Réf. Durand & al. 2009b

Référence bibliographique complète

DURAND, Y., GIRAUD, G., LATERNSER, M., ETCHEVERS, P., MÉRINDOL, L., LESAFFRE, B.  2009. Reanalysis of 47 years of climate in the French Alps (1958–2005): climatology and trends for snow cover. Journal of Applied Meteorology and Climatology, Vol. 48, 2487-2512.

Abstract: Since the early 1990s, Météo-France has used an automatic system combining three numerical models to simulate meteorological parameters, snow cover stratigraphy and avalanche risk at various altitudes, aspects and slopes for a number of mountainous regions (massifs) in the French Alps and the Pyrenees. This SAFRAN-Crocus-MEPRA model chain (SCM), usually applied to operational daily avalanche forecasting, is here used for retrospective snow and climate analysis. For this study, the SCM chain used both meteorological observations and guess fields mainly issued from the newly reanalysed atmospheric model data (ERA-40) of the European Centre for Medium-Range Weather Forecasts and ran on a hourly basis over a period starting in the winter of 1958/59 until recent past winters. Snow observations were finally used for validation and the results presented here concern only the main climatic features of the alpine modelled snowfields at different spatial and temporal scales.
The main results obtained confirm the very significant spatial and temporal variability of the modelled snowfields with regard to certain key parameters such as those describing ground coverage or snow depth. Snow patterns in the French Alps are characterised by a marked declining gradient from the northwestern foothills to the southeastern interior regions. This applies mainly to both depths and durations, which exhibit a maximal latitudinal variation at 1500m of about 60 days, decreasing strongly with the altitude.
Enhanced at low elevations, snow depth shows a mainly negative temporal variation over the study period, especially in the north and during late winters, while the south exhibits more smoothed features. The number of days with snow on the ground shows also a significant general signal of decrease at low and mid-elevation but this signal is weaker in the south than in the north and less visible at high elevation.. Even if a significant statistically test cannot be performed for all elevations and areas, the temporal decrease is present in all the studied quantities. Concerning snow duration, this general decrease can also be interpreted as a sharp variation of the mean values at the end of the 1980s, inducing a step effect in its time series rather than a constant negative temporal trend. The results have also been interpreted in term of potentiality for a viable ski industry especially in the southern areas and for different changing climatic conditions. Presently, French downhill ski resorts are economically viable from a range to about 1200 m a.s.l. in the northern foothills to 2000m in the south, but future prospects are uncertain.
In addition, no clear and direct relationship between the NAO or the ENSO indexes and the studied snow parameters could be established in this study.

Mots-clés
Climatology, Snow cover, Reanalysis, SCM model chain, French Alps

Organismes / Contact
GAME/CNRM-CEN (CNRS/Météo-France) - Corresponding author: Yves Durand, Météo-France, CNRM-CEN, 1441 rue de la Piscine, 38400 Saint-Martin d’Hères, France. E-mail: yves.durand@meteo.fr

(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
Temperature, Precipitation [see Durand & al 2009a] Snow cover (depht and duration)    

Pays / Zone
Massif / Secteur
Site(s) d'étude
Exposition
Altitude
Période(s) d'observation
Alpes françaises

23 massifs:

[See Durand & al 2009a]

Chablais
Aravis
Bauges
Chartreuse
Vercors
Mont-Blanc
Beaufortin
Haute-Tarentaise
Haute-Maurienne
Vanoise
Maurienne
Belledonne
Grandes-Rousses
Thabor
Oisans
Pelvoux
Champsaur
Dévoluy
Queyras
Parpaillon
Ubaye
Alpes-Azuréennes
Mercantour

600–2700 m
900–2700 m
600–2100 m
600–2100 m
600–2400 m
1200–3600 m
900–3000 m
900–3600 m
1200–3600 m
900–3600 m
600–3000 m
600–3000 m
900–3300 m
1500–3000 m
900–1200 m
1200–3600 m
1200–3300 m
600–3000 m
1200–3000 m
900–3300 m
1200–3000 m
600–2700 m
1200–3000 m
  1959–2005

(1) - Modifications des paramètres atmosphériques
Reconstitutions
 
Observations
 
Modélisations
 
Hypothèses
 

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

 


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

Snow climate variability at various spatial scales

  • Average snow depth temporal variability at an inter-massif scale

DJF (December, January, February) time-averaged snow depth over the entire area (i.e. averaged over the 23 massifs):
[...] The deviations from the mean value are quite significant (+29cm and –37cm on the average for a mean DJF value of 67cm) and that time variability is the prevailing phenomena with a chaotic succession of low and high values. The winter of 1966 stands out in the series both for its high mean value and its high spatial variability (very least snowy period in the South). Moreover the winter of 1975 was very snowy in the North and 1982 exhibits a snowy period in all the massifs. After the mid-1980s, the snow cover is more homogeneous over the whole area with decreased mean values and a reduced amplitude of extreme values, indicating less spatial variability. However the number of data points is too small to draw firm conclusions. This mid-1980s threshold is also mentioned by Beniston (2005a) along with the subsequent reduced snow amounts. The same author links this behaviour to a high and persistent pressure field over several years and its interpretation in terms of NAO index. In our own case (1800m, numerically analysed values, French Alps only), no clear and immediate relationship with the NAO index (averaged DJF values plotted in green in the figure) is observed (correlation coefficients of about –0.2). Within the framework of the present study, a clear relationship between temperature and NAO index was established in the companion paper [Durand & al 2009a], but not between precipitation and NAO index, which is in accordance with the low values obtained here for snow cover. [...]

Annual distribution of modelled monthly values of snow depths at 1800 m a.s.l. over the entire French Alps, Northern Alps and Southern Alps:
As already seen, the main marked maxima of the first 30 years are all clearly visible and are followed by a transition period extending until the late 1980s and a new set of years with less snowy magnitude and total duration. This behaviour is similar both in Northern and Southern Alps but with a large difference in the magnitude of the events and a significant lack of snowy winters in the South. Note that snowy winters (in terms of snow depth) do not always begin early but end later due to a longer melting period (initial temporal shift of nearly one month mainly in the South, less in the North). More globally, [one can] note the relatively chaotic features of the snow depth fields.

  • Snow depth temporal variability at the massif scale

Spatial and temporal evolution [relative deviation]of the mean winter snow depth at 1800 m a.s.l. from 1959–2005:
[...] In this section, “snowy” winter seasons [are considered] to be those for which the deviation from the mean exceeds at least 50% with a relatively homogenous spatial pattern. The “least snowy” winters correspond to a deviation below –50%. [...] Usually the far south differs most from all the other massifs. Least snowy winters are more homogeneously distributed than most snowy winters. In general, low elevations have stronger relative deviations from the long-term mean than high elevations.

At mid elevations, 1966 ranks first for most massifs except those in the far south (30th rank). For the entire French Alps, snowy winters were 1960, 1963, 1978 and 1982 while 1970 and 1975 were mainly snowy in the north and 1986 or 1996 in the south. Other detailed results indicate that 1986 was by far the snowiest winter at low elevations in the far south (+490% deviation). At high elevations, 2001 was the overall snowiest winter, particularly in central and southern areas, which does not appear in the figure, while 1960, 1977 and 1997 were also snowy. On the other hand, northern massifs had snowy high-elevation winters in 1966, 1975, 1993 and 1995. [see also detailed results...]

  • 100-day snow depth temporal variability at the massif scale

The minimum 100-day snow depth at 1800 m a.s.l. shows a very large year-to-year variability, but basically decreases over the studied period for all massifs. However, many northern and central massifs show an initial increase until the early 1980s followed by a decline. Some central-southern massifs show a relatively uniform but weak decline right from the beginning and Pelvoux, Dévoluy, Champsaur, Parpaillon and Ubaye even show a constant trend at mid altitudes. [...]

 

Long term Trend Analysis

  • Trends of average snow depth at the inter-massif scale

Annual averaged DJF snow depths at different elevations for the entire French Alps, Northern Alps and Southern Alps:
Within the high variability, no temporal trend is clearly visible at high elevation (2700 m a.s.l.) for any of the areas. At this altitude, the snow balance is mainly forced by precipitation and no trend was detected in the companion paper for this parameter. A general decline is visible at lower elevations, especially in the northern area, while the south exhibits more smoothed features with many snowy winters during the last decade. Extreme events such as the very low snowy 1964, 1988 or 1989 winters or the snowy 1966 or 1982 winters are also clearly visible. The transition zone at the end of the 1980s corresponds here to a widespread decrease for all elevations and areas as well as a weaker altitudinal variability. Indeed, at all medium and high elevations and especially in the Northern Alps, the beginning of the 1980s until about 1987 is a consistent period of snowy winters which, according to Beniston (1997), can be explained by a persistence of high pressures over Atlantic during these winter seasons. This period is followed by two consecutive very low winters, 1988 and 1989, similar in appearance to the recent 2002 and 2005 seasons.

In addition, a tendency towards low snow winters at low elevations and “more constant” snow winters at high elevations, particularly in the south, can be observed and could be related to the temperature and precipitation trends already mentioned. In autumn, temperatures become cooler while precipitation rises slightly. In spring, temperatures rise sharply while precipitation decreases. Temperature trends are most pronounced at around 2000 m a.s.l., making this altitude most critical for snow pack evolution. Whereas the northern massifs at this altitude are still cold enough to produce and maintain a regular winter snow cover, the southern massifs increasingly lack snow. Therefore, highly variable winters with large year-to-year fluctuations are a typical feature in the south, which may increase in the future both in the south and in the north under ongoing rising temperatures.

  • Trends of snow cover duration at the inter-massif scale

Annual averaged duration of snow cover (10cm threshold) at different elevations [...] over entire French Alps, Northern Alps and Southern Alps:
Generally speaking, the time variability is high and the trends are all graphically negative. The negative trend is enhanced at low elevations and persists also in early winter. The same negative trend is also visible at highest level (2700 m), especially in the southern area. The impact of the temperature increase at this elevation (as presented in the companion paper) may not be overridden by supplementary snowfalls. The extreme values correspond to those previously seen, such as the seasons 1984, 1990, 1993 or the recent 2007 for minimal values and 1966 or 1975 for maximal values. For this part of the study, two recent winter seasons, 2006 and 2007, were added to the data, especially because they exhibit particular features (2006: less snowy in terms of snow depth but with very cold weather causing a relatively long snow duration at low elevation as well as many accidents ; 2007: very mild with little precipitation).

[...] Statistical verifications [were done] of this generalised hypothetical decrease over the entire French Alps and its four main areas for elevations between 1500 and 2700 m a.s.l. [...] Spearman test of the temporal decrease, which does not imply a linear trend, gives a contrasted result when compared to the graphical representation. Only the 2100m level exhibits systematically a significant signal of snow duration decrease in all areas; this signal is also weaker in the south than in the north and absent at high elevation. [...] Globally, these tests do not formally prove a systematic significant downward trend as previously discussed. Beniston (1997) also observes no significant linear decrease trend over a longer period (starting in 1945) for either snow duration or maximum snow depth at different Swiss observation points, with even a slight increase after 1985. This last point could be consistent with the lowest elevation curves of the [present study]. Another link can be done with the companion paper, where the positive linear temperature trend is maximal only at mid elevation and where also no precipitation trend is detected.

Another possible reason for this lack of statistical significance of a long-term negative trend and of a generalised diagnostic of a negative decrease could be a lack of temporal consistency in the observed series, for instance due to a break (or step) in relation to an intervening year. [...] By looking at their behaviour before and after the middle of the 1980s, [...] most of the curves, when centred, exhibit a relatively flat pattern without any clear trend. [...] Presently, [the authors] cannot confirm with certainty this step hypothesis nor its possible causes with only [their] snow data; however [they] can quote the results of the companion paper which exhibits similar breaking-off behaviours at the same period in its smoothed temperature series, as is also the case for many other authors.

  • Trends at the massif scale

It is difficult to present accurate results at this scale, mainly because of the smoothing effects of the SAFRAN objective analyse software used as input. However, the main behaviours may be seen and corroborate our past spatial clustering attempt. At that elevation (1800m), no massif shows clear individual feature and, visually, one can appreciate the differences before and after the 1980s concerning snow depth or minimum 100-day snow depth. These differences are more marked at low elevation and can even be spatially inconsistent. Some central and southern massifs (Grandes Rousses, Pelvoux, Champsaur, Queyras, Ubaye, Vercors and Ht-Var/Ht-Verdon) feature also a strong decreasing trend from the beginning (not shown here), persisting in a weaker form up to high altitudes. No massif shows a rising trend in recent years except Parpaillon, which has a very strong decline at low elevations but starts to rise above 1500 m a.s.l.

Globally, the number of days with snow on the ground generally shows little evolution at medium and high altitudes, but is quite variable at low elevations (elevations not shown here). Below about 1000 – 1500 m a.s.l., most of the massifs experience a rise until the 1970s followed by a decline. Usually the decline is stronger than the rise except for some northern massifs (Chablais, Bauges, Mt Blanc, Belledonne). After an initial stagnant period, some central and southern massifs show a slight rise in recent years: Pelvoux, Champsaur, Thabor at mid altitudes only and Ubaye also at high altitudes. Parpaillon shows a strong decline at low altitudes but a rise at medium and high elevations. The far south shows the typical rising-falling trend with a rather strong amplitude and an ongoing decline up to high altitudes for Mercantour.

Observations
 
Modélisations
 
Hypothèses

[...] Analysing [the] sensitivity of snow cover to future climate variations, Martin et al. (1994 and 1997) found that temperature has the strongest impact on snow cover duration at low and medium altitudes, but precipitation becomes the key factor at high altitudes. Above a limit of about 1700 – 2000 m a.s.l., winter temperatures are mostly cold enough to produce snow (instead of rain) and the blocking of high pressure systems affects the amount of precipitation. This is consistent with the findings of Scherrer et al. (2004), who showed for the Swiss Alps that the recent decrease in low altitude snow cover can mainly be attributed to an increase in temperature (whereas precipitation trends are negligible). However, [the authors] have not been able to demonstrate a significant link with the NAO index over [their] area as did Beniston (1997) for snow cover and duration in the Swiss Alps, however [they] have observed that high NAO values globally correspond to weak snow depths. During the late 20th century, the dual effects of the NAO trends and the natural variability at smaller scales can be considered to have superimposed non-linearly. Given that global climate change could very well result in more frequent positive NAO index winters in the 21st century (Hurrell et al., 2003), low altitude snow cover, particularly in the south, may decrease even more than expected from a temperature increase alone. [...]


Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
Relationship with large scale forcing (NAO and ENSO)

The averaged snow depths, or derived quantities, over the different areas and at different elevations have been compared to the North Atlantic Oscillation index “NAO” variations over the whole study period in order to try to better explain observed features. As pointed out by Beniston (2005a), NAO is well representative of the decadal scale variability especially at high elevation and during the winter season when westerly meteorological flows are more intense. [...]

Correlation between the yearly NAO index and the modelled annual snow duration values over several regional areas at 1800 m. a.s.l. and averaged over different numbers of years:
At all averaging lengths, the correlations are negative and quite weak but significant for lengths up to 4 years, after which the variability of the involved variables is excessively reduced. This negative correlation between snow duration and NAO has already been pointed out by Beniston (2005b) on different observed Swiss series, as the result of a positive relationship between surface pressure and NAO combined with a negative relationship between pressure and snow duration (Beniston, 1997). The magnitude of the correlation is also close to the value computed with precipitation and presented in the companion paper.

These values can also be compared to those obtained by Scherrer et al. (2006) who found a significant negative correlation (~-0.3) between NAO and new snow sum measurements over Switzerland for the period 1931-1999. In addition, these authors [...] concluded that only observed low-lying snow amounts were primarily linked to NAO oscillations and probably to an ongoing climate change. Other features have already been identified, using observed series, by several authors such as Beniston (1997) who emphasises the role of large scale forcing on the snow cover characteristics.

However, no clear and direct relationship between NAO and snow related parameters seems to exist in our context. Even if the relation between NAO and high pressures and their induced circulations is obvious, they are not the only factors that condition snow cover. In the companion paper, precipitation is shown to be poorly correlated with NAO. Threshold effects due to the bounded quantity used (non-negative snow depth values), the high non-linearity of the snow equilibrium budget and the discrepancy between the different horizontal characteristic scales involved are perhaps at the root of this non-systematic relationship with NAO. However, [one can] note that high NAO values often match well with low snow depth and vice versa, as suggested by the previously mentioned negative correlation values.A complementary but limited similar evaluation of the ENSO (El Nino/Southern Oscillation) influence over snow parameters (daily snow depth and number of days with snow on the ground) has also been performed. [...] In term of correlation, results with ENSO have the same magnitude of non-significant values without any obvious signal for our both studied parameters.

This paper is the second part of an extensive climatological based study carried out for the French Alps [23 massifs, see Durand & al 2009a]. For the first time, long-term climate series (temperature, precipitation and various snow cover parameters) are presented for this area. Based on 40 years of newly reanalysed atmospheric model data (ERA-40) from the European Centre for Medium-Range Weather Forecasts and supplemented by 7 years of our own French analyses, the SAFRAN-Crocus-MEPRA model chain (SCM) has been used to retrospectively model meteorological and snow pack parameters over a period starting in the winter of 1958/59 until recent years (2005 for climatology and 2007 for temporal analyses).

Methods:
The usual assumption [is made] that it is possible, for a given set of elevations, aspects and slopes, to simulate the evolution of the main characteristics of the snow pack in a given region from the average meteorological conditions prevailing in this region. The method has been used operationally since 1992, providing good results (Brun et al., 1989 and 1992; Durand et al., 1993 and 1999).

The first requirement is to obtain the prevailing meteorological conditions for the considered regions at their different elevations, slopes and aspects. As no observed data is systematically available for each elevation and aspect, we use a meteorological analysis model, called SAFRAN, to compute the relevant meteorological variables from the overall meteorological information available inside and around the considered region. The output of SAFRAN is then used by the Crocus snow model to calculate the corresponding evolution of the snow pack. No snow-related observations are used in the full numerical scheme and the modelled snow cover is thus forced only by the meteorological analyses. This point is very important for the different result validations proposed. [...] [see details of the models used, p. 5 & 6...]

Studied parameters:
The main Crocus output parameter used in this study is daily snow depth (HS). This quantity is estimated inside the model by addition of all the elementary layers which are all individually modelled. So, even if a value of HS is not directly representative of all the global characteristics of the snow pack (e.g. water equivalent), it does take into account the entire stratigraphy. This HS parameter is used to determine several secondary parameters. Concerning the duration of snow cover, only the number of days with snow on the ground (n0) is considered. Other parameters as the maximum time period during which the ground is continuously covered with snow have also been studied but are more difficult to synthesise due to partial coverage or to a temporary vanishing of the snow cover especially at low elevations. From a climatological point of view, n0 is probably the more significant parameter (because it is related to a combination of low temperature and solid precipitation) and it will be associated to the beginning (b0) and end (e0) of continuous snow cover. Another parameter used is the minimum 100-day snow depth (hs100d). [...] [see also "Validation of studied parameters" p. 8 & 9...]

[...] The NAO index used here is the value computed daily by the NOAA/NWS/CPS and is constructed by projecting the daily (00Z) 500mb height anomalies over the Northern Hemisphere onto the first leading modes of the Rotated Empirical Orthogonal Function (REOF) analysis of monthly mean 500mb heights over the 1950-2000 period.


(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
 

(5) - Syntèses et préconisations

Summary and discussion:

[...] Snow depth trends are mainly negative for the entire winter half-year with maximum values at lower elevations in particularly in the northern areas but the signal is less visible in the southern zones at all elevations. Globally, a main decrease of the snow depth mean values can be observed at the end of the 1980s after rather snowy periods during the 1960s, late 1970s and early 1980s (less in the central-south), more clearly at low than at high altitudes. However, early winter snow depth has increased at medium and particularly at high altitudes, more significantly in the south than in the north, but decreased everywhere at low elevations. Central-southern areas still faintly show increasing tendencies during mid winter (leading to a slight overall increase for the entire winter half-year), whereas all other regions begin dropping by then. Much enhanced, this decline continues throughout late winter in all massifs and at all elevations.

The number of days with snow on the ground shows little change at medium and high elevations (except some central and southern massifs indicating a slight rise), but has generally declined at low altitudes. [The authors'] hypothesis of a possible step-like decrease of the modeled temporal series over the French Alps, mainly at low elevation, during the 1980s with no clear trend since then has also been recently observed and discussed by Marty (2008) with several long term Swiss observation series. The maximum magnitude of this decrease is a little less than one month at mid elevation and can be put in relationship with the maximum positive temperature trend (0.034 °C yr-1) presented in the companion paper at these elevations. The sensitivity of snow cover to temperature has also been investigated with observed and modeled series by Hantel et al. (2007) who found similar results but over a larger Alpine area and at a lower elevation. This difference with [the present] data in term of elevation comes mainly from a weaker positive trend (0.020 °C yr-1) at very low elevation in our area and discussed in the companion paper.

The tendency towards low snow winters at low elevations and “more resistant” snow winters at high elevations, particularly in the south, can be explained by temperature and precipitation trends. The underlying meteorological forcing has already been presented and discussed by Durand et al., (2009). In late summer and early winter, temperatures drop, particularly in the central-south. Together with constant to rising precipitation, this results in earlier and stronger snowfalls, particularly at high elevations. On the other hand, strongly rising temperatures in late winter and early summer, again predominantly in the central-south, together with decreasing precipitation, cause less snowfall and thus a reduced snow depth and earlier melt, particularly at low and medium elevations. The fact that temperature trends are most pronounced at around 2000 m a.s.l. makes this altitude the most critical for snow pack evolution. Whereas the northern massifs at this altitude are still cold enough to produce and maintain a regular snow cover, the southern massifs are increasingly deficient. Therefore, highly variable winters with major year-to-year fluctuations are a typical feature in the south, which may increase in the future and gain importance in the north as a consequence of ongoing rising temperatures.

Comparison with snow climatology studies carried out in Switzerland (Laternser and Schneebeli, 2003) and Austria (Fliri, 1992) shows close parallels to the general evolution in the French Alps and demonstrates that outstanding winters with much (1963, 1966, 1970, 1982) or little snow (1964, 1989, 1990, 1993) seem to be a phenomenon affecting the entire Alps. 1966 is of particular interest : the overall snowiest winter in the French Alps (except in the far south) extended into southwestern Switzerland (Valais), reduced its magnitude through central Switzerland before gaining intensity again to peak in western Austria (Tyrol). Many other smaller-scale trans-border ties between France and Switzerland exist, with neighbouring areas in the northeastern French Alps showing high correlations with southwestern Switzerland (e.g. 1971, 1980). Moreover, patterns observed in southern Switzerland (Ticino) presumably continue through northwestern Italy and appear again in the eastern/southeastern French Alps (1960, 1981). Such cases can be observed in 1961 with little snow in the north and much snow in the south or in 1999 with much snow in the north and little in the south. Further parallels between the Swiss and French Alps are that changes (overall negative trends over the whole period) are enhanced at low elevations and that a shorter snow duration is mainly caused by earlier melting in spring than by later first snowfalls in autumn. Laternser and Schneebeli (2003) also confirmed a similar general pattern of temporal snow variations occurred throughout the temperate and sub-polar northern hemisphere (e.g. Brown, 2000). [...]

Références citées :

Beniston, M., 1997, Variations of Snow Depth and Duration in the Swiss Alps over the Last 50 Years: Links to Changes in Large-Scale Forcings, Clim. Change 36, 281–300.

Beniston, M., 2005a. Mountain climate and climatic change: An overview of processes focusing on the European Alps. Pure appl. Geophys. 162 (2005), 1587-1606. [Fiche biblio]

Beniston, M., 2005b. Warm winter spells in the Swiss Alps: Strong heat waves in a cold season? A study focusing on climate observations at the Saentis high mountain site. Geophysical Research Letters, 32, L01812. [Fiche biblio]

Brown, R. D., 2000. Northern Hemispheric snow cover variability and change 1915-97. Journal of Climate 13: 2339-2355.

Brun, E., E. Martin, V. Simon, C. Gendre and C. Coléou., 1989. An energy and mass model of snow cover suitable for operational avalanche forecasting. J. Glaciol., 35 (121), 333-342.

Brun, E., P. David, M. Sudul and G. Brugnot, 1992. A numerical model to simulate snow cover stratigraphy for operational avalanche forecasting. J. Glaciol., 38 (128), 13-22.

Durand, Y., E. Brun, L. Mérindol, G. Guyomarc'h, B. Lesaffre and E. Martin, 1993. A meteorological estimation of relevant parameters for snow models, A. Glaciol., 18, 65-71.

Durand Y., Giraud G., Brun E., Mérindol L., Martin E., 1999. A computer based system simulating snowpack structures as a tool for regional Avalanche forecast. J. Glaciol., 45 (151), 469-485.

Durand, Y., M. Laternser, G. Giraud, P. Etchevers, B. Lesaffre, and L. Mérindol, 2009: Reanalysis of 44 Yr of Climate in the French Alps (1958–2002): Methodology, Model Validation, Climatology, and Trends for Air Temperature and Precipitation. , JAMC , 48, 3, March 2009, 429–449. [Fiche biblio]

Fliri, F., 1992. Der Schnee in Nord- und Osttirol 1895-1991: Ein Graphik-Atlas [Snow in North and East Tyrol 1895-1991: a graphical atlas], vol. 1 and 2. Wagner: Innsbruck (in German).

Hurrell, J. W., Kushnir, Y., Ottersen, G., Visbeck, M., 2003. The North Atlantic Oscillation: climatic significance and environmental impact. Geophys. Monogr. Ser. 134, AGU, Washington D.C.

Laternser, M., Schneebeli, M., 2003. Long-term snow climate trends of the Swiss Alps (1931-99). Int. J. Climatol.: 23: 733-750.

Martin, E., Brun, E., Durand, Y., 1994. Sensitivity of the French Alps snow cover to the variation of climatic variables. Annales Geophysicae 12: 469-477.

Martin, E., Timbal, B., Brun, E., 1997. Downscaling of general circulation models outputs: simulation of the snow climatology of the French Alps. Sensitivity to climate changes. Climate Dynamics 13: 45-56.

Marty, C. , 2008. Regime shift of snow days in Switzerland. Geophysical Research Letters, 35, L12501, doi:10.1029/2008GL033998.

Scherrer, S. C., Appenzeller, C., Laternser, M., 2004. Trends in Swiss Alpine snow days: The role of local- and large-scale climate variability. Geophysical Research Letter 31: L13215.

Scherrer, S. C., Appenzeller, 2006. Swiss Alpine snow pack variability: major patterns and links to local climate and large scale flow. Climate Research, 32: 187-199.