Réf. Schmidli & Frei 2005 - A

Référence bibliographique complète

SCHMIDLI, J., FREI, C. 2005. Trends of heavy precipitation and wet and dry spells in Switzerland during the 20th century. International Journal of Climatology, 25, 753–771.

Abstract: The low-frequency variability of heavy precipitation and drought conditions is investigated for 104 rain-gauge stations in Switzerland covering the 20th century. This constitutes an exceptionally dense data set of centennial time series. The investigation is based on a wide range of daily and multi-day precipitation statistics encompassing basic characteristics, measures of heavy precipitation and indices of typical and extreme dry and wet spells. Two different methods of trend analysis and statistical testing are applied, depending on the data nature of the statistic. Linear regression is used for statistics with a continuous value range, and logistic regression is used for statistics with a discrete value range. The trends are calculated on a seasonal basis for the years 1901–2000.
A clear trend signal is found for winter and autumn, with a high number of sites with a statistically significant trend. In winter, significant increases are found for all statistics related to precipitation strength and occurrence. The centennial increase is between 10 and 30% for the high quantiles and the seasonal 1 day to 10 day extremes. In autumn, statistically significant increases are found only for the statistics related to heavy precipitation, whereas precipitation frequency and spell-length statistics show little systematic change. Although the winter trend signal is strongest in northern and western Switzerland, the autumn trend signal is more uniform. In spring and summer, the heavy precipitation and the spell-duration statistics did not show statistically significant trends. Sensitivity tests indicate that the winter and autumn trends are robust with respect to inhomogeneities in the rain-gauge time series.

Mots-clés
Trend analysis; Climate variation; Precipitation extremes; Dry and wet spells; European Alps; Switzerland

Organismes / Contact

Atmospheric and Climate Science ETH, Winterthurerstrasse 190, 8057 Z¨urich, Switzerland - juerg.schmidli@env.ethz.ch


(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa
Precipitation      

Pays / Zone
Massif / Secteur
Site(s) d'étude
Exposition
Altitude
Période(s) d'observation
Switzerland         20th century

(1) - Modifications des paramètres atmosphériques
Reconstitutions
 
Observations

Several previous studies have investigated precipitation trends in Switzerland and the Alpine region. For mean precipitation, a statistically significant increase was found in winter in the northern and western parts of the Alps, including northwestern Switzerland, where the increase amounts to 15–30% in the 20th century (Widmann and Schär, 1997; Schmidli et al., 2002; Begert et al., 2005). In contrast, decreasing mean precipitation was found in the southern Alps in autumn, but southern Switzerland appears to be marginally affected (Brunetti et al., 2001; Schmidli et al., 2002). Only selected diagnostics of daily precipitation statistics have been examined so far: Frei and Schär (2001) find a statistically significant increase in the frequency of intense precipitation events (return period: 30 days) in winter and autumn for northern Switzerland (see also Courvoisier (1998)) during the 20th century. Rebetez (1999) finds that episodes of wintertime drought in southern Switzerland have increased in duration during the 20th century. Brunetti et al. (2001, 2004) find that the negative trend in mean precipitation in northern Italy is associated with a decrease of the wet-day frequency (in most seasons, but strongest in winter) and a slight increase of the wet-day intensity (all seasons). Several continental-scale analyses, covering a shorter time period though, conform with these regional trend patterns (Alpert et al., 2002; Klein Tank and Können, 2003; Haylock and Goodess, 2004).

In this study the authors have analysed precipitation data from a dense rain-gauge network in Switzerland for trends in daily precipitation statistics over the 20th century. The statistics encompass basic daily characteristics, measures of heavy precipitation and indices of wet and dry spells, including anomalously long dry and wet periods. Two different methods of trend analysis and statistical assessment were applied, depending on the data nature of the indices.

Statistic

Winter

  

Spring

 

Summer

 

Autumn

N

E

W

S

N

E

W

S

N

E

W

S

N

E

W

S

MEA

+

+

+

 

+

   

   

+

+

+

FRE

+

+

+

 

   

   

   

INT

+

+

+

 

+

   

+

       

+

+

+

+

Heavy precip

+

+

+

 

+

   

+

       

+

+

+

+

Wet spells

+

+

+

       

     

     
Dry spells

 

     

+

     

   

+

Table V. Summary of trend results for the 20th century daily precipitation statistics. The table contains only entries for indices and categories for which the trends for most stations in a region (and indices within a category) have the same sign. More specifically, a plus (minus) indicates that the lower (upper) quartile of the trend values is larger (smaller) than zero. A bold sign denotes that at least one-third of the stations have a significant trend at the 5% level

A compact summary of the trend results is provided in Table V. The analysis has identified spatially coherent and statistically significant trends for most indices of heavy precipitation. Increasing trends are found for winter and autumn and for the northern, western and eastern parts of the country. Yet there are differences in the trend characteristics between these two seasons. In winter, the change in high quantiles of daily precipitation is associated with an increase in the mean, the wet-day intensity and wet-day frequency and wet-spell length (though not statistically significant for the latter two). The centennial increase is between 10 and 30% for high quantiles and seasonal 1 day to 10 day extremes, and between 20 and 80% for the frequency of high-quantile exceedance. In autumn the situation is more complex. Here, the increase at high quantiles is associated with an increase of wet-day intensity, whereas wet-day frequency and mean spell-length have not changed significantly. In spring and summer the heavy precipitation indices did not show statistically significant trends and there was no evidence for long-term trends in the mean and seasonal extreme dry-spell duration. However, these latter results should not be interpreted as the absence of trends. Especially for rare extremes, the detectability is quite limited. For several of the extreme indices considered, a trend estimate of 20% per 100 years and more is required for statistical significance, even with the comparatively long observation period (100 years) available in this study.

The overall pattern of the present trend analysis conforms with findings from other regional and continental-scale precipitation analyses. Birsan et al. (in press) analysed mean daily streamflow records from undisturbed watersheds in Switzerland for the 20th century. They found an increase in annual streamflow mostly due to increases in winter and spring season runoff. Most of the changes, however, have occurred in the winter season, with significant increases in the winter maximum streamflow. They related the observed increases to changes in air temperature, snow cover, and precipitation. These results also tie in with an observed decrease in snow-days in the Swiss Alps (Scherrer and Appenzeller, 2004). At the larger European scale, from a European daily dataset (151 records) Klein Tank and Können (2003) find an increase in the annual number of moderate and very wet days (the latter corresponding to the NL95 index) during 1946–99. Using a more comprehensive European dataset (470 stations, 1958–2000), Haylock (2003) finds a coherent region over northern and central Europe with an increase of the 90% quantile of daily precipitation and the maximum 5 day rainfall in winter. His diagrams suggest that the trend magnitude is particularly large in the vicinity just north of the Alpine ridge. Similarly, a large proportion of stations showed statistically significant increases in high precipitation quantiles for autumn. In a regional analysis for Germany (more than 600 stations, just north of Switzerland), Bárdossy and Hundecha (2003) find (again for the second half of the 20th century) an increase in heavy precipitation indices in winter and, at lower significance, also in spring and autumn. The same study identifies a prevalence of decreasing trends for heavy precipitation indices in summer. With regard to the dry-spell characteristics, both analyses (Bárdossy and Hundecha, 2003; Haylock, 2003) show a tendency for increases in the maximum number of consecutive dry days in summer and a decrease in autumn. The marginal changes found for dry spells in this study may be due to regional differences, trend estimation errors and/or differences in the period considered. The present analysis supports a slowly emerging global picture of increasing trends in heavy precipitation during (parts of) the 20th century at middle to high northern latitudes during the winter half year (see also Groisman et al. (1999) and Knight and Karl (2001), Frich et al. (2002)).

Especially in winter, there is qualitative similarity between model projections [see below] and observed trends in Switzerland, and it is tempting to speculate that global climate change over the 20th century has contributed to the observed trend. For winter, in the region of the Alps, regional climate models exhibit changes in precipitation intensity of 5–15% by the end of the 21st century. These values are determined from several regional climate models from the PRUDENCE project, including the regional climate models of Christensen and Christensen (2003), Vidale et al. (2003) and Räisänen et al. (2004). The observed trend for the 20th century is of the same order of magnitude. The global mean temperature increase was about 0.6 °C in the 20th century, compared to more than 3 °C in the general circulation model boundary conditions used for the regional climate model projections. A detailed statistical comparison is yet to be conducted, but it appears that the observed trends in wintertime heavy precipitation are quite large compared with the global warming sensitivity of current climate models for this region. An interesting question, therefore, is to what extent the observed trends and decadal variations in the large-scale circulation regime over the North Atlantic (e.g. Hurrell, 1995) have contributed to the precipitation trends in the Alpine region.

Modélisations

Several studies from the newest generation of global and regional climate models have been published recently with scenarios of European climate change for the late 21st century (e.g. Durman et al., 2001; Semenov and Bengtsson, 2002; Voss et al., 2002; Christensen and Christensen, 2003; Huntingford et al., 2003; Räisänen et al., 2004; Schär et al., 2004). Although the results for precipitation are quantitatively variable and show differences in the subcontinental pattern of the change, there is a clear tendency for an increase in wintertime mean precipitation and a concomitant increase of high quantiles in daily precipitation. This pattern of change is found in the models typically north of 45 °N (i.e. including northern Switzerland). As for summer, most of the models exhibit a decrease in mean precipitation over most parts of the continent (except Scandinavia), which goes along with a decrease of precipitation frequency and more extended drought periods. Despite the decrease for mean precipitation, many model studies find little change or even an increase in quantiles of heavy precipitation for summer and autumn (e.g. Voss et al., 2002; Christensen and Christensen, 2003).

Hypothèses
 

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

In this study the authors investigate long-term variations and trends of precipitation extremes during the 20th century in Switzerland. For this purpose, daily precipitation data are analysed from a comparatively dense network of more than 100 rain gauges, all of which have been continuously operated since 1901. The trend analysis is based on a wide range of diagnostics, covering basic precipitation statistics, single to multi-day heavy precipitation events, spells of intense precipitation and anomalous dry spells. The authors undertake a trend analysis for a range of diagnostics with different data characteristics. To account for the different nature of the diagnostics, they apply two different statistical methods for trend estimation and statistical testing: linear regression and a non-parametric trend test for diagnostics with continuous data (e.g. precipitation amounts) and logistic regression for discrete data (e.g. number of events above a threshold).

The dataset for this study is composed of daily precipitation series at 104 rain-gauge stations in Switzerland. The station set embraces all Swiss rain gauges for which a continuous and complete daily record is available throughout the 100 year period 1901–2000. The data were kindly provided by the Swiss Federal Office of Meteorology and Climatology (MeteoSwiss, Zürich), which is responsible for the operation and quality control of the stations. The station sample is similar to earlier trend analyses for precipitation in Switzerland (Widmann and Schär, 1997; Frei and Schär, 2001), except for some recently discontinued stations. With a typical interstation distance of 20 km, the network considered constitutes a long-term observation system with exceptional density.

Although gross errors in the measurements have been carefully corrected through quality control procedures (testing for spatial continuity) by the original data provider, the instrumental changes, relocations and variations in observation practice have inevitably resulted in inhomogeneities in some of the records. A systematic test for and correction of inhomogeneities was undertaken by MeteoSwiss for a subset of nine stations (Begert et al., 2005). These were carefully selected to cover the range of climatic zones in Switzerland. The homogenized records from these nine stations were used in the present study. In these cases, the monthly correction factors derived from the homogenization procedure were simply applied to the daily precipitation totals. The procedure of homogenization is described in detail in Begert et al. (2005). The only widespread and systematic inhomogeneity evident from the homogenization exercise was that of converting conventional rain gauges into automatic stations (19 stations of the present set). The difference between the conventional and the tipping-bucket device amounts to about 5% in the annual mean, with a larger difference in winter and a smaller difference in summer. Therefore, there is a subset of 13 stations (six where homogenized) for which a spurious negative trend component is expected. However, among the remaining 82 uncorrected stations the authors expect a balance of positive and negative biases from inhomogeneities. Despite the possibility of inhomogeneities in some of the study records, theey decided to keep the full set of long-term records in order to maintain the full spatial detail of the network and to minimize random noise in the trend analysis. In their interpretations, they focus on the results of the entire station sample, compare these with the subsample of homogenized records, and not overemphasize results at single, possibly compromised, stations.

The set of indices is listed in Table I. They are grouped into four different categories: basic, heavy precipitation, wet spells, and dry spells. Category ‘basic’ is not necessarily indicative of extremes, but mean precipitation, wet-day frequency and wet-day intensity are useful statistics for interpretation. A threshold of 1 mm is chosen for wet days. Category ‘heavy precipitation’ encompasses several indices defined in terms of threshold exceedances. NLNN, for example, represents the number of days where station- and season-specific quantiles (the 90% and 95% quantiles) are exceeded. The quantiles represent long-term conditions and were determined from the base period 1961–90. Quantiles were determined from the wet-day precipitation amounts; hence, Q90 corresponds to the precipitation amount that is exceeded every 10th wet day on average. XND is the maximum precipitation amount measured on N (1, 3, 5, 10) consecutive days of the year. Finally, the ‘wet-spell’ and ‘dry-spell’ categories represent indices characteristic of the duration of consecutive dry- and wet-day sequences. Mean wet-day persistence, for example, represents the probability of a wet day following a wet day, and similarly for dry days. All indices are calculated individually for all stations and seasons, resulting in annual time series for 1901–2000.

Name

Description

Trend

MEA
FRE
INT

QNN
N10MM
NLNN
FL90
XND

PWW
XCWD
WSMEA

PDD
XCDD
DSMEA

Mean precipitation
Frequency of wet days (precipitation ≥1 mm)
Precipitation intensity (mean wet-day precipitation)

NNth percentile of wet-day amounts (NN = 90, 95)
No. of events ≥10 mm
No. of events > long-term NNth percentile (NN = 90, 95)
Fraction of total precipitation above long-term 90th percentile
Maximum N-day total precipitation (N = 1, 3, 5, 10)

Mean wet-day persistence
Max no. consecutive wet days
Mean wet-spell length

Mean dry-day persistence
Max no. consecutive dry days
Mean dry-spell length

Linear
Logit.probs
Linear

Linear
Logit.counts
Logit.counts
Linear
Linear

Logit.probs
Logit.counts
Logit.counts

Logit.probs
Logit.counts
Logit.counts

Table I. List of the extreme precipitation indices used in this study. The indices are grouped into four categories (from top to bottom): basic, heavy precipitation, wet spells, and dry spells. The last column indicates the method used to estimate the trend: linear regression (linear), logistic regression using event counts (logit.counts), and logistic regression using event probabilities (logit.probs)

Two different methods were used to calculate trend magnitudes and to test for statistical significance, depending on the value range of the index. For indices with a continuous value range (e.g. quantiles, seasonal precipitation maxima), trends were calculated with conventional linear regression. These indices are indicated by ‘linear’ in Table I. The trend magnitude is then expressed as a percentage change over the 100 years relative to the 100 year mean value of the index. In the ‘linear’ case, statistical significance is assessed following the non-parametric Kendall tau test (Kendall, 1970). This is a robust, rank-based test, which, unlike the conventional Student’s t -test, does not depend on the assumption of Gaussian distributed residuals.

For indices with a discrete value range, e.g. for counts of threshold exceedances and frequencies, we have used logistic regression for trend calculation. Logistic regression is a special case of a generalization of regression techniques (the generalized linear models; see McCullagh and Nelder (1989)). It is appropriate for dealing with number counts and probabilities, for which the assumptions of linear regression with uniform variance and Gaussian residuals are not satisfied. The logistic regression approach of this paper is similar to the application in Frei and Schär (2001). In particular, the logit function were used as a link function and the maximum likelihood method for parameter estimation. Moreover, in the assessment of statistical significance the authors have corrected for overdispersion in the data series, which is an implicit account of the serial correlation in the annual series. Their calculations are based on the software package R (Ihaka and Gentleman, 1996), where slightly different formal approaches are needed for counts compared with probabilities (see also Venables and Ripley (1997)). Indices for which trends are calculated and tested by logistic regression are labelled as logit.counts and logit.probs in Table I. For convenience in displaying the results, the trend magnitudes derived by logistic regression are converted into relative changes over the 100 years as for linear regression. Yet it should be noted that trend magnitudes between the two regression methods are not strictly comparable.


(2) - Effets du changement climatique sur le milieu naturel
Reconstitutions
 
Observations
 
Modélisations
 
Hypothèses
 

Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
 

 


(3) - Effets du changement climatique sur l'aléa
Reconstitutions
 
Observations
 
Modélisations
 
Hypothèses
 

Paramètre de l'aléa
Sensibilité des paramètres de l'aléa à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
 

 


(4) - Remarques générales

During recent decades, the Alpine region, and Switzerland in particular, has experienced several events of extreme precipitation conditions that have led to severe damage and fatalities. To name a few recent examples: during the record warm summer of 2003 only about 50% of the average summer precipitation fell in Switzerland, which caused large losses in agriculture. Heavy precipitation in November 2002 caused mudflows in the eastern Swiss Alps, and a flash mudflow in October 2000 during an event of heavy precipitation destroyed 10 houses and caused 13 fatalities in the mountain village of Gondo. Extended and intense rainfall and sudden snowmelt in May 1999 resulted in flooding along rivers and lakes over several weeks. These recent events have raised concern that the frequency of precipitation extremes has gradually changed over the 20th century, possibly in response to global climate change.

In this study the authors investigate long-term variations and trends of precipitation extremes during the 20th century in Switzerland. For this purpose, daily precipitation data are analysed from a comparatively dense network of more than 100 rain gauges, all of which have been continuously operated since 1901. The trend analysis is based on a wide range of diagnostics, covering basic precipitation statistics, single to multi-day heavy precipitation events, spells of intense precipitation and anomalous dry spells.

There are two main reasons for considering a wide range of diagnostics. Firstly, precipitation extremes in Switzerland can be of quite a variable nature: heavy convective rainfall lasting for a few hours and extending over typically a few hundred square kilometres occurs predominantly in summer (e.g. Schiesser et al., 1995). Trend signals for these events may reflect in the frequency distribution of 1 day precipitation totals. (Subdaily time resolution would be desirable, but is not available over such a long period.) On the other hand, multi-day heavy precipitation episodes occur throughout the year. One of the prominent examples of these types of event is heavy precipitation and flooding along the southern Alpine rim, occurring predominantly in autumn, in connection with moist southerly air flows (e.g. Massacand et al., 1998; Schär et al., 1998; Buzzi and Foschini, 2000). Trend signals for these types of event may be expected in multi-day precipitation totals or in the length and frequency of wet spells.

Secondly, there are limitations in our ability to identify trends in extreme events. The signal-to-noise ratio in a trend analysis depends on the record length, the trend magnitude, the ‘noise’ level (e.g. the magnitude of interannual variations), and the rarity of events under consideration. Unless the trend magnitude is very large, we must expect that a real trend may not be detectable in a statistical test when we focus on diagnostics for very rare extremes only (Frei and Schär, 2001). Therefore, it is advisable to investigate several diagnostics which sample different parts of the frequency distribution. Results for elementary diagnostics, such as the frequency of wet days and their average intensity, may be less indicative of trends in extremes, but may be statistically more robust and, nevertheless, interesting as a sign of long-term change in precipitation processes.

To cover long-term trends from a range of characteristics in precipitation extremes the authors consider a number of diagnostics evaluated from time series of daily precipitation totals. Many of these diagnostics were found to be useful as a reference for comparing trend results between different regions (Nicholls and Murray, 1999). The diagnostics are called indices in the present paper.


(5) - Syntèses et préconisations
 

Références citées :

Alpert P, Ben-Gai T, Baharad A. 2002. The paradoxical increase of Mediterranean extreme daily rainfall in spite of decrease in total values. Geophysical Research Letters 29: 1536. - [Fiche Biblio]

Bárdossy A, Hundecha Y. 2003. Trends of extreme precipitation and temperature associated climatic conditions in the German part of the Rhine basin from 1958 to 2001. Stardex report, University of Stuttgart.

Begert M, Schlegel T, Kirchhofer W. 2005. Homogeneous temperature and precipitation series of Switzerland from 1864 to 2000. International Journal of Climatology 25: 65–80. - [Fiche Biblio]

Birsan MV, Molnar P, Burlando P, Pfaundler M. In press. Streamflow trends in Switzerland. Journal of Hydrology. - [Fiche Biblio]

Brunetti M, Maugeri M, Nanni T. 2001. Changes in total precipitation, rainy days and extreme events in northeastern Italy. International Journal of Climatology 21: 861–871.

Brunetti M, Buffoni L, Mangianti F, Maugeri M, Nanni T. 2004. Temperature, precipitation and extreme events during the last century in Italy. Global and Planetary Change 40: 141–149.

Buzzi A, Foschini L. 2000. Mesoscale meteorological features associated with heavy precipitation in the southern Alpine region. Meteorological and Atmospheric Physics 72: 131–146.

Christensen JH, Christensen OB. 2003. Severe summertime flooding in Europe. Nature 421: 805–806. - [Fiche Biblio]

Courvoisier HW. 1998. Statistik der 24-stündigen Starkniederschläge in der Schweiz. SMI Report 194, MeteoSwiss. (Available from MeteoSwiss, Krähbühlstrasse 58, CH-8044 Zürich, Switzerland).

Durman CF, Gregory JM, Hassell DC, Jones RG, Murphy JM. 2001. A comparison of extreme European daily precipitation simulated by a global and a regional climate model for present and future climates. Quarterly Journal of the Royal Meteorological Society 127: 1005–1015.

Frei C, Schär C. 2001. Detection probability of trends in rare events: theory and application of heavy precipitation in the Alpine region. Journal of Climate 14: 1568–1584. - [Fiche Biblio]

Frich P, Alexander LV, Della-Marta P, Gleason B, Haylock M, Klein Tank AMG, Peterson T. 2002. Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate Research 19: 193–212.

Groisman PYA, Karl TR, Easterling DR, Knight RW, Jamason PF, Hennessy KJ, Suppiah R, Page CM, Wibig J, Fortuniak K, Razuvaev VN, Douglas A, Forland E, Zhai PM. 1999. Changes in the probability of heavy precipitation: important indicators of climatic change. Climatic Change 42: 243–283.

Haylock MR. 2003. Observed changes in European extremes, 1958–2000. Stardex report, University of East Anglia.

Haylock MR, Goodess CM. 2004. Interannual variability of European extreme winter rainfall and links with mean large-scale circulation. International Journal of Climatology 24: 759–776.

Huntingford C, Jones RG, Prudhomme C, Lamb R, Gash HHC, Jones DA. 2003. Regional climate-model predictions of extreme rainfall for a changing climate. Quarterly Journal of the Royal Meteorological Society 129: 1607–1621.

Hurrell JW. 1995. Decadal trends in the Northern Atlantic oscillation — regional temperatures and precipitation. Science 269: 676–679.

Ihaka R, Gentleman R. 1996. R: a language for data analysis and graphics. Journal of Computational and Graphical Statistics 5: 299–314.

Klein Tank AMG, Können GP. 2003. Trends in indices of daily temperature and precipitation extremes in Europe. Journal of Climate 16: 3665–3680.

Knight PYAGRW, Karl TR. 2001. Heavy precipitation and high streamflow in the contiguous United States: trends in the twentieth century. Bulletin of the American Meteorological Society 82: 219–246.

Massacand AC, Wernli H, Davies HC. 1998. Heavy precipitation on the Alpine southside: an upper-level precursor. Geophysical Research Letters 25: 1435–1438.

McCullagh P, Nelder JA. 1989. Generalized Linear Models. Monographs on Statistics and Applied Probability, 37, 2nd edn. Chapman and Hall: London.

Nicholls N, Murray W. 1999. Workshop on indices and indicators for climatic extremes: Asheville, NC, USA, 3–6 June 1997 Breakout Group B: precipitation. Climatic change 42: 23–29.

Rebetez M. 1999. Twentieth century trends in droughts in southern Switzerland. Geophysical Research Letters 26: 755–758.

Räisänen J, Hannson U, Ullerstig A, Döscher R, Graham LP, Jones C, Meier HEM, Samuelsson P, Willén U. 2004. European climate in the late twenty-first century: regional simulations with two global models and two forcing scenarios. Climate Dynamics 22: 13–31.

Schär C, Davies TD, Frei C, Wanner H, Widmann M, Wild M, Davies HC. 1998. Current Alpine climate. In A View from the Alps: Regional Perspectives on Climate Change, Cebon P, Dahinden U, Davies HC, Imboden D, Jàger C (eds). MIT Press: 21–72.

Schär C, Vidale PL, Lüthi D, Frei C, Haberli C, Liniger MA, Appenzeller C. 2004. The role of increasing temperature variability in European summer heatwaves. Nature 427: 332–336. - [Fiche Biblio]

Scherrer SC, Appenzeller C. 2004. Trends in Swiss Alpine snow days: the role of local- and large-scale climate variability. Geophysical Research Letters 31: L13 215.

Schiesser HH, Houze RA, Huntrieser H. 1995. The mesoscale structure of severe precipitation systems in Switzerland. Monthly Weather Review 123: 2070–2097.

Schmidli J, Schmutz C, Frei C, Wanner H, Schär C. 2002. Mesoscale precipitation variability in the region of the European Alps during the 20th century. International Journal of Climatology 22: 1049–1074.- [Fiche Biblio]

Semenov VA, Bengtsson L. 2002. Secular trends in daily precipitation characteristics: greenhouse gas simulation with a coupled AOGCM. Climate Dynamics 19: 123–140.

Venables WN, Ripley BD. 1997. Modern Applied Statistics with S-PLUS. Springer: New York.

Vidale PL, Lüthi D, Frei C, Seneviratne SI, Schär C. 2003. Predictability and uncertainty in a regional climate model. Journal of Geophysical Research–Atmospheres 108: (D18) 4586.

Voss R, May W, Roeckner E. 2002. Enhanced resolution modelling study on anthropogenic climate change: changes in extremes of the hydrological cycle. International Journal of Climatology 22: 755–777.

Widmann ML, Schär C. 1997. A principal component and long-term trend analysis of daily precipitation in Switzerland. International Journal of Climatology 17: 1333–1356.