globalchange  > 气候变化事实与影响
DOI: doi:10.1038/nclimate2468
论文题名:
Dramatically increasing chance of extremely hot summers since the 2003 European heatwave
作者: Nikolaos Christidis
刊名: Nature Climate Change
ISSN: 1758-1084X
EISSN: 1758-7204
出版年: 2014-12-08
卷: Volume:5, 页码:Pages:46;50 (2015)
语种: 英语
英文关键词: Attribution ; Climate-change impacts
英文摘要:

Socio-economic stress from the unequivocal warming of the global climate system1 could be mostly felt by societies through weather and climate extremes2. The vulnerability of European citizens was made evident during the summer heatwave of 2003 (refs 3, 4) when the heat-related death toll ran into tens of thousands5. Human influence at least doubled the chances of the event according to the first formal event attribution study6, which also made the ominous forecast that severe heatwaves could become commonplace by the 2040s. Here we investigate how the likelihood of having another extremely hot summer in one of the worst affected parts of Europe has changed ten years after the original study was published, given an observed summer temperature increase of 0.81 K since then. Our analysis benefits from the availability of new observations and data from several new models. Using a previously employed temperature threshold to define extremely hot summers, we find that events that would occur twice a century in the early 2000s are now expected to occur twice a decade. For the more extreme threshold observed in 2003, the return time reduces from thousands of years in the late twentieth century to about a hundred years in little over a decade.

Despite the slowdown in the global mean temperature increase since the late 1990s (refs 7, 8, 9), hot temperature extremes have continued to warm on both global and regional scales10, 11. Severe heatwaves in the past decade such as the ones in Moscow in 2010 (refs 12, 13), Texas in 2011 (ref. 14) and the Australian ‘angry summer’ of 2012–201315 were characterized by long duration, large spatial extent and catastrophic impacts. Research on event attribution aims to identify the drivers of such extreme events and determine how possible causes such as human influence on the climate might have altered their odds16, 17, 18. In this paper we revisit the first study of this kind6 that investigated the 2003 European heatwave and carry out a new analysis that is now extended to the present day. As in the original study, we concentrate on summer temperatures (average over June–August) in the land area bounded by 10° W–40° E and 30°–50° N, which, among a number of pre-defined climatic regions19, was mostly affected by the 2003 heatwave. The use of a pre-defined region helps minimize selection bias. The selected area largely includes the countries where heat-related mortality peaked (France, Germany and Italy), but is more extensive.

Summer temperature time series constructed with the best estimate of the CRUTEM4 observational data set20 show that the 2003 record was subsequently broken in 2012 (Fig. 1). Although a hot summer in the region cannot be directly linked to heatwave damage (for example, heatwave impacts in 2012 were less notable than in 2003), as records are being broken in a warming climate, hotter summers are generally expected to be associated with more severe impacts. Our analysis examines how the likelihood of very warm summers in the region has changed between the time the actual event occurred, using estimates of the forced climatic change in the 1990s, and the present day, using the more recent decade of 2003–2012. Interestingly, the recent decade is 0.81 K warmer than the 1990s (Fig. 1), indicating a shift of the summer temperature distribution towards higher values, which would increase the chances of new record-breaking temperatures.

Figure 1: Time series of the summer mean temperature anomaly relative to 1961–1990 in region 10° W–40° E, 30°–50° N.
Time series of the summer mean temperature anomaly relative to 1961-1990 in region 10[deg] W-40[deg] E, 30[deg]-50[deg] N.

Observed time series (red line) and the range of temperature anomalies from simulations with the seven CMIP5 models used in the analysis that include all forcings (red area, a), and from simulations that include natural forcings alone (blue area, b). The black line represents the time series of the mean of the model simulations. The observed mean anomalies in decades 1990–1999 (0.77 K) and 2003–2012 (1.58 K) are marked by the green horizontal lines.

Optimal fingerprinting.

Decadal summer temperatures averaged over the reference region in consecutive decades during the analysis period are estimated with data from CRUTEM4 and simulations with ALL and NAT forcings. Temperatures are expressed as anomalies relative to the 1961–1990 base period, that is, the same as in the CRUTEM4 data set. Using an earlier base period to approximate the pre-industrial climate was found to introduce uncertainty in the results without, however, changing the main conclusions of our study (Supplementary Information). The model data are first masked by the observations to include the same spatial coverage. For each experiment (ALL and NAT) we take the mean of all available simulations for each model and then compute the average of the ensemble means of the seven models used in the analysis23. The temperature time series constructed from this multi-model ensemble are organized into vectors xALL and xNAT and the observational time series are similarly organized into a vector y. The anthropogenic response, xANT, is approximated by the difference xALL − xNAT. Optimal fingerprinting decomposes the observations into the forced response and unforced variability:

The effect of internal variability in the observations and the simulated response is represented by the noise terms u0, uANT and uNAT. We carry out two analyses: the first covers the period 1920–1999 as in a previous study of the 2003 heatwave6 (that is, uses decadal temperature anomalies in eight consecutive decades 1920–1929, …, 1990–1999); the second spans years 1923–2012 (that is, uses anomalies in decades 1923–1932, …, 2003–2012). The variance–covariance matrix of the noise terms is derived from independent segments of equal length to the analysis period extracted from the control simulations of the unforced climate, processed and organized into vectors (Y) in the same way as the observations:

where n denotes the number of control segments and N the vector length. We use 75 segments for the 1920–1999 analysis and 68 segments for the 1923–2012 analysis. The control simulations are also used to estimate the uncertainty in the scaling factors and test whether the modelled internal variability and the regression residuals are consistent22, which is indeed found to be the case in our study. As common in optimal fingerprinting, the analysis is carried out in a transformed space defined by the leading empirical orthogonal functions of the internal variability estimated from the control segments. We retain the seven leading empirical orthogonal functions, which explain more than 90% of the observed variability. Each analysis outputs the best estimate and the 5–95% uncertainty range of the βANT and βNAT scaling factors. A signal is detectable, if the uncertainty range of the corresponding scaling factor does not encompass zero.

Temperature distributions and the likelihood of a threshold exceedance.

Optimal fingerprinting provides observationally constrained estimates of the response to anthropogenic and natural forcings, as well as the response to all forcings (from the linear combination ANT + NAT). Our fingerprinting algorithm yields 105 possible estimates of each scaling factor (by sampling each percentile of the distribution 1,000 times) representing the uncertainty in their value, from which we get 105 estimates of the temperature change in the reference region attributed to different forcings. By extracting the last decade we obtain samples of the climate response to all and natural forcings in the 1990s (from the 1920–1999 analysis) and the most recent decade of 2003–2012 (from the 1923–2012 analysis). We then combine these response estimates with a sample of ~3,500 summer temperature anomalies of individual years in non-overlapping decades from control experiments (representing the effect of internal variability) and construct the temperature distributions with and without the effect of human influence in the early 2000s and the present-day (Fig. 2c).

The probability of warm summers and the associated uncertainty are estimated as follows: we add each of the 105 estimates of the forced decadal temperature response to all the annual temperature estimates for the unforced climate and compute 105 estimates of the probability of exceeding a threshold using order statistics, or, the generalized Pareto distribution, if the threshold lies in the tail of the distribution. The resulting sample provides the 50th percentile (best estimate) and the 5th and 95th percentiles of the probability of an extreme event (that is, regional summer temperature above the threshold). The return time is computed as the inverse of the probability. When the threshold lies in the far tail of the temperature distribution, the probability estimates are very small and the corresponding return times are of the order of a thousand years or more (based on the 5th percentile). The precise return time of these extremely rare events cannot be reliably estimated because of the insufficient sample size and we report only a conservative estimate of the order of magnitude of the return time.

  1. Bindoff, N. L. et al. in Climate Change 2013: The Physical Science Basis (eds Stocker, T.et al.) 867–952 (IPCC, Cambridge Univ. Press, 2013).
  2. Stott, P. A. et al. in Climate Science for Serving Society: Research, Modeling and Prediction Priorities (eds Asrar, G. R. & Hurrell, J. W.) (Springer Science + Business Media, 2013).
  3. Schär, C. et al. The role of increasing temperature variability in European summer heatwaves. Nature 427, 332–336 (2004).
  4. Beniston, M. The 2003 heat wave in Europe: A shape of things to come? An analysis based on Swiss climatological data and model simulations. Geophys. Res. Lett. 31, L02202 (2004).
  5. Robine, J-M. et al. Death toll exceeded 70,000 in Europe during the summer of 2003. C. R. Biol. 331, 171–178 (2008).
  6. Stott, P. A., Stone, D. A. & Allen, M. R. Human contribution to the European heatwave of 2003. Nature 432, 610–613 (2004).
  7. Meehl, G. A. et al. Model-based evidence of deep-ocean heat uptake during surface-temperature hiatus periods. Nature Clim. Change 1, 360–364 (2011).
  8. Fyfe, J. C., Gillett, N. P. & Zwiers, F. W. Overestimated global warming over the past 20 years. Nature Clim. Change 3, 767–769 (2013).
  9. Kosaka, Y. & Xie, S. Recent global-warming hiatus tied to equatorial Pacific surface cooling. Nature 501, 403–407 (2013).
  10. Seneviratne, S. I., Donat, M. G., Mueller, B. & Alexander, L. V. No pause in the increase of hot temperature extremes. Nature Clim. Change 4, 161–163 (2013).
  11. Sillmann, J., Donat, M. G., Fyfe, J. C. &
URL: http://www.nature.com/nclimate/journal/v5/n1/full/nclimate2468.html
Citation statistics:
资源类型: 期刊论文
标识符: http://119.78.100.158/handle/2HF3EXSE/4912
Appears in Collections:气候变化事实与影响
科学计划与规划
气候变化与战略

Files in This Item:
File Name/ File Size Content Type Version Access License
nclimate2468.pdf（529KB）期刊论文作者接受稿开放获取
CC BY-NC-SA
View Download

Recommended Citation:
Nikolaos Christidis. Dramatically increasing chance of extremely hot summers since the 2003 European heatwave[J]. Nature Climate Change,2014-12-08,Volume:5:Pages:46;50 (2015).
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[Nikolaos Christidis]'s Articles
百度学术
Similar articles in Baidu Scholar
[Nikolaos Christidis]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[Nikolaos Christidis]‘s Articles
Related Copyright Policies
Null
收藏/分享
文件名： nclimate2468.pdf
格式： Adobe PDF
此文件暂不支持浏览
所有评论 (0)
暂无评论
 

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.