Introduction

Human activities are often threatened by snowdrifting. In our area this occurs principally in high mountain environments. Wind episodes distribute the snow irregularly, producing accumulation areas that increase the avalanche hazard.

Snowdrifting also affects the road and railway network, producing disturbances in both and increasing danger to traffic due to reduced visibility.

In ski resorts the effects of the wind on open areas are reflected in the erosion of snow from ski runs, causing important economic losses.

Snowdrifting has often been studied from its theoretical point-of-view (Reference KobayashiKobayashi, 1972; Reference MartinezMartinez, 1996; Reference Naaim-Bouvet and Naaim.Naaim Bouvet and Naaim, 1995). However, few of these studies have been undertaken in high mountain areas with complicated relief (Reference Castelle, Roussel and Clappier.Castelle, 1990; Reference DuclosDuclos, 1993; Reference Mases, Vilaplana., Arnáez, GarCía-Ruíz and Gómez VillarMases and Vilaplana 1994; Reference Mases, Buisson, Good and Vilaplana.Mases and others, 1995).

An experimental plot has been equipped in the eastern Spanish Pyrenees on the highest slopes of the La Molina ski resort. Working in a ski area considerably simplifies access to the site in all weather conditions.

This paper describes the gauges used at the La Molina plot, the measurements done, the type of data obtained and the results inferred from these with respect to the distribution of snow in a natural snowdrift.

Study Area

The experimental site is on the highest slopes (Torrent Negre) of the La Molina ski resort in the Catalan Pyrenees, in northeastern Spain (Fig. 1). It is accessible by skilift.

Fig. 1. The study area.

The Torrent Negre area is characterized by high erosion surfaces or platforms and by a flat topography. However, there are some slope changes that produce snowdrifts. At one of these snowdrifts have been monitored. The gauges are installed in this area.

The altitude range in the study area is between 2000 and 2400 m on a slope mainly exposed lo the north. Mean annual precipitation is about 1211.2 mm of water; mean annual snow precipitation is about 414mm of water and die average snow-accumulation thickness is about 2.5–3 m at 2050 m. Precipitation comes mainly from the east and is often followed by strong north winds that immediately re-distribute all the fresh snow. The snow lies on the ground about 137 days per year at 1711 m. The mean annual temperature is about 5.4°C, and the February average temperature is about −1.6°C (Reference SalvadorSalvador, 1985).

Wind is characterized by high-intensity episodes that come from the north and northwest. Precipitation comes mainly from the east and is often followed by strong north winds that immediately redistribute all fresh snow.

Experimental Equipment

Several gauges were installed to measure wind episodes and the resulting redistribution of snow during the 1992–93, 1993–94, 1994–95 and 1995–95 seasons. They were the following:

The poles in the snowdrift were measured once a week. The snow collectors were sampled once a week and immediately after snowdrifting occurred, when the snow collected was weighed.

Different measurements were also made on the snow-pack. Once a week a snow-stratigraphy profile was done and each day weather conditions were measured and the type of snow on the surface recorded.

These data allowed us to characterize the episodes of snow transport. Following the data analysis, we consider that a wind episode begins when the speed exceeds 6 ms⁻¹; below this snowdrifting very seldom occurs.

Comparison Between Drifted Snow and the Development of a Snowdrift

Measurements on the poles indicated the evolution of the thickness of the snowdrift.

In Figure 2 the snowdrift evolution on a north-south orientation for the 1992–93 season is shown. The snowdrift grew slowly until the end of February. On the 26 February 1993 an important increase was observed, and also at the next measurement on 9 March 1993. From mid-March until 27 April snow loss was the main process in the snowdrift. This loss could be produced either by melting or by winds redistributing die snow.

Fig. 2. Snowdrift evolution (north-south) orientation during the 1992–93 season.

From interpretation of the profiles obtained during the last three seasons we observed:

A comparison between volume variations of the snowdrift and the snow transported by the wind would be of great interest. Distances between the poles in the snowdrift were too great to calculate the total volume of the snowdrift. In order to discover the growth and changes in the snow-drift thickness, an imaginary vertical plane through the two lines of snow poles was used to calculate surface variations. These surface changes are representative of the volume variations in the snowdrift (Fig. 3).

Fig. 3. An imaginary vertical plane through the lines of poles allows us to calculate surface variations. T₀ = Time 0 (before a snow-drifting event) and T₁ = Time 1 (after the snow-drifting event). S = total incitase in m² of the snowdrift in one of the lines (northwest—southeast or north-south).

In Figure 4, the number of episodes (snowdrifting events vs the variations (m²; Fig. 3) of the snowdrifl thick-ness for each wind episode is presented for the 1992–93 and 1993–94 seasons. The first two snowdrifting episodes during a winter season are those that produce the most important increase in area (Fig. 3). However, loss of snow due to a wind episode from the opposile direction or high temperatures, producing a decrease in the snow-cover thickness by melting, changes the snowdrifl profile and thus the evolutionary stage of the snowdrift.

Fig. 4. The number of episodes (snowdrifting events) vs variations of the snowdrift thickness (m²). Refer to Figure 3. Lines are fit to the data. Seasons of 1992–93 and 1993–94 for the north-south line are plotted.

Reference TablerTabler (1975) pointed out that the increase of snow in a snowdrift is inversely proportional to the amount of snow accumulated. The efficiency of snow trapping from a slope change (snowdrift) decreases while the natural snowdrift is being built. This is what was observed at the La Molina snowdrift during the 1992–93, 1993–94 and 1994–95 seasons.

Table 1 shows the amount of snow collected in the two snow-collector columns (oriented north and northwest), the growth of the snowdrift (Fig. 3) and the main meteorological parameters (wind velocity, air temperature, snow-grain types and the occurrence of precipitation).

Table 1. Description of episodes

The wind episodes of 28 December, 1 April 1993 and 22 April 1994 show similar features. The air temperature ranged between 3° and −20°C during the three episodes. While snowdrifting occurred there was precipitation during the three episodes and the amount of available snow to be drifted away was also similar. Wind velocities were similar during the first two episodes, above 25 m s⁻¹, but during the third one the velocity was much higher. The snow-collector columns collected 5875g during die first episode, 5255g during the second and 5330g during the third. This indicates that the quantity of drifted snow was also very similar during the three episodes. However, during the 28 December 1993 episode the snowdrift grew 64.8 m² in the northerly direction and 52.6 m² in the northwesterly direction. During the 4 January 1993 episode, the growth was 5.75m² and 8.85m² and in the 22 April 1994 episode it was 10.35m² and 13.2m², respectively in the northerly and northwesterly direction (Fig. 3 and Table 1). The differences found by comparing these three episodes are assumed to be different stages of evolution between the first case and the other two. The second and third episodes represent similar evolutionary stages, in both cases the snowdrift was approaching its equilibrium profile (Reference TablerTabler, 1975).

Snowdrifting episodes with similar features produce difierent rates of growth in a snowdrift depending on its evolutionary stage. We have assumed that the amount of transported snow is not the main control during snowdrift evolution.

Considering the proximal zone (windward area), and the distal zone (leeward area), of the snowdrift, it is in the distal zone where volume changes are more pronounced.

Wind intensity principally affects the relative position of the maximum rate of growth of snowdrifting. Strong winds tend to increase the volume of the snowdrift in the distal zones rather than in the proximal ones (Fig. 5).

Fig. 5. Variations of snowdrift depending on the wind intensity.

Application of Reference TablerTabler’S (1975) Snowdrift-Development Model

Reference TablerTabler (1975) developed a statistical model to predict the maximum profiles of snowdrifts. He studied their evolution in Wyoming. His model fits in flat environments but cannot be applied to high-mountain environments or to snowdrifts with high angles.

Reference Taillandier, Verreault and Vachon.Taillandier and others (1990) proved that the same requirements are preserved while working with scale models in real conditions. The La Molina snowdrift is smaller than those studied by Reference TablerTabler (1975) so to apply his model we have increased the size of our snowdrift by three times (Reference Taillandier, Verreault and Vachon.Taillandier and others, 1990).

In Figure 6, we show profiles derived for three seasons 1992–93, 1993–94 and 1994–95, and also the predicted profile using Tabler’s model for the La Molina snowdrift. The three seasons show differences when compared with the model. In the 1992–93 season, Tabler’s model fits with our snowdrifl profile. In the 1993–94 season case b), we observed that in the 11 February 1994 measurement the snowdrift presented a profile above Tabler’s predicted equilibrium profile. This situation was produced by a snow-drifting episode characterized by low-intensity winds (Fig. 5). However, later episodes (18 March 1994) removed the snow that exceeded Tabler’s profile and the snowdrift reached its equilibrium profile. Finally, in the 1993–94 season, the snowdrift did not reach Tabler’s equilibrium profile. This season was characterized by very low precipitation rates, which meant there was little snow to be drifted. We also observed that, for the three seasons presented in Figure 6, Tabler’s model fitted with the topographic features of our study area.

Fig. 6. Profiles of the La Molina snowdrift the 1992–93, 1993–94 and 1994–95 seasons, compared with those predicted by Reference TablerTabler’s (1975) model.

Conclusions

From the experimental data obtained at La Molina we observed that the factor controlling the growth of a snowdrift is its evolutionary stage. The amount of snow transported during a wind episode is noi the main factor. The evolutionary stage is directly related to the number of snowdrifting episodes that had occurred previously. These observations are in agreement with those of Reference TablerTabler (1975).

An identical blowing-snow event produces different rates of growth in a snowdrift depending upon whether it occurs in the initial or final stage of the evolution.

Wind velocity principally affects the location of the snow deposit, the stronger the wind velocity the further away (distal) the snow accumulates as a drift, and the lower the velocity the nearer (proximal) it accumulates.

By comparing the evolution of the snowdrift with Reference TablerTabler’s (1975) model we observed that for the three seasons presented it fitted with the topographic features of our study area.