International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013

Wet snow avalanche activity in the Swiss Alps –
trend analysis for mid-winter season
Christine Pielmeier, Frank Techel, Christoph Marty, Thomas Stucki
WSL Institute for Snow and Avalanche Research SLF, 7260 Davos Dorf, Switzerland
ABSTRACT: During the winter 2011-2012 the Alps experienced repeated periods with high activity of
wet snow and full-depth glide avalanches during mid-winter season (December to February). Damage
to infrastructure but also fatalities were the consequence. Looking back 5 years, there was at least
one intense and widespread wet snow and full-depth glide avalanche period in four out of five midwinter seasons. To study the long-term trend, changes in wet snow and full-depth glide avalanche activity during mid-winter season were analyzed from 1952 to 2013, based on long-term observation stations. Robust time series analyses showed a positive trend in number and proportion of wet snow and
full-depth glide avalanche records. The trend coincides with the trend in increasing air temperature in
Switzerland. A break in the data series is shown that originates probably from a major revision of the
snow and avalanche recording system in 2002. For the 50 year period before the revision, the proportion of wet snow avalanches increased by 0.4% per year. With the anticipated increasing temperatures
in the European Alps, the positive trend in wet snow avalanche activity will most likely continue, which
requires adaptations in risk management.
KEYWORDS: Wet snow avalanche, full-depth glide avalanche, snowfall limit, avalanche forecasting,
climate change
1

INTRODUCTION

lanches has increased significantly during
the mid-winter season.
The aim of this study is to quantify correlations and trends in avalanche activity using robust time series analysis. It is based on the
snow and avalanche records of long-term SLF
observation stations. These records distinguish
between dry and wet or partially wet snow avalanches from the winters 1951-1952 to 20122013. The trends are discussed in the realm of
snow climate change in Switzerland. Adaptations in local and regional avalanche forecasting
are shown and future challenges are discussed.

An extensive study in the mid 1990s did not
provide evidence for a trend in extreme snowfall
and severe avalanching in the Swiss Alps during
the period 1500-1990 (Laternser and Pfister,
1997; Schneebeli et al. 1997; Laternser, 2002).
More recently, Marty and Blanchet (2012) studied temporal trends of annual maximum snow
depth and 3-day snowfall sum during the last 80
winters, where low elevation stations in particular showed a significant decrease in extreme
snow depth. This was attributed to the observed
decrease in the snow/rain ratio (Serquet et al.,
2011) due to increasing air temperatures. But
does the increase in air temperature impact the
wet snow avalanche activity during mid-winter
(December till February)?
Subjectively, wet snow and full-depth glide
avalanches became more noticeable during
mid-winter in recent years in Switzerland. For
example, the snowy winter 2012 (i.e. the season
2011-2012) lead to high avalanche activity with
damage to infrastructure and forest, mostly due
to full-depth glide avalanches during December
and January (Techel et al., 2013). And last winter on 23 December 2012, it rained to above
2000 m which lead to the highest activity during
the whole winter with 90% wet and partially wet
avalanches (Darms, 2013). From these and other observations it is our hypothesis that the proportion of wet snow and full-depth glide ava--------------------------------

2

DATA AND METHODS

2.1 Avalanche records 1952-2013
Avalanche activity in the Swiss Alps is recorded today by more than 150 observers of the
Swiss avalanche warning service. Since the winter 1952, the observed avalanche data is stored
in the SLF snow and avalanche database.
Laternser (2002) rated the long-term observation stations according to avalanche observation
quality. In our study, we use 24 long-term stations (Table 1) which are well spread over the
seven regions of the Swiss Alps, are all located
at medium elevation (1195 to 1800 m a.s.l.),
have the most complete datasets, are still reporting today and have medium to high avalanche observation quality ratings according to
Laternser (2002).

Corresponding author address: Christine Pielmeier,
SLF, Flüelastrasse 11, CH-7260 Davos Dorf,
tel. +41 81 4170125, email: pielmeier@slf.ch

1240

 International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013
Table 1: Details of the 24 long-term SLF snow
and avalanche observation stations. In brackets
is the number of years, the data are missing for
the whole year.
Code Station
Elevation Observation
name
(m a.s.l.) since
1MR Mürren
1660
1951
1MN Moleson
1520
1965
1AD
Adelboden
1350
1954
1WE Wengen
1310
1969
1GS
Gsteig
1195
1954
2AN
Andermatt
1440
1952
2ME
Meien
1320
1954
2ST
Stoos
1280
1952
3BR
Braunwald
1340
1954
3SW Schwägalp
1290
1967(11)
4BP
Bourg-St1610
1952
Pierre
4GR
Grimentz
1570
1954
4UL
Ulrichen
1350
1953(1)
5AR
Arosa
1820
1954(3)
5BI
Bivio
1770
1953(2)
5SA
St. Antönien 1510
1952
5SP
Splügen
1460
1952
5SE
Sedrun
1420
1952(1)
6SB
San Ber1640
1952
nardino
6BG
Bosco Gurin 1490
1952
7MA
Maloja
1800
1952
7SN
Samnaun
1750
1952
7LD
La Drossa
1710
1968
7ZU
Zuoz
1710
1952

tem, we manually regrouped the avalanche records from 2002-2013 into the previous coding
scheme. Nevertheless, the change of the observation and coding system does coincide with a
marked decrease (Figure 1) in number of avalanche records from the 24 long-term observer
stations, which will be discussed in Chapter 4.
The type and number of avalanches was
taken from the SLF database according to the
codes L2 (type of avalanche) and L5 (number,
size and impact of avalanche) for 1952-2001.
According to the observation guidelines (SLF,
1987) the avalanche codes L2 and L5 were determined by the observer by going through the
coding tables from top to bottom and choosing
the first applicable code according to Table 2.
Since L5 contains data of mixed orderedcategorical scale we cannot count each individual avalanche. However, it is possible to group
and count avalanche records according to snow
wetness and magnitude of activity, where:
 Dry snow avalanches of small magnitude
correspond to L2 = 2, 4, 7, 9 and
L5 = 1, 2, 7, 8, 9;
 Dry snow avalanches of large magnitude
correspond to L2 = 2, 4, 7, 9 and
L5 = 3, 4, 5, 6;
 Wet snow/mixed avalanches of small
magnitude correspond to L2 = 1, 3, 5, 6, 8
and L5 = 1, 2, 7, 8, 9;
 Wet snow/mixed avalanches of large
magnitude correspond to L2 = 1, 3, 5, 6, 8
and L5 = 3, 4, 5, 6.

For the winters 1952-2001, avalanche data
are available according to the coding system
shown in Table 2. In the early years, the observers transmitted their coded snow and avalanche observations by telegram, telephone,
telex or telefax to the SLF where it was eventually entered into the database. In the late 1980’s
digital transmission took over. In 1987, the coding system for avalanche observations was
slightly altered. The old codes were converted
into L2 and L5 codes (Table 2) as far as possible.
A yet larger change of the coding system
happened after the avalanche winter 1999 after
which the Intercantonal Early Warning and Crisis Information System - IFKIS (Bründl et al.,
2004) was launched in Switzerland. This implied
that the observation guidelines as well as the
codes for snow and avalanche observations
were completely revised at the start of the winter
2002. The platform became internet based and
the templates became much more detailed. Because the detail of avalanche records in the new
coding system is greater than it is in the old sys-

By definition, avalanches with unknown type
are considered as dry snow avalanche records.
Mixed avalanches (dry and wet) and full-depth
glide avalanches are considered as wet snow
avalanche records. For 2002-2013, all avalanche records in the database were manually
regrouped accordingly.
Avalanches causing damage are stored
separately from the above mentioned snow and
avalanche data-base. In this data-base 4340
avalanches from all regions of Switzerland,
which caused damage to buildings, infrastructure or forest during mid-winter season (19522013) are recorded. However, 75% of these recordings contain no information on wetness of
the avalanche. Therefore, these data cannot be
used to supplement the long-term analysis of
wet and dry snow avalanche records.

1241

 International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013

Tab. 2: SLF code system for avalanche observations from 1952-2001 (SLF, 1987).
Code
Type of Avalanche
Code
Number, size and impact of avalanche
L2
L5
/
observation impossible
/
observation impossible
0
no avalanches
0
no avalanches
9
type of avalanche unknown
9
extent unknown
8
ground or full-depth glide ava8
avalanche with fatalities
lanche
7
slab avalanche dry
7
avalanche with caught or buried persons
6
slab avalanche wet
6
avalanche with property damage (buildings,
forest, road, railway)
5
slab avalanche dry and wet
5
several (more than two) big avalanches,
without damage
4
slab avalanche and loose snow
4
few (one or two) big avalanches, without
avalanche dry
damage
3
slab avalanche and loose snow
3
several (more than two) medium avalanches,
avalanche wet
without damage
2
loose snow avalanche dry
2
few (one or two) medium avalanches, without
damage
1
loose snow avalanche wet
1
few or several small avalanches, without
damage

Figure 1: Number of avalanche records for dry snow and wet snow avalanches for the 62 mid-winter
seasons from 1952-2013. The grey vertical line marks the winter 2002 where the new observation and
database coding system was launched. This coincides with a systematic ‘drop’ in the amount of data.
The dashed, blue and red horizontal lines are the median of 10 mid-winters of the avalanche records
until the winter 2001 and afterwards.
the monthly anomalies (dTA) for December,
January and February for 1952-2013 and the
1961-1990 mean (MeteoSwiss, 2013). For the
24 long-term SLF snow and avalanche observation stations (Table 1) the means of the following
snow parameters were calculated for the midwinter seasons: median snow depth (HSmedian)

2.2 Temperature and snow data 1952-2013
The monthly anomalies of the air temperature for the mid-winter seasons in Switzerland
are based on homogeneous data series of 13
long-term MeteoSwiss stations. The mid-winter
anomaly is the difference between the mean of

1242

 International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013
and maximum snow depth (HSmax). Figure 2
shows the means of dTA and HSmax for the 62
mid-winter seasons. Large inter-annual variations exist as well as trends to warmer temperatures and lower snow depth starting in the late
1980s.

coding system in 2002 than the absolute numbers of avalanche records. An increase in the
proportion of wet avalanches is obvious in the
10-year moving average (Figure 3) and, since
the late 1980’s it coincides with the increasing
temperature anomaly (Figure 2).

2.3 Statistical analysis
The correlations are analyzed with the nonparametric Spearman rank order correlation (ρ)
testing for a monotonic relationship (Crawley,
2007). Robust time series analysis is performed
by a non-parametric trend test using the statistical software R (R, 2011). The MannKendall
trend test (McLeod, 2011) is applied for significance testing and the Theil Sen statistic as
slope estimator (Bronaugh and Werner, 2009).
The statistical analysis was calculated for the 62
year (1952-2013) and 50 year (1952-2001) series of dry and wet avalanche records in midwinter, for temperature anomaly and snow parameters. Trends are considered significant if
the level of significance α ≤ 0.05.

The absolute number of wet and dry snow
avalanche records as well as their relative frequencies, the temperature anomalies and the
snow parameters described above were analyzed with the Mann-Kendall trend test (MK) and
the Theil-Sen slope estimator (TS). Table 3
shows the p-values and slope statistics for the
significant parameters. In view of the change in
the avalanche coding system in the winter 2002,
trends were calculated for the whole period of
62 years as well as for the 50 years before the
introduction of the new coding system.

3

The Mann-Kendall trend test showed a highly
significant positive trend in the number of wet
snow avalanche records, the proportion of wet
avalanche records and the temperature anomalies over the 62- and 50-year period. In the 50year period, wet snow avalanche records increased by 2.2 records per year, the proportion
of wet snow avalanche records by 0.4% per
year. The trend in the number of dry snow avalanche records was significant only in the 50year period. The temperature anomaly increased by 0.04 °C per year.
None of the snow parameters showed a significant trend. However, when looking at descriptive statistics for discrete time periods, the
median of HSmax was lower for the years 20022013 (106 cm) than for the 50 years before (117
cm, p=0.06), yet not significantly lower than the
10 years directly before 2002. The dTA also
showed no significant differences in these periods.

RESULTS

3.1 Correlation of data series
For the 62-year period, the number of dry
avalanche records is correlated with snow depth
measurements, most with the mean of HSmax
(ρ=0.72, p<10-7). Wet avalanche activity correlates to both, snow depth measurements and
temperature anomaly. However, the correlation
is weak (HSmax: ρ=0.35, p<0.01, dTA: ρ=0.45,
p<10-3). The proportion of wet to dry avalanches
has a moderate correlation to dTA (ρ=0.61,
p<10-6), but no significant correlation to snow
depth variables.
3.2 Trends for 62 mid-winter periods (19522013) and 50 mid-winter periods (1952-2001)
The absolute numbers of dry and wet snow
avalanche records are shown in Figure 1, their
relative frequencies in Figure 3. The proportion
of dry and wet snow avalanche records is much
less influenced by the introduction of the new

Tab. 3: Parameters with significant trends from the Mann-Kendall trend test (MK) and the Theil-Sen
slope estimator (TS) for the 62 year (62a) and the 50 year (50a) series of mid-winter seasons.
Parameter
MK: 62a TS: 62a trend MK: 50a TS: 50a trend
Wet snow avalanche records
p=0.02
+ 0.8 / a
p<10-4
+ 2.2 / a
Proportion wet snow avalanche records p=0.01
+ 0.2% / a
p<0.01
+ 0.4% / a
Temperature anomalies
p=0.04
+ 0.02°C / a
p<0.01
+ 0.04°C / a
Dry snow avalanche records
p>0.05
- 0.1°C / a
p=0.04
+ 1.5 / a

1243

 International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013

Fig. 2: Means of snow depth maxima and temperature anomaly for the mid-winter seasons of the
years 1952-2013. The black curve shows the simple moving average over a 10-year period. An increase in the temperature anomaly and a decrease in maximum snow depth are obvious in the late
1980s.

Fig. 3: Proportion of the number of dry snow (blue area) and wet snow (red area) avalanche records
for the mid-winter season of 1952-2013. The black curve shows the simple moving average over a 10year period. An increase in the proportion of wet snow avalanche records (red area) is obvious in the
late 1980s.

1244

 International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013

4

(Baggi and Schweizer, 2009, Techel et al.,
2011). Intensified snowpack and avalanche observations also by remote sensing (Durand,
2012; Lato et al., 2012) as well as process understanding through improved meteorological
and snow models can assist local and regional
avalanche experts in the challenges present and
ahead. Furthermore special products such as
the Swiss danger ratings for full-depth glide avalanches, special media communications and
new guidelines can assist local safety personnel.

DISCUSSION AND CONCLUSION

We have analyzed a long-term dataset on
avalanche observations for detecting trends in
dry and wet snow avalanche activity during the
mid-winter season (December-February) over
62 years. 24 observer stations were selected
according to length and quality of avalanche
observations. We found a significant temporal
trend in the proportion and absolute number of
wet snow avalanche records in the mid-winter
season. The trend was more pronounced for the
50-year period from 1952 till 2001 than for the
whole 62-year period till 2013. During the 50year period, the median proportion of wet snow
avalanche records increased by 12%, from 35%
in the first 10 years to 47% in the last 10 years.
Also, the mean temperature anomaly increased
significantly, by 0.04 °C per year, which coincides with the proportion and number of wet
snow avalanches in the mid-winter seasons, in
particular since the late 1980’s (Figure 2 and 3).
The median and maximum snow depth did not
show any significant trends.
The observation and coding system experienced a fundamental revision in 2002. This coincided with a decrease in the number of large
avalanches reported by the 24 long-term stations starting in the winter 2002. In addition, also
the numbers of avalanche records decreased in
2002. But the proportion of wet to dry snow avalanche records was not as influenced. This
‘drop’ can possibly be explained with the introduction of an additional group of observers in
the same regions who shared the task of avalanche observations with the long-term observers and are not considered in this study. Changes in observation guidelines and recording
methods will remain a major challenge when
interpreting long-term avalanche data.
Gobiet et al. (2013) and Bavay et al. (2007)
predict snow depletion for the European Alps by
a decreasing snow/rain ratio due to increasing
temperature. Presumably, the trend to increasing wet snow avalanche activity will continue
and avalanche forecasters and safety authorities
will be more often confronted with periods of
high wet snow and full-depth glide avalanche
activity during December, January and February. Predicting avalanche release remains difficult, particularly for full-depth glide avalanches.
The prediction of wet snow avalanche activity
due to rain on snow at high elevation or due to
solar radiation strongly depends on precise
short-term meteorological predictions, e.g.
snowfall line and amount of precipitation (Teich
et al., 2012), on monitoring of the energy balance (Mitterer and Schweizer, 2013) and on information on snowpack stability and wetness

5

ACKNOWLEDGEMENT

We thank the long-term SLF observers who
provide us consistently with snow and avalanche observations. We also thank Michel
Bovey for assistance with database queries as
well as Kurt Winkler and Jürg Schweizer for valuable comments on the manuscript.
6

REFERENCES

Baggi, S. and Schweizer, J., 2009. Characteristics of
wet snow avalanche activity: 20 years of observations from a high alpine valley (Dischma, Switzerland). Natural Hazards, 50(1): 97-108.
Bavay, M., Grünewald, T., Lehning, M., 2013. Response of snow cover and runoff to climate
change in high Alpine catchments of Eastern
Switzerland. Advances in Water Resources, 55: 416.
Bronaugh, D. and A. Werner, 2009. zyp: Zhang +
Yue-Pilon trends package. R package version 0.91. http://www.R-project.org/
Bründl, M.; Etter, H.-J.; Steiniger, M.; Klingler, C.;
Rhyner, J.; Ammann, W.J., 2004. IFKIS - a basis
for managing avalanche risk in settlements and on
roads in Switzerland. Nat. Hazards Earth Syst.
Sci. 4: 257-262.
Crawley, M., 2007. The R Book. Wiley, 942 p. ISBN:
978-0-470-51024-7
Darms, G., 2013: Wetter, Schneedecke und
Lawinengefahr in den Schweizer Alpen.
Hydrologisches Jahr 2012/13. Davos, WSL-Institut
für Schnee- und Lawinenforschung SLF, 30 S.
Durand, Y., Giraud, G. Goetz, D., Maris, M., Payen,
V., 2012. Modeled snow cover in Pyrennees
mountains and cross-comparisons between remote-sensed and land-based observation data.
Proceedings of the Int. Snow Science Workshop,
September 16-21, 2012, Anchorage, AK, USA.
998-1004.
Gobiet, A., Kotlarski, S., Beniston, M., Heinrich, G.,
Rajczak, J., Stoffel, M., 2013. 21st century climate
change in the European Alps. Science of the Total
Environment,
doi:org/10.1016/j.scitotenv.2013.07.050.

1245

 International Snow Science Workshop Grenoble – Chamonix Mont-Blanc - 2013
Laternser, M., 2002. Snow and Avalanche Climatology of Switzerland. Dissertation ETH Zürich, CH,
No. 14493, 139 p.
Laternser, M., Pfister, C., 1997. Avalanches in Switzerland 1500-1990. In: Frenzel, B., Matthews, J.,
Gläser, A., Weiss, M. (Eds.): Rapid mass movement as a source of climatic evidence for the Holocene. Palaeoclimate Research, 19: 241-266.
Lato, M.J., Frauenfelder, R., Bühler, Y., 2012. Automated avalanche deposit mapping from VHR optical imagery. Proceedings of the Int. Snow Science
Workshop, September 16-21, 2012, Anchorage,
AK, USA. 392-394.
Marty, C., Blanchet, J., 2012. Long-term changes in
annual maximum snow depth and snowfall in
Switzerland based on extreme value statistics,
Climatic Change,doi:10.1007/s10584-011-0159-9.
McLeod, A.I., 2011. Kendall: Kendall rank correlation
and Mann-Kendall trend test. R package version
2.2. http://CRAN.R-project.org/package=Kendall
MeteoSwiss, 2013. Federal Office of Meteorology and
Climatology, Zürich, CH. Monthly temperature
anomalies of 13 homogeneous Swiss meteo stations. Online http://www.meteoschweiz.admin.ch)
Mitterer, C. and Schweizer, J., 2013. Analysis of the
snow-atmosphere energy balance during wetsnow instabilities and implications for avalanche
prediction. The Cryosphere, 7(1): 205-216.
R, Development Core Team (2011). R: A language
and environment for statistical computing. R
Foundation for Statistical Computing, Vienna,
AUT.
ISBN
3-900051-07-0,http://www.Rproject.org/
Schneebeli, M., Laternser, M., Ammann, W., 1997.
Destructive snow avalanches and climate change
in the Swiss Alps. Eclogae geol. Helv., 90: 457461.
Serquet, G., Marty, C., Rebetez, M., Dulex, J.P.,
2011. Seasonal trends and temperature dependence of the snowfall/precipitation day ratio in Switzerland, Geophysical Research Letters, 38,
L07703, doi:10.1029/2011 GL046976.
SLF Handbuch für Beobachter (SLF observer guidelines), 1987. Eidg. Institut für Schnee- und
Lawinenforschung SLF, Davos, CH, 35 p.
Techel, F., Pielmeier, C., Darms, G, Teich, M.,
Margreth, S., 2013. Schnee und Lawinen in den
Schweizer Alpen. Hydrologisches Jahr 2011/12.
Davos,
WSL-Institut
für
Schneeund
Lawinenforschung SLF, WSL Ber. 5, 118 p., ISSN
2296-3448.
Techel, F.; Pielmeier, C.; Schneebeli, M., 2011. Microstructural resistance of snow following first wetting. Cold Reg. Sci. Technol. 65: 382-391.
Teich, M., Marty, C., Gollut, C., Grêt-Regamey, A.,
Bebi, P., 2012: Snow and weather conditions associated with avalanche releases in forests: Rare
situations with decreasing trends during the last
41
years,
Cold
Reg.
Sci.
Technol.,
doi:10.1016/j.coldregions. 2012.06.007.

1246