High mountain area in La Grave

Our case study focuses on the high mountain area of La Grave, located in the Oisans massif in the French Alps (Figure 1A). La Grave is located within the perimeter of Ecrins National Park. By car, it takes 2 hours to reach it from Lyon and 1.5 hours from Grenoble (Figure 1B). Conventional ski resorts offer many types of infrastructure, such as lifts and ski runs. However, La Grave differs greatly from this. It offers almost exclusively off-piste itineraries, where skiers can make their paths in barely marked areas, called the high mountain area.

FIGURE 1

Multiscale map of La Grave, 45°02′39.6″N; 6°18′12.2″E. (Map by Lucas Berard-Chenu; photos by Nao Yoshizawa)

The high mountain area in the La Grave operation is equipped with a single cableway that allows skiers to ascend from 1400 to 3200 masl (Poulain and Garcia 2024). It consists of 2 sections (Figure 1C). The first connects the village, which is situated at 1400 masl, to the intermediate station at 2400 masl, and the second connects the intermediate station to the upper station at 3200 masl. The same cableway equipment has been in operation since the opening of the cableway in July 1976. It has a capacity of 30 people, with a group of 5 gondola cabins arriving at the upper station every 5 minutes. This relatively low capacity limits the number of skiers. At the end of the second section, at 3200 masl, skiers can take a T-bar ski lift, which was installed in 1990 on the Girose glacier, to reach the top of the mountain at 3550 masl. The cableway company is responsible for its customers' safety until they exit the cableway station, as well as inside the marked and secured area of the glacier tracks. Outside the boundary, skiers are free to ski at their own risk. The terrain for skiing is ungroomed and unmanaged, and there is no avalanche control (Figure 1D). Freeriding is the main activity in La Grave, but there are some routes reached by hiking from the 3200-masl station; therefore, ski touring is also practiced. As a result of this particular configuration, La Grave mainly attracts visitors who are capable of managing risks by themselves.

Despite a vertical drop of more than 2000 m, La Grave has attracted relatively few visitors, with 25,000 to 35,000 skier visits in the past 5 seasons. In comparison, Les Deux Alpes and L'Alpe d'Huez, 2 ski resorts located near La Grave, each had approximately 1.1 million skier visits during the 2018/19 season (Lagneux 2019). Figure 2 shows the annual number of La Grave cableway ticket sales in winter from 1976/77 to 2022/23. The number of ticket sales refers to the number of people who purchased one of the various offers for the cableways marketed by the operator. The cableway company markets 4 types of products: single pass, 1-day pass, several-day pass, and season pass. In the company's commercial statistics, each ticket sale is counted as 1 regardless of the ticket type. The number varied each year, with an increasing trend from 1976/77 to 1992/93 and a decreasing trend after the 2010s. Cableway operations were suspended because of insolvency in 1986/87 and the COVID-19 crisis in 2020/21. Figure 3 shows the average ratio of ticket sales by month for each decade. It indicates a seasonal shift in ticket sales from spring to winter.

FIGURE 2

Annual number of ticket sales in winter in La Grave (1976/77–2022/23).

FIGURE 3

Average ratio of ticket sales by month per decade (1980/81–2022/23).

Variables selected and data collected

Analyzing the determinants of off-piste skiing demand requires a finer scale of observation than the annual number of skiers. The dataset encompassed 5 ski seasons, 2017/18–2022/23, for which daily skier visit data were available, incorporating variables identified as follows.

Number of skier visits: This was provided by the cableway company and represented the number of people attending the ski area for all or part of a day. The cableway was occasionally closed because of strong winds or high avalanche risk. On the morning of a day when the avalanche danger level rises to 3 or higher, an inspection and meeting to decide on closure is held by local residents and guides, without representatives of the cableway company. In the case of closures, the number of skiers was 0; therefore, we excluded 52 such days over the 5 seasons. We also excluded 6 days of exceptional attendance because of local events (eg Derby de La Meije in early April).

Avalanche risk assessment data: The Avalanche Risk

Assessment Bulletin for the Oisans massif by Météo-France provides data for 3 independent variables: avalanche risk, fresh snow, and weather.

Months: A categorical variable for months of the season was included; this could take 5 modalities: December, January, February, March, or April. We also included categorical variables for the 5 specific periods we considered in comparison to nonspecific periods: Christmas vacations, winter vacations, spring vacations, weekends, and holidays outside of vacation periods.

Lagged dependent variable: We included skier visits on the previous day as a lagged dependent variable, because several studies on ski areas have shown that skier visits persist over time. For example, Malasevska et al (2015) and Falk et al (2022) showed that past frequentation values have a positive and significant effect.

Before and after the COVID-19 pandemic: To incorporate changes in visitation patterns before and after COVID-19, a dichotomous variable with pre- and post-COVID values of 0 and 1, respectively, was integrated into the analysis. We defined post-COVID as after the 2020/21 season.

In addition, we conducted an interview of approximately 1 hour with the manager of the cableway company on 17

February 2022 to understand the characteristics of the clientele.

Model specification and estimation method

We used a regression model to analyze the relationship between skier visits and the following selected factors:

where t denotes time; skier_visit_t represents the dependent variable; skier_visit_{t-1_} represents skiers' attendance during previous periods, capturing the influence of past trends and patterns; fresh_snow_t refers to snowfall measured during the day and night; weather_t represents the forecasted weather; avalanche_risk_t represents the estimated avalanche risk of the day; month_t represents the current month; specific_period_t captures whether the current day is in a specific period; post_covid_t indicates whether the season is in a post-COVID period; β₀ to β₆ are the regression coefficients, representing the expected change in skier visits associated with each respective independent variable, and ε represents the error term, accounting for unobserved factors and random variations in skier visits that are not explained by the independent variables. We applied a natural logarithm transformation to the visit variables relating to skiers and past skiers to address potential heteroscedasticity or nonlinear relationships between the variables. This transformation was intended to stabilize the variance and allowed the regression coefficients to be interpreted as approximate percentage changes in skier visits.

We used the statistical software STATA (v15.1) to estimate the regression model coefficients with the ordinary least squares (OLS) estimator, which minimizes the sum of the squared residuals to obtain parameter estimates. Following OLS regression model estimation, we conducted several diagnostic checks to assess the validity of the model's assumptions and evaluate the reliability of the estimated results. (See Narayan 2003 for more information regarding the diagnostic tests to model tourism demand.) We used robust standard errors to avoid an autocorrelation or heteroscedasticity bias in our estimates. We assessed the presence of unit roots in the dependent variable using the augmented Dickey–Fuller test (Dickey and Fuller 1979). Model misspecification was tested using the Ramsey regression specification error test (Ramsey 1969).

Descriptive results

From 2017/18 to 2022/23, skier visits in La Grave were highly variable as the seasons progressed, and they did not follow a clear trend. They were similar for the winters of 2017/18 and 2018/19, with 35,287 skiers during the 111 open days in 2017/18 and 35,786 skiers during the 119 open days in 2018/19 ( Figure S1 (Yoshizawa_MRD2025_Supplemental_material_FigS1-S3.pdf), Supplemental material,  https://doi.org/10.1659/mrd.2023.00032.S1). The winter of 2020/21 had the lowest numbers, 25,071 skiers for 75 days, because of the COVID-19 pandemic, which shortened the season. The post-COVID seasons of 2021/22 and 2022/23 were similar, with 28,770 skiers during the 123 open days in 2021/22 and 26,905 skiers during the 122 open days in 2022/23. The number of skier visits remained fairly low (around 27,800) compared with the pre-COVID winters (around 34,500). On average, skier visits were higher in February and March than at the beginning or end of the season ( Figure S2 (Yoshizawa_MRD2025_Supplemental_material_FigS1-S3.pdf), Supplemental material,  https://doi.org/10.1659/mrd.2023.00032.S1) and higher on weekends and Fridays than during the rest of the week ( Figure S3 (Yoshizawa_MRD2025_Supplemental_material_FigS1-S3.pdf), Supplemental material,  https://doi.org/10.1659/mrd.2023.00032.S1).

Table 1 shows the distribution of fresh snow days on which the cableway was open. On 63.8% of the days, there was no fresh snow. There was rain on 5.3% of the days, and large amounts of snow (≥21 cm) occurred on only 1.5% of the days.

TABLE 1

Distribution of fresh snow/precipitation categories across all days in the study period on which the cableway was open.

Figure 4 shows skier visits based on different amounts of fresh snow. The visits tended to increase with more fresh snow, with the highest number of visits occurring on days with 21 cm or more. Rainy days had the lowest number of visits.

FIGURE 4

Violin plot of skier visits according to amount of fresh snow. The width of each violin represents data density, while the central box plot indicates the interquartile range and median.

Figure 5 illustrates skier visits based on avalanche risk levels. A notable difference was observed in the distribution of days when La Grave was open across different risk levels: 31 days (5.7%) were classified as risk level 1, 274 days (50.3%) as risk level 2, 216 days (39.6%) as risk level 3, 23 days (4.2%) as risk level 4, and 1 day (0.2%) as risk level 5. Although the average number of visits changed little across risk levels, the variability in visits increased significantly at risk levels 3 and 4.

FIGURE 5

Violin plot of skier visits according to avalanche risk. The width of each violin represents data density, while the central box plot indicates the interquartile range and median.

Empirical results

This section presents the model analysis results. Table 2 shows the estimated coefficients of each independent variable's effect on off-piste skiing demand. The lagged dependent variable (skier visits on the previous day) was statistically significant, with a coefficient of 0.28. It indicates that skier visits on one day moderately influence the number of skier visits on the next day. Because the other variables were categorical, we interpreted the estimates of each variable based on the base level defined for each variable. We observed that the amount of fresh snow positively and significantly affected the number of skier visits compared with days without fresh snow. The intensity of this effect increased with the amount of fresh snow. For example, a day with 11 to 20 cm of fresh snow increased the number of skier visits the following day by 54.5%, compared with a day with no fresh snow, with all other factors being equal. In the case of 21 cm or more of fresh snow, the number of skier visits increased by 121.5%. A day at risk level 2, 3, or 4 was not significantly different from one at risk level 1. These results confirm hypothesis H1 that off-piste skiing demand depends on the availability and amount of fresh powder snow. However, our results oppose hypothesis H2, because high avalanche risk did not affect off-piste skiing demand.

TABLE 2

Regression analysis results for factors influencing off-piste skiing.

For the weather control variables, we observed that sunny days had a more significant and positive effect on skier visits than cloudy days. In contrast, snowy and rainy days affected the number of skier visits negatively compared with cloudy days. These negative effects were more pronounced on rainy than on snowy days.

Regarding the periods tested in our model, although February and March had significantly more skier visits than December, weekends and holidays had significantly more skier visits than nonspecific periods. This relationship is significant and negative for spring holidays, but not significant for winter and Christmas vacations. On average, skier visits have fallen by 13.6% since COVID-19.

The adjusted R² value indicates that our model explains 44.39% of the variance in skier visits to La Grave. Finally, because we used robust standard errors, autocorrelation and heteroscedasticity are not of concern. Therefore, we only conducted diagnostics for a unit root in the dependent variable and for misspecification of the model. Table 3 presents the results of the unit root and misspecification tests. The presence of a unit root in the dependent variable was excluded by the rejection of the null hypothesis of the augmented Dickey–Fuller test, and the misspecification test cannot reject the null hypothesis of missing variables in the model. The results of these tests indicate the reliability of the estimated results.

TABLE 3

Results of diagnostic tests.

Off-piste skiing demand patterns

The results indicated that an increase in fresh snow at La Grave led to an increase in skier visits. This is similar to the case of Gunstock Mountain Resort in Gilford, NH, USA, described in Hamilton et al (2007), where snowfall had a significant effect on alpine skier attendance. In contrast, it differs from the case of Cannon Mountain Ski Area in Franconia, NH, USA, in the same study, where snowfall did not consistently show a significant effect, as well as the results of 2 ski resorts in Michigan, USA (Shih et al 2009). In La Grave, where off-piste skiing is a key tourism resource (Corneloup 2009; Poulain and Garcia 2024), the skiers seek fresh, deep snow after snowfall. Consequently, the skiers exhibited heightened responsiveness to the prevailing amount of snowfall. Other results were similar to those of other ski resorts: skier visits were higher on sunny days, and skiers visited the previous day (Malasevska et al 2015; Falk et al 2022).

Contrary to hypothesis H2, we determined that avalanche risk did not affect skier visits. Hallandvik et al (2017) showed that users' comprehension of avalanche risk information varies based on their experience level. Engeset et al (2018) showed that comprehension of information can vary depending on which types of avalanche problems are present. We suggest that the choice to visit La Grave for off-piste skiing, even if the avalanche risk is high, is a decision made by the individual and involves a more detailed risk assessment on the day. In addition, previous familiarity with slopes, the presence of ski tracks, and other skiers on the slopes increase skiers' propensity to enter slopes exposed to avalanche risks (McCammon 2004; Furman et al 2010; Marengo et al 2017). La Grave, being an area specializing in off-piste skiing, typically has these factors. Given this specificity of La Grave, we should not generally assume that higher avalanche risks have no impact on off-piste skiing demand. La Grave's unique characteristics may explain this counterintuitive off-piste skiing demand.

Christmas and winter vacation periods significantly affected ski tourism demand in previous studies (Malasevska et al 2015; Domaines Skiables de France 2023), but this was not the case in La Grave. Moreover, spring vacations negatively affected the skier visits. Conversely, weekends and holidays had a significant positive impact on skier visits. According to our interview with the cableway company manager, La Grave's clientele comprises a significant proportion of local skiers and those residing within a 2-hour radius, such as in Grenoble and Briançon, whereas other nearby resorts depend on vacationers. Bausch et al (2021) suggested a clear impact of weather on the number and duration of short-term trips to alpine destinations, whereas no clear indication existed about the weather playing a significant role in mid- or long-term trips. We believe that the proximity of skiers' residences probably accounted for their heightened sensitivity to the prevailing conditions.

Post-COVID negatively affected skier visits in La Grave. This was a different trend from the evolution of skier visits in French ski areas increasing as a whole post-COVID from pre-COVID levels (Domaines Skiables de France 2023). What caused this difference remains uncertain. Post-COVID inflation in France (Plane and Vermersch 2024), might have affected the behavior of La Grave's specific clientele. There are some studies showing changes of tourist flow during and after COVID-19 (Shin et al 2022; Yu et al 2023); however, their findings were derived from macroscale analyses. They could not provide insights into niche tourism for off-piste skiing. Thus, we should consider this in a more localized context. In France, during the closure of ski areas because of the COVID-19 crisis, ski touring, in which skiers climb without ski lifts, gained popularity (Cazé 2021). We suggest that the decrease in skier visits in La Grave resulted from competition between ski touring and lift-accessed off-piste skiing. Even after the restrictions were lifted, skier visits in La Grave decreased—likely because many skiers have continued the habit of ski touring rather than off-piste skiing.

Given the limited data availability, only 5 seasons were analyzed. Further analysis with a larger dataset would be desirable. We used weather forecast data because of the unavailability of observational data. This may have recorded the frequency of skiers' attendance more sensitively than an analysis based on observational data.

Impact of climate change and adaptation strategies in La Grave

We considered the likely impact of climate change on off-piste skiing in La Grave. On the demand side, based on our results regarding the importance of fresh snow at the 1800-masl elevation, we suggest that climate change will negatively affect skier visits to La Grave in the future. On the supply side, recent analyses of ski businesses facing climate change have emphasized the adaptive importance of snow management practices, such as grooming and snowmaking (Spandre et al 2019; Berard-Chenu et al 2023). However, snow management practices cannot support off-piste skiing, because this activity seeks fresh, natural snow on unmanaged slopes. Therefore, the challenge of adapting to climate change is even more pronounced for off-piste skiing than for alpine skiing, because off-piste skiing highly depends on natural snow conditions. This is an important insight for the ski area management to consider. Moreover, extreme snowfall, increased by climate change at high elevations (Le Roux et al 2021), can complicate risk management in La Grave. It could heighten the frequent disruption of cableway operations. Hence, we highlight the future vulnerability of La Grave, which relies on off-piste skiing as a major tourist resource.

Facing these situations, a new exploitation plan, replacing the T-bar lift on the glacier with a cableway, is being implemented by the cableway company (Poulain and Garcia 2024). It will enable tourists to reach an elevation of 3600 masl instead of 3200 masl all year, along with the experience of riding the cableway, which will allow them to observe the Girose glacier from above. The project will promote tourism beyond winter, repositioning La Grave as a glacier tourist site (Salim et al 2021). This tourism diversification strategy will strengthen tourism resources in seasons other than winter (George et al 2019). Moreover, the new cableway will be powered by electricity and therefore will reduce carbon dioxide emissions, because the T-bar lift consumes 5000–6000 L of fuel per season (SATG 2020). From the perspective of the cableway company, the development of the glacier zone and its use beyond the winter season will allow some adaptation to climate change.

However, a group of residents called La Grave Autrement has opposed the cableway company's project since 2019 (Poulain and Garcia 2024). The group has expressed its opposition to the new construction plan on the glacier zone. They have criticized the more tourism-dependent future model, which is contrary to La Grave's tradition of maintaining minimal development since the cableway opening. The project could also enhance the destination's international and domestic attractiveness, potentially increasing its carbon footprint because of tourist travel, although not directly emitted by the cableway company. As an alternative, the group proposes removing the existing T-bar lift and constructing a mountain hut at 3200 masl to keep the high mountain environment above this elevation undeveloped (Plas and Virilli 2022). They advocate for the glacier zone to be used for nonfacilitated recreational or educational activities, such as hiking, alpine climbing, ski touring, and scientific tourism.

Our analysis was limited to demand for off-piste skiing in winter, making it difficult to suggest constructive alternatives. The issue of the cableway project represents the 2 climate change adaptation strategies for La Grave, both of which lower dependence on off-piste skiing. This decision also represents a choice for La Grave about the extent to which it will rely on tourism in the future.