Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Climate Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 17 April 2021

Elevation-Dependent Climate Change in the European Alpsfree

  • Michael KuhnMichael KuhnInstitute of Atmospheric and Cryospheric Sciences, University of Innsbruck
  •  and Marc OlefsMarc OlefsCentral Institution for Meteorology and Geodynamics (ZAMG), Austria


Elevation-dependent climate change has been observed in the European Alps in the context of global warming and as a consequence of Alpine orography. It is most obvious in elevation-dependent warming, conveniently defined as the linear regression of the time series of temperatures against elevation, and it reaches values of several tenths of a degree per 1,000 m elevation per decade. Observed changes in temperature have forced changes in atmospheric pressure, water vapor, cloud condensation, fluxes of infrared and solar radiation, snow cover, and evaporation, which have affected the Alpine surface energy and water balance in different ways at different elevations. At the same time, changes in atmospheric aerosol optical depth, in atmospheric circulation, and in the frequency of weather types have contributed to the observed elevation-dependent climate change in the European Alps. To a large extent, these observations have been reproduced by model simulations.


In the time since the 1980s, the scientific community has given increased attention to the phenomenon of elevation-dependent warming (EDW) in various mountain areas across the globe. EDW is calculated as the slope obtained by linear regression of the time series of temperature against elevation (Palazzi, Mortarini, Terzago, & von Hardenberg, 2018). Observed facts and explanations have accumulated and have shown that more than temperature alone is involved in elevation-dependent warming: the changing climatic forcing has provoked changes in the atmosphere, in the surface energy budget, in altered convection, cloudiness, precipitation and water cycle, and air pollution. It is thus preferable to speak about elevation-dependent climate change (EDCC) and include changes in boundary layer stability; earth-atmosphere fluxes of energy, water, greenhouse gases and aerosols; and increasing insight into the physical processes involved. In view of the extremely dense coverage of the Alps with climatological observations and studies of the ecological and economic implications of EDCC, this review is restricted to those processes and parameters that are most typical for the European Alps, with only a few references to conditions in mountain areas elsewhere.

Global warming affects temperature T, pressure p, and vapor pressure e (Figure 1) in a direct way, each of them having typical distributions with height in the atmosphere and on the surface. While the vertical coordinate z is used for atmospheric processes, h is used to describe changes with elevation following the surface. Note that elevation dependence d/dh comes close to vertical gradients d/dz in station pairs that are close together horizontally, while it may include the horizontal changes of meteorological conditions between stations at large horizontal distance. EDCC is not linear with height, it may display transient peaks at cold air pools, in clouds, or at the snow line.

Embedded in global warming, the Alps have seen temperatures increase faster than surrounding lowland Europe. A basic explanation for this is the fact that high elevation ground surfaces absorb more solar radiation than the atmosphere at the same elevation, and they have a more intense energy exchange by turbulent mixing. However, exactly this interference of free air with Alpine surfaces depends strongly on topography, surface slope, exposure to solar radiation, and orientation with respect to prevailing winds, so that within the Alps, regional and seasonal differences of EDCC must be expected.

Figure 1. Air pressure, vapor pressure, air temperature and virtual temperature for the period 1887–2018 (anomalies relative to 1887–2018) for the high (2804 m a.s.l.) and low level (445 m a.s.l.) subgroup of stations as well as virtual temperature (Tv) calculated from air pressure at the upper and lower boundary of the “East alpine standard air column” and the directly measured mean virtual temperature of the column in the Greater Alpine Region, 1887–2018, smoothed with a 21 years Gauss filter.

Recalculated and updated following Böhm, 1998, personal communication ©ZAMG.

EDW is expressed by two variables, the rate of change or the trend ∂/∂t and the vertical gradient ∂/∂z. EDW is thus the combined effect of the change of the trend with elevation ∂/∂z(∂T/∂t) and the rate of change or trend of the vertical gradient ∂/∂t(∂T/∂z). EDW is positive if it proceeds faster at high elevation and negative if lower levels warm faster, for example, on account of a decreasing frequency of inversions. Observed changes of vertical temperature gradients are often short-lived, diurnal, or seasonal. They may, however, be more persistent when the frequency of NW flow with saturated adiabatic lapse rate changes to SW flow across the main Alpine divide with dry adiabatic lapse rate in the lee side. That would have an influence on the mean vertical temperature gradient and on the elevational gradients of precipitation. When associating altitude and temperature, note that the dry adiabatic temperature gradient of −9.8 K/km is a constant, while the saturated adiabatic temperature gradient changes with temperature, it is about −4 K/km in humid, warm air, and −6 to −7 K/km in a dry, cold environment (Wallace & Hobbs, 2006, p. 84).

Following the direct greenhouse gas forcing, a number of indirect forcings arise from feedbacks in the atmosphere and at the surface. In the decades before 1980, anthropogenic aerosols increased close to their sources near the ground and were reduced by international efforts thereafter. This gave a strong, decadal signal in solar radiation and sunshine duration at low elevation (Philipona, 2012, Philipona, Behrens, & Ruckstuhl, 2009).

These two direct, anthropogenic effects gave rise to indirect changes of elevation-dependent warming by feedbacks with water vapor, cloudiness, temperature, and aerosol optical depth in the atmosphere and with snow, soil moisture and vegetation at the surface, all of them leading to transient or long-term elevation-dependent changes. Under the influence of orography and of atmospheric circulation, EDCC is not the same at all places in the Alps and not in all seasons.

The following sections survey data sources of Alpine climatology and EDCC; discuss forcings and feedbacks contributing to EDCC; present prominent reviews of observations and model simulations of elevation-dependent warming; present records of the increase in sunshine that contributes to the observed decrease of snow and glaciers; consider the observed parallel changes of temperature and pressure; explain the role of orography and wind in the observed distribution of precipitation; and give an example of the outstanding dependence of glacier mass balance on elevation. This sequence provides a guideline through the many aspects of EDCC.

Data Sources

Among the many data sources that have been used to study EDCC, a few are mentioned here. Some of them are based on station pairs, on groups of stations, and on data grids, which make area coverage more convenient, but may contain some uncertainty for specific locations. Frei (2014) introduced a non-Euclidean method of interpolation following topography rather than flat maps.

The carefully homogenized database HISTALP (Auer et al., 2007) was developed in cooperation with weather services of all Alpine countries and contains monthly means of temperature and pressure back to 1760, of precipitation since 1800, cloudiness since 1840, and sunshine since 1880 for the so-called greater Alpine region (GAR, 4–19° E, 43–49° N, 0–3500 m asl). See Figure 2.

Figure 2. Mean seasonal air temperatures of Austrian HISTALP low level stations (colored lines) and high level stations (black lines), relative to the means of 1961–1990. Bold lines are smoothed with a 21 years Gauss filter.

Updated from Auer et al., 2007, ©ZAMG.

Efthymiadis et al. (2006) constructed a 5 arcmin grid of precipitation in the greater Alpine region for 1800–2003. Chimani, Böhm, Matulla, and Ganekind (2011) and Chimani, Matulla, Böhm, and Hofstätter (2012) used the HISTALP data set to construct a monthly temperature, precipitation, and solid precipitation grid of the greater Alpine region back to 1780 (temperature) and 1801 (precipitation) at a resolution of 6 × 9 km. Daily temperature and precipitation were gridded at 1 km resolution by Hiebl and Frei (2016, 2018) for Austria; daily precipitation for the Alps at 5 km by Isotta et al. (2013). Begert and Frei (2018) presented a new area-mean temperature series since 1864 for Swiss regions below and above 1000 m elevation which will well serve further EDCC studies.

Principles of Forcings and Feedbacks Contributing to Elevation-Dependent Climate Change

The observed greenhouse gas warming of the atmosphere and of the ground has promoted several physical consequences, among them increased melting of snow and ice, increased evaporation and convective processes that lead to tropospheric cloud condensation, and associated warming of the troposphere (Ohmura, 2012). The increase of saturation vapor pressure e* in the troposphere with feedback via the increased downward longwave radiation (e.g., Held & Soden 2000, who embedded this feedback in the global radiation balance). The value of e*, which is a function of T, determines the long wave emissivity (e.g., Ruckstuhl, Philipona, Moreland, & Ohmura, 2007). The radiation balance requires warming of the surface until radiative equilibrium is re-established (Ohmura, 2012).

The melting of snow and ice will change surface albedo and temperature at a certain elevation (e.g., Pepin et al., 2015), and cloud formation releases heat of condensation to the troposphere at confined layers (e.g., Beniston & Rebetez, 1996; Ohmura, 2012), so that associated climate change will be irregular with elevation. However, the emission of blackbody radiation at the surface and the increase of downward longwave radiation due to increased water vapor change continuously with elevation, although at different rates. Both are temperature sensitive and both follow power functions: A given increase in long wave downward radiation Δ‎E requires a larger temperature increase at low temperature (high elevation) than in a warm environment (low elevation), while it is more sensitive (requires less temperature change) to an increase in water vapor at low temperature (high elevation).

Table 1. The Sensitivity of Longwave Downward Radiation Δ‎E to an Increase in Temperature Δ‎T and to an Increase in Water Vapor ∆Q at Low and High Elevation


Low Elevation, Warm

High Elevation, Cold







These two effects may be of similar absolute magnitude, but are different in sign, and may thus give rise to seasonal and local differences in EDCC. A modeling experiment of Rangwala (2013) presented an example for this basic physical situation: Using data typical of the Tibetan Plateau (5138 m asl) and neighboring China (114 m asl)—increases of specific humidity q by 0.4 g/kg and of T by 2 K at both high and low elevation—he showed that the sensitivity of long wave downward radiation to q in winter was much higher than to T, and opposite in summer.

In a comprehensive survey, Rangwala and Miller (2012) listed the following prime feedback mechanisms contributing to elevation-dependent warming: Snow, clouds, water vapor, reflecting and absorbing aerosols, and soil moisture. They confirmed that the snow/ice albedo feedback is most effective for EDCC at the snow line or the upper limit of surface melting and pointed out that it affects predominantly Tmax. In their Table 2, they gave a thoroughly referenced list of “mechanisms that can produce an elevation sensitive temperature response at the land surface which will be dependent on elevation-based changes in the climate drivers.”

The Mountain Research Initiative (Pepin et al., 2015) contributes to the growing awareness and understanding of EDW. They listed “snow albedo and surface-based feedbacks; water vapour changes and latent heat release; surface water vapour and radiative flux changes; surface heat loss and temperature change; and aerosols” as important mechanisms that contribute towards EDW. From a thorough study of existing literature on EDW, they confirmed the opinion of previous authors, that an elevational gradient in warming rates is found in many, but not all mountain ranges. This is true even for local scales, in particular, if valley stations prone to cold air pools are compared with slope stations. Their Table 1 gives an overview of published “Increases/Decreases/Indifferent” of warming with elevation for minimum, maximum, and average temperatures for both Observations and Models. This is addressed again in the section “Models Versus Observations of EDCC.”

Ceppi, Scherrer, Fischer, and Appenzeller (2012) separated observed trends of Swiss temperature data, 1959–2008, for an elevation range up to 2700 m into RCM-based circulation effects and residual components. They attributed 53% of the observed trend to the influence of large-scale circulation in winter and 12% in fall. Observed elevational profiles of trends show maxima at low level in fall and winter, intermediate in spring, and on top in summer. Vautard and Yiou (2009) found that in Europe “during the last 60 years atmospheric circulation changes are the main drivers of surface weather trends in winter but not in summer where temperature strongly interacts with the water cycle.”

In their review of elevation-dependent warming in the Swiss Alps, 1981–2017, Rottler, Kormann, Franke, and Bronstert (2019) used daily, homogenized data of 28 Swiss meteorological stations separated in three elevation categories. They summarized that

elevation-based differences in temperature trends occur during autumn and winter with stronger warming at lower elevations. We attribute this elevation-dependent temperature signal mainly to elevation-based differences in trends of incoming solar radiation, cloud cover, air humidity, snow/ice and elevation dependency of temperature trends.

(Rottler et al., 2019, p. 1.)

This is not in contrast to earlier research. In the annual averages, temperature and sunshine duration have a clear elevation dependence, snow cover to a lesser degree, as it depends more on local topography. In the period 1981–2017, the yearly cycles of trend magnitudes show cooling in winter at high elevation versus warming at low elevation, in agreement with Hedenig’s (2019) analysis of temperature in the Austrian Alps. Rottler et al. (2019) attribute significant changes in the trends of all parameters in March/April and June to changes in large-scale circulation.

Topography and exposure play a significant role in the elevation dependence of climatic changes in the Alps. In their analysis of patterns of temperature trends at high elevation, Pepin and Lundquist (2008) gave evidence to a peculiarity of mountain meteorology: trends at high, well-ventilated peaks like Jungfraujoch, Säntis, Zugspitze, or Sonnblick, are more homogeneous, have less variability than those of slopes and incised valleys. This is in agreement with Rottler et al. (2019, Figure 4) and Salzmann, Scherrer, Allen, and Rohrer (2015).

It is appropriate to refer to the influence of vegetation at the end of this section, as it is involved in rather slow feedback mechanisms. Tudoroiu et al. (2016) reported negative EDW (stronger warming at lower elevation) from the Trentino region in the NE Italian Alps. They “analyzed weather records for the period 1975–2010 . . . that show that warming occurred both at high and low elevations, but it was less pronounced at high elevations” and make changes in aerosol optical thickness, in cloudiness, and in land use including reforestation at high elevations, responsible for the negative elevation dependence.

Johnston et al. (2018) gave an interesting sideline to the present EDW from speleothems of the last interglacial (130–115 ka BP). By comparing specimens of that time from high-elevation Alpine caves to those of low-elevation European caves, they arrived at values of EDW in the last interglacial similar to the Alpine conditions around the year 2000.

Models Versus Observations of EDCC

Early modelling of EDCC, such as Giorgi, Hurrel, Marinucci, and Beniston (1997) and Beniston, Diaz, and Bradley (1997), based on the model resolution possible at that time, confirmed the expectation that warming increased with elevation and was strongest in winter and spring. However, Pepin and Lundquist (2008, p. 1) remarked that “most climate models suggest amplification of global warming in high mountains, but observations are less clear,” and Palazzi’s view (2019, slide 2) is that “observations of EDW are less in agreement with each other than models,” which reflects the many situations described in the section “Principles of Forcings and Feedbacks Contributing to Elevation-Dependent Climate Change.” In general, models have fewer variables and fewer interconnections and feedbacks than actual nature. Thus, expectations should be modest and gratitude should be shown because the models do allow a better understanding than measurements alone could at this time.

In their review of the 21st century climate change in the European Alps, Gobiet et al. (2014) used simulations of the Ensemble Project with a 25 km grid. They devoted an entire section of their work to the uncertainties of their models and parametrizations. Palazzi et al. (2018), in their study of elevation-dependent warming in global climate simulations at high spatial resolution, gave a thorough summary of models applied for the study of elevation-dependent warming, their resolution, and performance in the Greater Alpine Region, the Colorado Rocky Mountains, and the Tibetan Plateau—Himalayas.

In contrast to Rottler et al. (2019), who give snow albedo feedback a minor role from the analysis of Swiss records, simulations of both Palazzi et al. (2018) and Minder, Letcher, and Liu (2018) placed snow albedo on top of the list of EDCC mechanisms responsible for the warming at intermediate elevations. While water vapor and its IR emissivity has a nearly continuous vertical distribution in the Alpine atmosphere, a strong change of albedo appears in rather concentrated elevation bands around the snow line.

Minder et al. (2018) did not find any evidence for a contribution of elevation-dependent water vapor feedbacks and pointed out that model EDW “depends strongly on certain aspects of RCM configuration.”

These are discrepancies between observations and simulations that may be solved by models of higher resolution and better downscaling. Lüthi, Ban, Kotlarski, Steger, and Schär (2019), who simulated Swiss snow cover at 2.2 km made a valuable step in that direction. Until model resolution reliably reaches the spatial scales of orography and associated convective activity, of the order of 1 km, care must be taken to be aware of the possible bias and simplifications of regional climate models in support of EDCC values.

Solar Brightening: Increase In Sunshine

The intensity of solar radiation at the Earth’s surface and the duration of sunshine are determined by the concentration and type of atmospheric aerosols and by the presence and type of clouds. The global increase of industry in the second half of the 20th century has increased aerosol optical depth (AOD) of the atmosphere in the boundary layer and lower troposphere, which contributed a noticeable component to EDCC, the so-called global dimming. (Global in this context does not mean worldwide, but refers to the term global radiation meaning shortwave radiation from the upper hemisphere). With strong and successful international efforts to curb air pollution in the 1970s, this effect was substantially reduced since, increasing in turn sunshine and incoming shortwave radiation, the so-called solar brightening. The increase in solar radiation at the ground implied increased absorption of radiation and thereby increased warming of the ground. While greenhouse gas forcing gives the strongest warming at top elevations, the reduction of aerosols at their highest concentration near the sources at the ground contributed to stronger warming at low elevations (Philipona, 2012; Philipona et al., 2009).

This is supported by the decreased frequency of inversions (Figure 4). Since increased warming at low elevations is transferred upward to higher elevation by convection and up-slope winds this may be an explanation why the mean temperature of the Alpine atmosphere has risen more than that of lowland Europe. In general, a ground surface at a given elevation has a stronger positive or negative energy exchange than air at the same elevation. The net result is most likely positive due to the absorption of solar radiation by the ground, but that is not universally so.

Ramanathan, Crutzen, Kiehl, and Rosenfeld (2001) gave a compact description of atmospheric aerosols and their radiative forcing. Aerosols like sulfates scatter and reflect solar radiation, while black carbon particles absorb radiation (both are referred to as direct radiative forcing). Aerosols act as cloud condensation nuclei and thereby increase atmospheric reflection (first indirect radiative forcing). Large numbers of particles create many small cloud droplets that are less likely to fall out as precipitation and thus prolong the time with high atmospheric albedo (second indirect radiative forcing).

Scattering on dust, air pollution particles and fog is most pronounced on valley bottoms and at low elevation, clouds may reach and exceed the elevation of Alpine peaks, both giving elevational dependence to radiative forcing and sunshine duration. Marty, Philipona, Fröhlich, and Ohmura (2002) presented details of the mean Alpine radiation balance and aerosol optical depth and its strong elevational dependence. Philipona (2012) associated the stronger mean warming observed in the Alps compared to lowland Europe with the stronger increase of water vapor and found it supported by stronger advection of moisture towards the Alps. For EDCC, however, the declining aerosol optical depth since the 1980s is the dominant cause (Manara et al., 2016; Philipona et al., 2009; Ruckstuhl et al., 2008; Tudoroiu et al., 2016; Wild, 2009; Zeng et al., 2015).

Figure 3. Mean annual sunshine duration 1884-2018 at low level (445 m a.s.l.; blue lines) and high-level stations (2074 m a.s.l.; orange lines) of the Austrian HISTALP stations. Deviation from the mean of 1961–1990) in percent. Thick lines are smoothed with a 21 years Gauss filter.

Updated from Auer et al., 2007. ©ZAMG.

The time series of mean annual sunshine duration given in Figure 3 supports the observations of solar surface radiation: at low elevations (blue curve) there is a stronger decrease (dimming) from 1940 to 1980 than at high elevation (orange curve), and there is a stronger increase (brightening) since then.

Stocker et al. (2013) indicated the total aerosol effect in the atmosphere including cloud adjustments due to aerosols as −0.9 to −1.9 W m−2, compared to +3 W m−2 for the effect of well-mixed greenhouse gases.

A Barometric Proof of Rising Air Temperatures

By necessity, warming of the Alpine atmosphere, even if uniform at all elevations, leads to elevation-dependent pressure changes. The hypsometric equation (e.g., Salby, 2012; Wallace & Hobbs, 2006) establishes a link between the elevation difference of a mountain and a valley station, their respective pressures, and the mean virtual temperature Tv of the layer between the two stations. The virtual temperature accounts for the varying humidity of the air, but Böhm, Auer, Schöner, and Hagen (1998) have shown that “all results for Tv are highly similar to temperature itself.” This equation implies that changes of the mean temperature of the valley atmosphere may be reconstructed from recorded changes of pressure at low and high elevation observatories. A further advantage of this procedure is that past records of pressure are more representative, regionally, than past temperature records.

In Figure 1, an increase of the amplitude of p(t) with elevation is obviously modulated by concurrent variation of mean valley temperature and, to a much lesser degree, of humidity. Both Böhm et al. (1998) and Toumi, Hartell, and Bignell (1999) noted that the pressure signal in reaction to warming increases with height. Böhm et al. (1998) have shown the agreement between their calculations of Tv from pressure versus observed values for the period 1887–1996. They arrived at a correlation R2 = 0.86 between observed and calculated column mean of virtual temperature between high and low stations in the Eastern Alps. An application of the method to updated data until 2018 increases the correlation toward R² = 0.89 (see Figure 1).

Compared to a perfect correlation R2=1.00, the small difference ∆R2=0.11 leaves the question whether it is only due to random fluctuations. Certainly elevational pressure differences are not determined by local thermodynamics alone but react to changes in larger scale synoptic patterns as well. Beniston and Jungo (2002) found shifts in the distribution of pressure, temperature, and moisture, and changes in the typical weather patterns in the Alpine region in response to the behavior of the North Atlantic Oscillation. With respect to EDCC, they noted, “At low elevations, the NAO signal may be weak or absent in the Alps, higher elevation sites are sensitive to changes in NAO patterns.” Rottler et al. (2019) analyzed the effect of 26 Weather Type Classes on temperatures of three elevation categories in the Swiss Alps and found significant seasonal variability. Investigating surface pressure and 2m air temperature of 60 Swiss stations, Salzmann et al. (2015) found the correlation coefficients between surface temperature and surface pressure increasing, from 0.2 at 500 m elevation to 0.85 at 3500 m.

You et al. (2017) investigated the relationship between observed warming and surface pressure in the Tibetan Plateau. They noted that the correlation between temperature, pressure, and elevational amplification of warming is strongest in winter, which may be due to more frequent synoptic disturbances in summer.


Figure 4. Seasonal trends in daily temperature minima 1961–2017 as a function of elevation in Austria. Absolute changes over 57 years from Theil-Sen linear trend slopes are shown. The median line, the dark shaded interquartile range and the light shaded 95% confidence interval of the median, based on 10 by 10 km subdomains, indicate statistical distribution.

Modified from Hiebl and Schöner, 2018. ©ZAMG.

Changes in elevational temperature gradients in mountain areas are part of global warming due to greenhouse gases; they depend on local topography, changes of aerosol optical thickness, cloudiness, and large-scale atmospheric circulation. In this situation, Salzmann et al. (2015) pronounced that the “main conclusion of the altitudinal dependence of daily temperature anomalies in complex terrain is that due to local exposition and weather processes that can be of similar importance as the station altitude, there is no universally valid altitude variability relation,” and this leads to the altitude variability of trends as well.

There are, however, a number of features that modern studies have in common. Ceppi et al. (2012), among others, reported the largest linear temperature trends at low elevations, in particular in winter and in fall. By principal component regression of temperature trends with geopotential heights, they attributed 53% of the winter trend and 12% of the fall trend to changes in atmospheric circulation. Hiebl and Schöner (2018) investigated the trends of temperature inversions in Austria based on a data set by Hiebl and Frei (2016). In that period, the frequency of inversion days per year decreased by 11%, their intensity by 0.02 K/100m, and the mean magnitude by 0.3 K. Figure 4 displays the elevation dependence of changes of monthly mean Tmin, 1961–2017. Pronounced warming at low elevation and highest spatial variance occurred in December, January, July, and October.

The complexity of Figure 4 reflects the variability of shallow cold air pooling and the stabilizing effect of a valley-bottom snow cover. The authors point at the significant difference between elevational changes of surface records and valley free air temperature (Dreiseitl, 1988; Nickus & Vergeiner, 1984; Pepin & Seidel, 2005).

The elevational change of the daily temperature range (DTR) was addressed by Weber et al. (1997). This is again a parameter that is dominated by the surface energy balance so that largest DTRs are found in valleys where radiation and latent heat fluxes lead the energy balance and smallest DTRs at peak stations where sensible heat loss in stronger winds reduces the surface-free air differences. Figure 2 of their article illustrates these effects for mean monthly values of DTR of 23 Alpine stations.


Precipitation in the Alps depends greatly on the season, the elevation, and the screening by orography. The North-South transect of annual precipitation across the Alps in Figure 5 (Frei & Schär, 1998) clearly shows a large screening effect with dry interiors and maxima at intermediate elevation at either side of the Alpine chain. With significant horizontal precipitation gradients, the determination of elevational precipitation gradients requires short-distance station pairs.

Houze (2012) has given a comprehensive survey of orographic effects on precipitation. While there are many publications on the regional distribution of precipitation, notably by Fliri (1975) and Frei and Schär (1998) for the Alps, and while a limited number consider the local dependence of precipitation on elevation (e.g., Auer, 1992; Schwarb, Daly, Frei, & Schär, 2001; Sevruk, 1997; Sevruk & Mieglitz, 2002), no publication has been found on the change of elevational precipitation gradients with time, based on observations.

Gobiet et al. (2014) have projected precipitation change for 100 m elevation bands in the Alps from 1961–1990 to 2070–2099, based on ten GCM-RCM chains assuming the SRES A1B emission scenario. In the elevation range up to 2500 m, this resulted in increases by +10 to 20% in winter (DJF) and by −30 to −10% in summer (JJA), with the strongest changes appearing at low elevation. Wastl and Zängl (2008) simulated the mountain-valley differences in precipitation with the MM5 model and found strong dependence on wind direction.

Figure 5. Mean annual precipitation (box plot) and topographic height (solid line) along a north-south section across the eastern Alps. The data represents the zonal mean of the analysis between 10.2°E and 12.6°E. The symbols (see insert for legend) depict the interquartile range and the minima and maxima of the distribution of mean annual precipitation in the corresponding latitude belts.

From Frei and Schär, 1998.

Snow and Ice

Figure 6. Elevation dependent seasonal course of snow depth 1961–1990 vs. 1981–2010. Stations situated at the North side of the Eastern Alps (Zürs, St. Anton), in the dry interior (Umhausen) and on the South side (Villacher Alpe, Kanzelhöhe, Bad Bleiberg).


Figure 7. Relative changes of average snow depth (left) and snow cover duration (right) derived from linear trends (Theil-Sen slope) in the period 1961 to 2018 as a function of elevation for the winter season Dec to Feb (top) and Nov to Apr (bottom) for 65 Austrian stations. The trend significance (95% confidence interval) was estimated using the Mann-Kendall trend test (Kendall, 1975; Mann, 1945). Red dots are significantly negative trends, gray represents no significant trends.


The temporal evolution of the snow cover in the course of one season is spatially characterized by differential snow accumulation and ablation processes. Whereas the altitude of snow/rain transition, the precipitation amount, and redistribution by wind and avalanches dominate snow accumulation, the energy balance of the snow surface governs snow ablation. Thus, snow climatology is governed by local topography (altitude, exposition), larger scale weather patterns (air masses and their provenance), and other regional to local scale effects (e.g., frequency of Föhn winds, cold air pooling, orographic effects of precipitation, snowline depression effects).

As both the snow accumulation and ablation processes are influenced by redistribution of snow already lying on the ground and preferential deposition of solid precipitation caused by wind, small scale spatial variability of snow cover is generally high and flat field stations or precipitation gradients cannot explain observed snow amounts (Grünewald & Lehning, 2011). Orographic effects of precipitation reflect the altitudinal distribution of the observed snow cover (Grünewald, Bühler, & Lehning, 2014) together with terrain features (Lehning, Grünewald, & Schirmer, 2011) that also determine the interannual persistence of snow distribution (Helfricht, Schöber, Schneider, Sailer, & Kuhn, 2014). Elevation-dependent correlations of snow depth with air temperature and precipitation in Austria, for the period 1961–2016, indicate that the snow cover is very sensitive to temperature changes up to 2000 m asl, which can be compensated for by precipitation (increasing sensitivity with rising altitude) above around 1300 m asl (Olefs, Koch, & Gobiet, 2019; Schöner et al., 2018). Due to the strong temperature dependence of snow ablation (e.g., Hock, 2003), the decrease of snow depth (SD), especially of snow water equivalent (SWE), and the number of snow covered days (snow cover duration—SCD) is strongly and approximately linearly related to altitude and ranges from −30% to −90% (SD) and 0% to −70% (SWE, SCD) in the period 1961–2018 (Klein, Vitasse, Rixen, Marty, & Rebetez, 2016; Marke, Hanzer, Olefs, & Strasser, 2018; Marty, Tilg, & Jonas, 2017a; Schöner et al., 2018; see Figure 6 and Figure 7).

Projected future changes of the Alpine snow cover are most reliable in altitudes below around 1500 m asl, where temperature sensitivity is high and precipitation sensitivity is low (a systematic model study of sensitivity is Schlögl, Marty, Bavay, & Lehning, 2016). Moreover, climate model uncertainties of precipitation scenarios are generally much larger than those for temperature (Gobiet at al., 2014; Olefs et al., 2019). Expected snow cover decrease in the far future (2071–2100) is strongly elevation dependent and ranges from −20% to −70% (high, middle, low elevation—3000, 2000, 1000 m asl) for the climate mitigation scenario (RCP 4.5), and from −30% to −90% (high, middle, low elevation) without mitigation measures (RCP 8.5) (CH2018, 2018; Frei, Kotlarski, Liniger, & Schär, 2018; Gobiet et al., 2014; Kotlarski et al., 2018; Schmucki, Marty, Fierz, & Lehning, 2015; Steger, Kotlarski, Jonas, & Schär, 2012; Verfaillie et al., 2018). This may yield a SCD reduction of 2 weeks to 1 month at the beginning and 1 to 3 months at the end of the winter season (corresponding to a halving of SCD at 1500 m asl, Marty, Schlögl, Bavay, & Lehning, 2017b).

Compared to annual snow cover, the reaction of glaciers to climate is slow, but its elevation dependence is strong (e.g., Beniston et al., 2018). The surface elevation h of a glacier is a function of space and time h(x, y, t). It is determined by the local mass balance of snow and ice and by the slow flow of the glacier (e.g., Benn & Evans, 2010; Charalampidis et al., 2018; Cuffey & Paterson, 2010; Oerlemans, 1989). The mass balance is directly connected to climate and topography. Accumulation supplies mass by snowfall, snowdrift, and avalanches; ablation consumes snow and ice, predominantly by melting. In the equilibrium state of a glacier, the annual loss in the lower ablation area is resupplied by ice flow from the higher accumulation zone, which typically is about twice as large as the ablation area. Figure 8 shows the elevation profile of the specific mass balance (expressed as kg/m2 or mm of water equivalent) of 50 m elevation bands of Hintereisferner, a well-developed valley glacier in the Austrian Alps. Details of elevation-dependent mass balance profiles have been presented by, among others, Greuell and Böhm (1998), Greuell, Knap, and Smeets (1997), Huss, Farinotti, Bauder, and Funk (2008b), Kuhn (1979), and Petersen, Pellicciotti, Juszak, Carenzo, and Brock (2013). In the period 1971–1980, glacier mean specific mass balance was positive in many Alpine glaciers, while it was generally negative since 1993. The elevation profile Δ‎ b(h) of the difference between the two curves in Figure 8 is influenced by the topography of the glacier (Charalampidis et al., 2018).

At the beginning of the 21st century, many Alpine glaciers had become so thin that resupply by ice flow from their accumulation areas to their tongues was greatly reduced. Since then, it was rather a decay than a retreat of glaciers and elevation-dependent loss of ice was not a straightforward function of EDCC anymore.

Figure 8. Mean specific mass balance of Hintereisferner (46°48’ N, 10°46’ E) expressed as kg/m2 or mm of water equivalent, in 50 m elevation bands for the period 1971–1980 which was close to equilibrium, and for 1993–2002, which is typical for the recent, warm decades. Values above 3500 m elevation react more to snowdrift and avalanches than to warming.



In summary, there are many mechanisms for elevation dependence in operation at the same time, some independently, most of them interconnected by feedbacks. Some of them are more effective at cold, high elevation, others at warm, low elevation, some change with the season or the time of the day. Many of them depend on topography/orography—valley bottom, slope exposure, or peak situation, blocking or channeling, or screening in the dry interior of the Alps. This multitude of regional boundary conditions explains the occasional disagreement between observations and model simulations. EDCC is not a consequence of local or regional conditions alone—it is influenced also by the elevation-dependent effects of general circulation and the frequency of weather types and changes in the aerosol optical depth.


This review is dedicated to the memory of Reinhard Böhm (1948–2012), pioneer in Alpine climatology.

We are thankful to Zentralanstalt für Meteorologie und Geodynamik in Vienna, in particular to Ingeborg Auer, Johann Hiebl, and Wolfgang Schöner for their support with data and figures.

Further Reading