This paper investigates the recent spate of record-breaking warm seasons that have affected Switzerland in less than a decade and compares the seasonal statistics to those simulated for a “greenhouse-gas” climate by the end of the 21st century. The peaks of minimum and maximum temperatures observed during some the record seasons enter well into the 25%–75% temperature quantile range for the scenario climate simulated by a set of regional climate models. The anomalously warm seasons allow a “preview” of conditions that may occur with greater frequency in the future. The use of current data as a form of proxy for the future enables an assessment of the possible impacts on the natural and socio-economic environments, and can help in considering possible adaptation strategies to reduce some of the associated risks of climatic change.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Over the past few years, large positive departures of temperatures from their mean values have become commonplace in many parts of the world. In Switzerland, where the exceptional heat wave of the summer of 2003 still remains in public memory, new temperature records have been broken in a recent past, in particular the warmest July on record in 2006, the warmest fall in many parts of the country in 2006, and one of the warmest winters on record in 2006–2007. The spate of anomalously warm weather over a relatively short time span has inevitably caught public attention, with a legitimate questioning as to whether this constitutes further evidence of global warming. Clearly, mean temperatures are higher than they were several decades ago. It can thus be expected that with increasing average temperatures the upper extremes of temperature probability density function (PDF) can reach levels rarely experienced under cooler conditions. In some parts of the country, temperatures have increased during the 20th century by 1.5°C or more, particularly at high elevations, as reported by Beniston .
 In the wake of the spate of recent record warm months or seasons, many publications have attempted to identify the causal mechanisms underlying such events [e.g., Beniston, 2004a, 2005; Schär et al., 2004; Luterbacher et al., 2007]. While the enhanced greenhouse effect is certainly partially responsible for the long-term increasing trends in both maximum and minimum temperatures, it is more difficult to attribute the recent large positive temperature anomalies solely to the greenhouse effect. Enhancement of low-level atmospheric temperatures by sensible heat fluxes at the surface associated with dry soils has recently attracted much interest [Seneviratne et al., 2006]. Vautard et al.  show that northward propagation of drought conditions that begin in the Mediterranean region during winter generate systematic preconditions contributing to the intensity and/or persistence of heat waves in Europe. The positive feedback effect of dry soils also helps explain the severity of a heat wave.
 This paper revisits the Swiss observational record and compares the data of a greenhouse climate simulated by regional climate models (RCMs). The extremes associated with some of the recent record events will be seen to enter well into the medium to upper quantiles of future minimum and maximum temperature PDFs. By analyzing the impacts that these extreme conditions currently impose on environmental (e.g., the cryosphere, hydrological systems, and the biosphere) and socio-economic systems (e.g., human health, agriculture, hydropower), analogies with the future can be drawn for timely decision-making in terms of risk-reduction and adaptation strategies.
 The Swiss weather service (MeteoSwiss) maintains a dense network of observation stations and manages the acquired daily data in the form of a digital data base. Switzerland is an ideal locale for analyzing long time-series, because the data is generally homogenous [Begert et al., 2003] and available in digital form from 1901. The Swiss data has been used in many statistical studies of climate and climatic change in Switzerland [e.g., Beniston, 2004a], and the climatological stations span a range of altitudes from 200–3600 m above sea level. Jungo and Beniston  have shown through cluster analysis that, despite individual site heterogeneity, there is a close resemblance between climatological variables in different parts of Switzerland, particularly on an altitudinal basis, such that even if some sites may be biased by local characteristics, the long-term trends are in general agreement among all stations.
 The observational record used in this investigation comes from Swiss observational stations, distributed horizontally and vertically over the territory. For the sake of conciseness, only two sites representative of low elevations (Basel, in NW Switzerland at 369 m above sea level) and high elevations (Saentis at 2500 m in NE Switzerland) will be discussed, and because they have proven their quality in a number of previous studies [e.g., Beniston and Stephenson, 2004]. The conclusions for this limited sample are however consistent with the findings at the other Swiss sites.
 Global and regional climate models allow an insight into possible climate futures as a function of greenhouse gas emissions and concentrations. The IPCC scenarios developed by Nakićenović et al.  are based on various emission pathways as a function of economic and population growth, societal and technological choice, e.g., the speed with which energy sectors reduce their dependency on fossil fuels, and shifts in deforestation and land-use trends. In this paper, results for the high emissions IPCC SRES A2 scenario only will be discussed to emphasize that even for one of the “worst-case” climatic change scenarios, current extreme temperatures in the Alps exhibit close statistical resemblance to those of the A2 climate. CO2 levels for this scenario reach about 800 ppmv by 2100 (i.e., three times their pre-industrial values), and are in the upper range of climate futures published by the Intergovernmental Panel on Climate Change .
 The European Union project “PRUDENCE” (http://prudence.dmi.dk) and the ongoing “ENSEMBLES” project (http://www.ensembles-eu.org/) have used a suite of regional climate models (RMCs) to simulate climatic change over Europe for the last 30 years of the 21st century. Shifts in a number of key climate variables can be assessed through comparisons with the reference 1961–1990 climate. Déqué et al.  report a generally close agreement between many of the model results. For this study, four different RCMs have been employed: the Swiss CHRM; Danish HIRHAM; Swedish RCAO; and Italian RegCM models (see Beniston et al.  for further references). Initial and boundary conditions for the RCMs are derived from a two-step process involving the fully-coupled ocean-atmosphere general circulation model of the UK Hadley Centre HadCM3 [Johns et al., 2003], followed by the higher-resolution atmospheric HadAM3H model [Pope et al., 2000] driven by HadCM3.
 Raw data from RCM simulations has been used in this study because the bias is very small between RCM results and observations for the control period, at least for the region studied, as previously discussed by Beniston [2004a, Figure 4]. Basel and even many locations within the Alps thus constitute one of the geographical locales where raw RCM data can be used with a reasonable degree of confidence.
Figure 1 illustrates the evolution of seasonal maximum temperatures (Tmax) from 1901–2006 at the high-elevation site of Saentis; superimposed on Figure 1 are the 10 warmest seasons of the series, to highlight the clustering of record-warm seasons that occurs in the most recent part of the record, in particular during summer and winter. Summer and autumn show strong increases in temperature, and while springs have not generally been as warm as they were in the 1940s and 1950s, 2007 has set another record. Winter temperatures peaked in the late 1980s and show a cooling trend that was abruptly reversed during the 2006–2007 winter, which in many locations, including Saentis was the warmest on record. Figure 1 also shows the large year-to-year variability of mean seasonal temperatures, where these can undergo changes by as much as ±5°C from one year to the next. In the past, outbreaks of anomalously warm seasons have occurred on numerous occasions, but when such “flares” of temperature are superimposed on the general century-scale temperature trends, it is not surprising that in the most recent part of the record, temperatures exceed earlier record high values. Figure 2 highlights the seasonal distribution of minimum temperatures (Tmin) at the low-elevation site of Basel, the clustering of the 10 most extreme seasons is not always the same as for Saentis, but in both cases it is seen that in less than a decade, new temperatures records have been set for all for seasons. The differences in the sequence of record-high years between Tmin and Tmax, and between low and high elevations, can be attributed to site differences, e.g., the presence or absence of inversion layers in winter, differences in cloudiness, and snow cover and its feedbacks with the atmosphere. In less than a decade, new temperatures records have been set for all for seasons.
 It was shown by Beniston [2004a] and Schär et al.  that the extreme European summer heat wave of 2003 may be a forerunner of summers to come later this century. In view of the spate of record warm seasons in Switzerland in recent years, it is of interest to assess how close some of the latest events come to resembling statistically those that are expected in a future climate.
 In order to address this issue, RCM data from the four models mentioned in the preceding section were used to compute an inter-model average for the model grid-points closest to Basel and Saentis. Using the daily data simulated by the models, it was possible to compute the daily and monthly means, and the lower and upper quantiles of the Tmin and Tmax distributions.
Figure 3 illustrates the particular case of the Tmin distribution at the Basel grid-point, for the 30-year period from 2071 to 2100. Superimposed upon the scenario climate Tmin statistics are three sets of curves: the bold line is the observed monthly-average Tmin at Basel for the 1961–1990 period, the bold dashed line illustrates the monthly values for the year 2003, and the bold dotted curve provides the daily Tmin values during the exceptional 2003 heat wave from June–August. The following broad observations are summarized below:
 1. The average monthly values for the reference climate of the 20th century are at or below the 10% quantile limit of the A2 scenario climate.
 2. Minimum temperatures between current and future climates increase from 3.8°C to 4.9°C on average according to the month.
 3. For a particular year such as 2003, monthly mean temperatures make broad incursions into the middle range of the future Tmin quantiles, i.e., for example June and August 2003 temperatures are warmer than the 50% quantile of the A2 climate.
 4. Certain days of the 2003 heat wave, particularly in June and August, are seen to largely exceed the upper quantiles of the scenario climate, i.e., isolated days of the 2003 summer are as warm or warmer than the 90% quantile of the future climate. In other words, events such as the 2003 heat wave, have been as intense as what would constitute the upper range of the Tmin PDF in the scenario climate.
 In order to push this analysis further, a count of the number of incursions of observed temperatures into the quantile range of future climate has been undertaken to assess whether any significant trends are discernible. Figure 4 provides an example of such counts, where the exceedances of summer Tmax at the Saentis site beyond the 10%, 50% and 90% quantiles of future climate have been plotted from 1901–2006. Values of Tmax beyond the 10% quantile level are a fairly common occurrence, averaging about 80% of the time and exhibiting a slightly increasing trend. The trend is much more conspicuous for the number of counts beyond the 50% quantile of future climate, and even ignoring the strong bias imposed by the hot summer of 2003, the exceedances have doubled since 1901 from less than 20% of the summer period (15 days per summer) to more than 40% today (over 35 days). This is to be expected, since in the intervening period mean summer temperatures have warmed by about 1.5°C at this site; the threshold of the 50% quantile in the scenario climate is 13.1°C for the three summer months, which corresponds roughly to the 97% quantile of summers for 1901–1910 but “only” to the 85% quantile for 1997–2006. Figure 4 shows that some summers in the late 1940s exceeded the 90% quantile threshold of the A2 climate for a small number of days; in 2003, however, the exceedance was more than 10%. The number of counts beyond the 90% quantile for a decade centered on the warmest 10-year period prior to today, i.e., 1947–1956 is 13 at Saentis, while in the 10 years since 1997 this has risen to 23. Similar conclusions can be reached for the other seasons at all other time-series investigated in this study, but are not shown here for reasons of conciseness.
Table 1 shows for Tmin and Tmax for each season at Basel and Saentis, how frequently such events may occur in the future based on the comparisons with future quantiles of temperature. It is seen that the record seasons observed in the last decade could occur 50–70% of the time in the A2 climate. This clearly suggests that Switzerland has already experienced conditions that would occur at least 1 year in 2 by 2100.
Table 1. Frequency of Occurrence of the Conditions Observed in the Record-High Seasonal Temperatures in the Recent Past When Projected Into a Future Scenario Climate
 The statistics that underlie Figure 4 and Table 1 raise the question as to whether climate model projections are not underestimating observed trends. This is a question that has recently attracted much attention [e.g., Scheffer et al., 2006; Torn and Harte, 2006], and is perhaps related to the failure by many models to reproduce the strength of certain positive feedback mechanisms. It is not the purpose of this paper to enter into the details of these processes, but the results portrayed here, already at the high end of the quantiles for an A2 climate, may need to be put into the perspective of an even warmer climate than that resulting from A2 emissions.
 The discussion in the previous section emphasizes the point that the recent spate of anomalously warm seasons observed in Switzerland may well be forerunners of climatic conditions that regional climate models are projecting for the latter part of the 21st century. While there is cause for concern in view of the observed environmental, social and economic impacts during the events themselves and often long after (e.g., in the case of ground-water deficits or the inertia of melting glaciers), the recent extremes have provided a “preview” of future events and impacts.
 A succinct summary of the manner in which very warm seasonal anomalies can affect the alpine region are outlined below [e.g., Beniston, 2003, 2004b]:
 1. Winters: the 2006–2007 season has been characterized by a sharp reduction in snow amount and duration at elevations below 2,000 m. Many low to medium sites suffered from significant shortfalls in revenue resulting from a reduction in the number of days where skiing is possible by a factor of 2–3. Warm winters at lower elevations also reduce the period of dormancy that many plants require, and often fail to destroy the precursor condition for pest and disease outbreaks later in the year.
 2. Springs: anomalously warm springs leads to early runoff in hydrological basins. Further downstream, if sustained precipitation occurs, this can lead to flood situations, as already experienced in 1995 along the Rhine River. Warm springs also enable an early start to the vegetation period of certain species, which can then be disrupted with the return of a cold spell after the snow has melted.
 3. Summers: possibly the most adverse social and environmental impacts occur during the summer. Summer heat waves observed in Switzerland and many parts of the alpine domain result in acute problems of public health, large losses for agriculture, sharp reductions in river discharge (compounding problems of water use in the agricultural and energy sectors), a decline of glacier mass and degradation of permafrost (with collateral effects on slope instability events).
 4. Autumns: warm, dry conditions can prolong droughts that may have begun earlier in the year, with further drops in river discharge. A persistently warm autumn also delays the time of the first winter snowfalls.
 Without some form of economic or technological adaptation, such impacts will increasingly be felt as climate warms. The lessons learned from the current spate of anomalously warm seasons can thus be used as a proxy for future seasons and the impacts they generate. Impacts analyses can help in devising adaptation measures to minimize the associated risks by the end of the century. For example, strategies for avoiding conflicting uses of water between hydropower and irrigation in summer, or optimization of tourism across all seasons in the Alps taking into account climatic change, can be formulated on the basis of these recent previews of future climate.
 This paper has provided a brief overview of the anomalously warm seasons that have broken records in Switzerland in a very recent past. The spate of record-breaking seasons is not surprising in view of the rise of mean temperatures that have occurred since the beginning of the selected time-series in 1901. However, when comparing the warm-season data with monthly or seasonal data in the future, it was shown that the very warm seasons observed in current climate already mimic what may occur with increasing frequency in a greenhouse climate. Indeed, the present investigation has shown that Switzerland has experienced climatic conditions that models suggest may occur at least 50% of the time by 2100, with a number of impacts on environmental and socio-economic systems that would be replicated whenever such temperature thresholds are exceeded. By using the current-climate data as a form of proxy for future conditions, risks associated with some of the more adverse impacts could be reduced by putting into effect appropriate adaptation schemes in a timely manner.