/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
/n
image/svg+xml
Figure 2: Switzerland temperature records (JJA) with heatwave of 2003 (red lines)
Figure 1: European weather anomaly in 2003
Figure 4 c,d: Changes in temperature and variability for SCEN compared to CTRL
Figure 4 a,b: Simulated temperature distributions CTRL and SCEN
Figure 5:Precipitation anomalies vs. temperature for a) JJA datab) Climate simulations CTRL (blue) and SCEN (red) Figure 3: Analysis of temperature records for two separate periods of time
Influence of precipitation During the simulations it was found that increased variability occurs mainly in central and eastern Europe (Figure 4d), and that the warm summers of SCEN show signs of drought, nonlinearly amplifying temperature anomalies. To examine this further, summer precipitation and temperature anomalies are plotted against each other (Figure 5), both for long-term statio data (5a) and simulated data (CTRL and SCEN, 5b). Both data sets exhibit a similar relationship between temperature and precipitation anomalies. Fig. 5b demonstrates that in terms of temperature and precipitation the climatic conditions in JJA 2003 were not unlike those simulated by SCEN for the period 2071–2100. The 2003 observation is located near the middle of the SCEN data points (Fig. 5b) — thus, at the end of the century about every second summer could be as warm or warmer (and as dry or dryer) than 2003.
Conclusion Climate models have shown a significant increase of environmental variability This drastically increases the likelihood of summer 2003 (50% increase in standard deviation increases the probability by a factor of about 150). In this process, feedback from precipitation seems to play a crucial role. Simulations The hypothesis of an increasing climate variability was found to be supported by various climate models. In Figure 4, a regional climate model (RCM) driven by a greenhouse-gas scenario representing 2071-2100 conditions (SCEN). The scenario is compared against a control integration covering the period 1961–90 (CTRL). The temperature distribution for CTRL agrees notably well with observations. In the SCEN simulation, the distribution is shifted by 4.6 °C towards warmer temperatures, and also shows a pronounceded widening of its statistical distribution, with the standard deviation increasing by 102% (P < 1%). Influence of climate changeIn order to examine whether the summer 2003 event becomes more likely when considering climate change, the data set is separated into two different time intervals. Figure 3 shows the analysis of the data records for the time intervals 1864-1923 (blue curve) and 1941-2000 (red curve). Results show a mean warming of 0.8 °C and an increased variability; the likelihood of an excess temperature of T'=3 °C increases by about 100 % (Figure 3c). This effect gets even more significant if an even shorter interval is considered, 1990-2002, resulting in a return period of 46,000 years for the 2003 event, which is much less than before, but still high. Temperature analysisTemperature records are typically approximated with a gaussian distribution, introducing mean temperature and its standard deviation (as a measure for environmental variability) as sole parameters of the model. As soon as the distribution is known through measurement, it is possible to examine the extreme heatwave that affected nearly all of Europe during summer 2003 in terms of probability. In Figure 1, the distribution of the European temperature anomaly is shown for the summer of 2003, with respect to the 1961-90 mean. Figure 2 shows temperature records for Switzerland since 1864 for the sommer months (JJA) with fitted Gaussians. The red line indicates the summer 2003 heatwave in this distribution (mean temperature anomaly of T'=5.1 °C), emphasizing its highly unusual nature. In this distribution, the resulting return period of the summer 2003 event is as high as millions of years.
Summary The extreme heatwave experienced in Europe during summer of 2003 was very unlikely to occur due to an increase of mean temperature alone. By also increasing the width of the temperature distribution (i.e., increasing environmental variability), such extreme events become much more likely. Indeed, simulations suggest a rising variability with increasing mean temperature due to feedback through precipitation, leading to the conclusion that by the end of the 21st century, heatwaves like in summer 2003 could become much more frequent. Conclusion Climate models have shown a significant increase of environmental variability This drastically increases the likelihood of summer 2003 (50% increase in standard deviation increases the probability by a factor of about 150). In this process, feedback from precipitation seems to play a crucial role. Conclusion Climate models have shown a significant increase of environmental variability This drastically increases the likelihood of summer 2003 (50% increase in standard deviation increases the probability by a factor of about 150). In this process, feedback from precipitation seems to play a crucial role.
Ulrich Weber
Dion Häfner
Institute of Environmental Phyics, Heidelberg
Poster design by
Students' work @ IUP
The role of increasing temperature variability in European summer heatwaves
Based on a paper by Schär et al.1
1 Published in Nature 427, 332-336 (Jan. 2004). Scan code for more information: