemes and/ or different definitions for extreme cyclones (Ulbrich et al., 2009; Neu et al., 2012). The fidelity of research findings also rests largely with the underlying reanalyses products that are used (Box 2.3). See also Section 14.6.2.
Over longer periods studies of severe storms or storminess have been performed for Europe where long running in situ pressure and wind observations exist. Direct wind speed measurements, however, either have short records or are hampered by inconsistencies due to changing instrumentation and observing practice over time (Smits et al., 2005; Wan et al., 2010). In most cases, therefore wind speed or storminess proxies are derived from in situ pressure measurements or reanalyses data, the quality and consistency of which vary. In situ observations indicate no clear trends over the past century or longer (Hanna et al., 2008; Matulla et al., 2008; Allan et al., 2009; Barring and Fortuniak, 2009), with substantial decadal and longer fluctuations but with some regional and seasonal trends (Wang et al., 2009c, 2011). Figure 2.35 shows some of these changes for boreal winter using geostrophic wind speeds indicating that decreasing trends outnumber increasing trends (Wang et al., 2011), although with few that are statistically significant. Although Donat et al. (2011) and Wang et al. (2012h) find significant increases in both the strength and frequency of wintertime storms for large parts of Europe using the 20CR (Compo et al., 2011), there is debate over whether this is an artefact of the changing number of assimilated observations over time (Cornes and Jones, 2011; Krueger et al., 2013) even though Wang et al. (2012h) find good agreement between the 20CR trends and those derived from geostropic wind extremes in the North Sea region.
SREX noted that available studies using reanalyses indicate a decrease in extratropical cyclone activity (Zhang et al., 2004) and intensity (Zhang et al., 2004; Wang et al., 2009d) over the last 50 years has been reported for northern Eurasia (60°N to 40°N) linked to a possible northward shift with increased cyclone frequency in the higher latitudes and decrease in the lower latitudes. The decrease at lower latitudes was also found in East Asia (Wang et al., 2012h) and is also supported by a study of severe storms by Zou et al. (2006b) who used sub-daily in situ pressure data from a number of stations across China. SREX also notes that, based on reanalyses, North American cyclone numbers have increased over the last 50 years, with no statistically significant change in cyclone intensity (Zhang et al., 2004). Hourly SLP data from Canadian stations showed that winter cyclones have become significantly more frequent, longer lasting, and stronger in the lower Canadian Arctic over the last 50 years (1953–2002), but less frequent and weaker in the south, especially along the southeast and southwest Canadian coasts (Wang et al., 2006a). Further south, a tendency toward weaker low-pressure systems over the past few decades was found for U.S. east coast winter cyclones using reanalyses, but no statistically significant trends in the frequency of occurrence of systems (Hirsch et al., 2001). Using the 20CR (Compo et al., 2011), Wang et al. (2012h) found substantial increases in extratropical cyclone activity in the SH (20°S to 90°S). However, for southeast Australia, a decrease in activity is found and this agrees well with geostrophic wind extremes derived from in situ surface pressure observations (Alexander et al., 2011). This strengthens the evidence of a southward shift in storm tracks previously noted using older reanalyses products (Fyfe, 2003; Hope et al., 2006). Frederiksen and Frederiksen (2007) linked the reduction in cyclogenesis at 30°S and southward shift to a decrease in the vertical mean meridional temperature gradient. There is some inconsistency among reanalysis products for the SH regarding trends in the frequency of intense extratropical cyclones (Lim and Simmonds, 2007; Pezza et al., 2007; Lim and Simmonds, 2009) although studies tend to agree on a trend towards more intense systems, even when inhomogeneities associated with changing numbers of observations have been taken into account (Wang et al., 2012h). However, further undetected contamination of these trends owing to issues with the reanalyses products cannot be ruled out (Box 2.3) and this lowers our confidence in long-term trends. Links between extratropical cyclone activity and large-scale variability are discussed in Sections 2.7 and 14.6.2.
Studies that have examined trends in wind extremes from observations or regional reanalysis products tend to point to declining trends in extremes in mid-latitudes (Pirazzoli and Tomasin, 2003; Smits et al., 2005; Pryor et al., 2007; Zhang et al., 2007b) and increasing trends in high latitudes (Lynch et al., 2004; Turner et al., 2005; Hundecha et al., 2008; Stegall and Zhang, 2012). Other studies have compared the trends from observations with reanalysis data and reported differing or even opposite trends in the reanalysis products (Smits et al., 2005; McVicar et al., 2008). On the other hand, declining trends reported by Xu et al. (2006b) over China between 1969 and 2000 were generally consistent with trends in NCEP reanalysis. Trends extracted from reanalysis products must be treated with caution however, although usually with later generation products providing improvements over older products (Box 2.3).
In summary, confidence in large scale changes in the intensity of extreme extratropical cyclones since 1900 is low. There is also low confidence for a clear trend in storminess proxies over the last century due to inconsistencies between studies or lack of long-term data in some parts of the world (particularly in the SH). Likewise, confidence in trends in extreme winds is low, owing to quality and consistency issues with analysed data.