Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection Droughts and wet spells

In the SAR, an intensification of the hydrological cycle was projected to occur as the globe warms. One measure of such intensification is to examine whether the frequency of droughts and wet spells are increasing. Karl et al. (1995c) examined the proportion of land areas having a severe drought and a severe moisture surplus over the United States. Dai et al. (1998) extended this analysis to global land areas using the water balance approach of the Palmer Drought Severity Index. Long-term global trends for 1900 to 1995 are relatively small for both severe drought and wet area statistics. However, during the last two to three decades, there have been some increases in the globally combined severe dry and wet areas, resulting from increases in either the dry area, e.g., over the Sahel, eastern Asia and southern Africa or the wet areas, e.g., over the United States and Europe. Most of the increases occurred after 1970. Except for the Sahel, however, the magnitude of dry and wet areas of the recent decades is not unprecedented during this century, but it should be noted that rainfall in the Sahel since the height of the drought has substantially increased. In related work, Frich et al. (2001) found that in much of the mid- and high latitudes, there has been a statistically significant increase in both the number of days with precipitation exceeding 10 mm per day and in the number of consecutive days with precipitation during the second half of the 20th century.

Recent changes in the areas experiencing severe drought or wet spells are closely related to the shift in ENSO towards more warm events since the late 1970s, and coincide with record high global mean temperatures. Dai et al. (1998) found that for a given value of ENSO intensity, the response in areas affected by drought or excessive wetness since the 1970s is more extreme than prior to the 1970s, also suggesting an intensification of the hydrological cycle. Tornadoes, hail and other severe local weather

Small-scale severe weather phenomena (SCSWP) are primarily characterised by quasi-random temporal and spatial events. These events, in turn, have local and regional impacts, often with significant damage and sometimes loss of life. Tornadoes and thunderstorms and related phenomena such as lightning, hail, wind, dust, water spouts, downpours and cloudbursts belong to this group. In the light of the very strong spatial variability of SCSWP, the density of surface meteorological observing stations is too coarse to measure all such events. Moreover, areally consistent values of SCSWP are inherently elusive. Statistics of relatively rare events are not stable at single stations, observational practices can be subjective and change over time, and the metadata outlining these practices are often not readily available to researchers. For these reasons, monitoring the occurrence of local maxima and minima in smoothed SCSWP series, as well as checking for trends of the same sign for different but related SCSWP (e.g., thunderstorms, hail, cloud bursts), are important for checking inconsistencies. Because of the inherent difficulty in working with these data, there have been relatively few large-scale analyses of changes and variations in these events. Nonetheless, a few new regional analyses have been completed since the SAR.

A regional analysis by Dessens (1995) and more recent global analysis by Reeve and Toumi (1999) show that there is a significant interannual correlation between hail and lightning and mean minimum temperature and wet bulb temperatures. Using a three-year data set, Reeve and Toumi (1999) found a statistically significant relationship between lightning frequency and wet bulb temperature. They show that with a 1°C increase in global wet-bulb temperature there is a 40% increase in lightning activity, with larger increases over the Northern Hemisphere land areas (56%). Unfortunately, there are few long-term data sets that have been analysed for lightning and related phenomena such as hail or thunderstorms, to calculate multi-decadal hemispheric or global trends.

Figure 2.38:
Annual total number of very strong through violent (F3-F5) tornadoes reported in the USA, which are defined as having estimated wind speeds from approximately 70 to 164 ms-1. The Fujita tornado classification scale was implemented in 1971. Prior to 1971, these data are based on storm damage reports (National Climatic Data Center, NOAA).

A regional analysis assessed the temporal fluctuations and trends in hail-day and thunder-day occurrences during a 100-year period, from 1896 to 1995, derived from carefully screened records of 67 stations distributed across the United States. Upward hail day trends were found in the High Plains-Rockies and the south-east, contrasting with areas with no trend in the northern Midwest and along the East Coast, and with downward trends elsewhere (Changnon and Changnon, 2000). The major regions of decrease and increase in hail activity match regions of increased and decreased thunder activity for 1901 to 1980 well (Changnon, 1985; Gabriel and Changnon, 1990) and also crop-hail insurance losses (Changnon et al., 1996; Changnon and Changnon, 1997). In general, hail frequency shows a general decrease for most of the United States over the last century, with increases over the High Plains, the region where most of the crop-hail damage occurs in the United States. So, despite an increase in minimum temperature of more than 1°C since 1900 and an increase in tropospheric water vapour over the United States since 1973 (when records are deemed reliable), no systematic increase in hail or thunder days was found.
In south Moravia, Czech Republic, a decreasing linear trend in the frequency of thunderstorms, hailstorms and heavy rain from 1946 to 1995 was related to a significant decrease in the occurrence of these phenomena during cyclonic situations, when 90% of these phenomena occur in that region (Brázdil and Vais, 1997). Temperatures have increased in this area since 1946.

Since 1920, the number of tornadoes reported annually in the United States has increased by an order of magnitude, but this increase reflects greater effectiveness in collecting tornado reports (Doswell and Burgess, 1988; Grazulis, 1993; Grazulis et al., 1998). On the other hand, severe tornadoes are not easily overlooked. Restricting the analysis to very strong and violent tornadoes results in a much different assessment (Figure 2.38) showing little long-term change, though some years like 1974 show a very large number of tornadoes. Furthermore, consideration of the number of days with tornadoes, rather than number of tornadoes, reduces the artificial changes that result from modern, more detailed damage surveys (e.g., Doswell and Burgess, 1988). The data set of "significant" tornado days developed by Grazulis (1993) shows a slow increase in number of days with significant tornadoes from the early 1920s through the 1960s, followed by a decrease since that time.

2.7.4 Summary

Based on new analyses since the SAR, it is likely that there has been a widespread increase in heavy and extreme precipitation events in regions where total precipitation has increased, e.g., the mid- and high latitudes of the Northern Hemisphere. Increases in the mean have often been found to be amplified in the highest precipitation rates total. In some regions, increases in heavy rainfall have been identified where the total precipitation has decreased or remained constant, such as eastern Asia. This is attributed to a decrease in the frequency of precipitation. Fewer areas have been identified where decreases in total annual precipitation have been associated with decreases in the highest precipitation rates, but some have been found. Temperature variability has decreased on intra-seasonal and daily time-scales in limited regional studies. New record high night-time minimum temperatures are lengthening the freeze and frost season in many mid- and high latitude regions. The increase in global temperatures has resulted mainly from a significant reduction in the frequency of much below normal seasonal mean temperatures across much of the globe, with a corresponding smaller increase in the frequency of much above normal temperatures. There is little sign of long-term changes in tropical storm intensity and frequency, but inter-decadal variations are pronounced. Owing to incomplete data and relatively few analyses, we are uncertain as to whether there has been any large-scale, long-term increase in the Northern Hemisphere extra-tropical cyclone intensity and frequency though some, sometimes strong, multi-decadal variations and recent increases were identified in several regions. Limited evidence exists for a decrease in cyclone frequency in the Southern Hemisphere since the early 1970s, but there has been a paucity of analyses and data. Recent analyses of changes in severe local weather (tornadoes, thunder days, lightning and hail) in a few selected regions provide no compelling evidence for widespread systematic long-term changes.

Other reports in this collection