Study sites

Study sites included all 14 ski resorts in Utah (Figure 1). Three of these resorts currently do not have the capacity to make snow, while 11 do. Base elevations range from 1646–2926 m with peak elevations ranging from 1951–3353 m. The total skiable lift-served area ranges from 0.49–29.54 km². Specific characteristics of each resort can be found in Table 1.

FIGURE 1

Locations of the 14 ski resorts across Utah, United States. The 2 boxes represent areas used to explore temperature projections to 2100 (classified as northern and southern Utah).

TABLE 1

Characteristics of each resort in Utah (ordered by region and elevation at the main lodge).

Historical data and climate projections

As the ski season is different for each resort, we acquired the opening and closing dates for each resort over 5 recent seasons (2014–2015 through 2018–2019) from resort websites, social media pages, local news outlets, and skicentral.com. This enabled us to analyze weather data within the dates that encompass a recent typical ski season at each resort. After identifying opening and closing dates for each resort in Utah over 5 recent seasons, we defined the season for each resort as the earliest opening date and the latest closing date during this period. We chose the longest recent season to represent the season length, given that historical seasons may have been longer. We included an additional 2 weeks on both ends of the season to account for the variability in operations and to account for conditions leading up to the opening of the resort each year.

To assess recent historical temperatures and snowmaking opportunities, we downloaded daily historical maximum and minimum temperatures at each resort from 1980–2019 from Daymet version 3, using the R package daymetr (Hufkens et al 2018; Thornton et al 2018). Daymet has temperature data on a 1 km grid, and we used the grid cell that contains the main lodge to represent each resort. We aggregated daily data at the seasonal level for each resort, only using days within each resort's season (as defined above). For each resort, we used daily minimum temperatures to find the proportion of the early season with a daily minimum temperature at or below –5°C, indicating favorable snowmaking conditions (Scott et al 2006, 2008). Here, we defined the early season as 2 weeks before opening through January 2. This time period contains holidays (Thanksgiving, Christmas, New Year's) that tend to see high visitation, making snowmaking critical during this period. We ran Mann–Kendall trend tests with Sen's slope on these data to identify trends in temperature from 1980–2018 (Sen 1968). We tested for statistical significance at the α ≤ 0.05 level.

To assess snow quality, we used snow water equivalent (SWE) and snow depth data from Snow Telemetry (SNOTEL) sites across the state to explore whether freshly fallen snow density has already been altered as a result of climate change (USDA NRCS nd). These data are available for individual SNOTEL stations across Utah; time periods for each sensor vary, but the data range from 1999–2018. We used the 7 stations across the state that had data for this entire period of record (station names: Tony Grove Lake, Brighton, Louis Meadow, Ben Lomond Peak, Timpanogos Divide, Midway Valley, and Big Flat). As there were only 7 stations across the state with data for this period, we analyzed the data at the state level rather than at individual resorts. All measurable snow events were identified for each selected station. We measured snow events by subtracting the start-of-day SWE values from the previous start-of-day SWE. For days with recorded changes in SWE and snow depth, the density of the snow was estimated using Equation 1.

SNOTEL data reports SWE rounded to 0.254 cm (0.1 inch) and snow depth rounded to 2.54 cm (1 inch); this rounding limits precision, but is consistent over time. This density analysis was done only for freshly fallen snow at SNOTEL monitoring sites.

We used the probability of exceedance (Equation 2) to find the probability that a snow event was equal or greater to a specific density for each winter season and compared density probabilities from the 1999–2000 through 2018–2019 winters. We ranked each snow event for each selected station. The data between the 10th and 90th percentile (probability of exceedance between 0.10 and 0.90) were used for the analysis to accommodate noise due to outliers or measurement errors.

For future climate projections, we used data for Utah from 2021–2100 to explore how winter (December–March) temperature is expected to change through the 2099–2100 winter season under climate change scenarios RCP 2.6, RCP 4.5, and RCP 8.5 of the Intergovernmental Panel on Climate Change (IPCC) (van Vuuren et al 2011). RCP 2.6 represents a mitigation scenario that aims to keep warming below 2°C globally, RCP 4.5 is an intermediate scenario, and RCP 8.5 assumes high greenhouse gas emissions with little to no effort to reduce emissions (IPCC 2014). Monthly projections for the RCP scenarios between 2021–2100 were acquired using the EC-Earth general circulation model for boundary conditions (Hazeleger et al 2012) and the Rossby Centre Regional Climate ensemble model (RCA4) (Samuelsson et al 2011) through NA-CORDEX (Mearns et al 2017). This analysis is at a larger spatial scale, rather than at individual resorts, because climate projections do not currently have fine enough spatial resolutions to generate resort-specific projections. The spatial resolution of projection data is 50 km, and Figure 1 shows the 2 areas (northern and southern Utah) we used to calculate climate projections. We analyzed future projections for minimum temperature, maximum temperature, and precipitation.

We ran Mann–Kendall trend tests with Sen's slope on the winter temperature projection data for each scenario to understand the projected temperature anomalies by the 2099–2100 winter season. All analyses and visualizations were performed in R and the data and code are publicly available (Akbar et al 2021).

Key informant interviews

We conducted semistructured key informant interviews with Utah ski resort managers or senior employees (Ayres 2008). We used a mixed-methods sampling technique that included criterion sampling (ie contacting people who met our criteria for a year-round management position at any Utah ski resort) and snowball sampling to reach other people identified by interviewees as possibly helpful (Palinkas et al 2015). We contacted managers at all 14 resorts; representatives from 2 resorts declined to participate, and 4 did not respond to emails or phone calls. We interviewed 2 individuals at 1 resort and 1 individual at 7 resorts (n = 8 resorts; n = 9 interviews). These represent 1 (out of 2) resort in the southern part of the state, and 7 (out of 12) resorts in the northern part of the state.

The full interview script can be found in  Appendix S1 (mred-41-03-05_s01.pdf) (Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). Interview questions focused on background information, general questions relating to the impact of climate change, adaptation measures, barriers to adaptation, and how various climate-related scenarios would impact the resort. Most of the questions were open-ended, but 3 questions had Likert-type response scales (ie ordered categories, such as “not a barrier” to “extreme barrier”). For the questions on adaptation perceptions, some question wording was adapted from Wolfsegger et al (2008). All participants were given a copy of the questions at least 48 hours before the interview. Interviews took between 20 minutes–1 hour and were recorded if participants agreed; recorded interviews were transcribed. The Institutional Review Board at Utah State University approved this study under protocol #9773, and interviews were conducted in 2019.

Responses to open-ended questions were coded using semantic coding and coding categories (Braun and Clarke 2006; Lune and Berg 2016). We conducted a top-down thematic structural coding analysis approach in 2 rounds (Gorden 1992). Codes are defined as “tags or labels for assigning units of meaning to the descriptive or inferential information compiled during a study” (Miles and Huberman 1994: 56). Coding categories, including major and minor themes, can be found in  Appendix S2 (mred-41-03-05_s01.pdf) (Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). Interviews were coded independently by 2 authors and compared for accuracy (Lune and Berg 2016). Resulting themes and patterns were summarized and explored in the context of the literature and research questions (Attride-Stirling 2001).

Past temperature trends

Analysis of historical weather data indicates the minimum daily temperatures by ski resort season increased from 1980–2018 for all 14 resorts (Figure 2). These trends are statistically significant for all resorts at α = 0.05, with N = 39 seasons (for full statistical results by resort, see  Appendix S3 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). Sen's slope for these trends ranges from 0.067–0.172; these values can be interpreted as the expected annual increase (°C) in the mean minimum daily temperature over this time period during the ski season. A slope of 0.067 suggests a 2.6°C increase in minimum temperature from 1980–2018, while a slope of 0.172 suggests an increase of 6.7°C. Mean minimum daily temperature by ski resort season from 1980–2018 has been increasing faster than maximum temperature ( Appendix S4 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). The increase in mean maximum daily temperature by season is statistically significant for 6 of the 14 resorts, with Sen's slope ranging from 0.039–0.106 for those that are significant (see  Appendix S3 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1).

FIGURE 2

Trends in the mean minimum daily temperature (°C) by ski resort season (1980–2018). Numbers on each panel represent Sen's slope; * indicates the value is statistically significant at α ≤ 0.05.

The proportion of most resorts' early/holiday season with a minimum daily temperature at or below –5°C has decreased steadily, with annual variation (Figure 3). These trends are statistically significant at 12 of the 14 resorts, with coefficients ranging from –0.003 to –0.008 (see  Appendix S5 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). The 2 resorts that did not have significant trends were the 2 farthest-north resorts (Cherry Peak and Beaver Mountain). A coefficient of –0.005 indicates the proportion of the early season with daily temperatures at or below –5°C has decreased by 0.005 each season, or by 0.195 from 1980–2018. For example, if a resort had a 0.900 proportion of the early/holiday season with daily temperatures at or below –5°C during the 1980 season (eg Deer Valley), an average annual decline of 0.005 would bring this proportion to 0.705 by the 2018 season. Assuming an early/holiday season of 44 days, this would have dropped the snowmaking days from 40 to 31 between 1980–2018. Trends in the reduction in the proportion of days with minimum daily temperature at or below –5°C are even greater when considering the full ski season and not just the critical early/holiday period (see  Appendix S6 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1).

FIGURE 3

Trends in the proportion of days in the early season with a minimum daily temperature below –5°C by ski resort (1980–2018). The early season is defined as 2 weeks before opening (varies by resort) through January 2, to capture the holiday season. Numbers on each panel represent Sen's slope; * indicates the value is statistically significant at α ≤ 0.05.

Ski resort manager perceptions

The interviews with managers revealed varying perceptions of what constitutes a “successful season,” with visitation (maintaining current numbers and increasing visitors), good-quality snow, and season pass sales all mentioned frequently. Respondents indicated that snow quality and snow quantity play a key role in a successful ski season. Resort managers reported their resort would need to be open 97–120 days each season to remain viable.

Several managers knew that increasing temperatures and shifts in snowfall patterns would affect the length of their ski season. One resort manager said:

A shift in the ski season and negative impacts were mentioned by another manager, who noted:

Another resort manager intimated that although high temperatures and low snowfall have happened in the past: “I would say they're becoming more frequent as far as later opening dates . . . I mean, you name it, warmer temperatures, less precipitation, we're seeing them more frequently.”

Adaptation measures: We asked the resort managers to rank various adaptation measures on a scale with response options ranging from very inappropriate to very appropriate. Specific measures and their responses are in Table 2. Most resort managers indicated they have already diversified winter and all-season activities at their resort in some capacity. Seven of the 8 Utah ski resorts represented in our interviews have multiple winter activities besides skiing and snowboarding, such as snow tubing, snowshoeing, educational programming, and cross-country skiing. Activities in other seasons range from sporting activities, such as mountain biking (sometimes extensive trail systems), obstacle courses, horseback riding, and disk golfing, to events such as conferences and festivals and complementary services such as spas, restaurants, and rentals. These offerings are diversified to adapt to climate change and to help expand revenue sources alike.

TABLE 2

Responses from resort managers about adaptation measures at their resort. n = 8 resorts.

Seven of the resort managers we interviewed have snowmaking at their resorts, either for most of their terrain or just to supplement critical areas. Resorts either have storage ponds for water or purchase water for snowmaking. Methods to increase snowmaking capacity are common at many resorts, as a resort manager noted that it has “been a major push in the last 10 or so years.” Various methods were noted to increase snowmaking capacity at resorts. One resort manager said the resort used chemical additives, while another said the resort might consider using them in the future; others are against it. Four resort managers spoke about alternative ways to increase snowmaking efficiency at their resort, such as cooling water storage ponds before snowmaking and investing in snow gun technology. One resort manager indicated the resort did not have snowmaking capacity and did not foresee a necessity for it in the future.

Resort managers reported a variety of issues related to slope maintenance and snow coverage. Avoiding slopes that require too much snow is a consideration for many resorts; however, these areas are usually just closed for longer periods or, if they are essential, given more snowmaking capacity. Three resort managers stated that in the future they may have to close slopes for longer periods of time. Slope maintenance included clearing brush and obstacles so less snow was needed for a safe level of coverage. One resort manager reported the resort “also implemented mowing the brush, that way you can open up with minimal snow.” Another manager discussed contouring the slopes so the coverage needed is equal across the skiing area.

Most resorts are as high in elevation as they can feasibly go, with lease agreements often limiting moving to higher elevations, but a few have plans to move higher. Two resort managers mentioned having higher base lodges, giving them an advantage in the future.

Barriers to adaptation: We asked the resort managers to rate potential barriers to climate change adaptations at their resort on a scale with response options ranging from not a barrier to extreme barrier. Specific barriers and managers' responses are visualized in Table 3. Resort managers face moderate to extreme barriers in financial costs to adaptation. Prioritizing the need to receive a return on investment combined with large overhead costs limits the ability of ski resorts to invest in additional adaptation methods. One manager noted, “It can be an extreme barrier because you see the potential of being able to do and adapt all these different things at once, but then you have to prioritize and it takes time.” Some resorts also face additional challenges based on their size and lack of corporate backers, which limits their financial abilities to adapt.

TABLE 3

Responses from resort managers about barriers to adaptation at their resort. n = 8 resorts.

At resorts that have snowmaking capability, water availability and temperature required to make snow were noted as major barriers to snowmaking. Some resorts own artificial water storage ponds; others receive a certain allotment from public utilities, setting a limitation on snowmaking ability. Resorts that receive water from public utilities receive a reduced price because it is untreated water from natural creeks. Using water for snowmaking is considered a nonconsumptive use, although sublimation losses can be considerable (4–41%) (Reba et al 2012). Snowmaking also requires snowmaking guns, plumbing, increased electricity, and labor.

Possible scenarios: Resort managers were asked to rate the impact of possible future scenarios on a scale with response options ranging from no impact to extreme impact (Table 4). In Utah, the 2017–2018 season was a particularly warm and low-snowfall year; we specifically asked about this year given its recency.

TABLE 4

Responses from resort managers about impacts of different future conditions on their resort. n = 8 resorts.

Resorts managers had varying responses when asked about a potential decrease in snowpack. Two thought a decrease would have slight or no effect because of their annual snowfall being extremely high, with some resorts reporting over 1200 cm annually. One manager responded, “[other resort's] average is 230 inches [585 cm] . . . but our average is 500 inches [1270 cm] plus and at 20% decrease [in snowpack] we still have got decent coverage. So, it's a different scenario.” Some resort managers thought less snow would not affect their resort if they still had the temperature to make snow. Increases in temperatures to a level above where snowmaking is efficient would be more of an issue with one manager, who stated, “Sometimes along with less snowfall comes warmer temperatures, which doesn't allow you to make snow.” Less snowfall at the end of the season is less of a concern than snowfall at the beginning of the season when the base is being built at the resort. According to one resort manager, short periods with less snow may not have much of an impact, but longer stretches of no snow could have a much larger impact. One manager noted: “A couple of seasons ago when a lot of ski areas couldn't even open because there wasn't enough snow, especially in the Northwest and the Midwest and parts of the East and stuff, we ended up getting higher skier visits.” The manager explained that, in contrast to nearby ski resorts, the manager's resort drew visitors, as it could open many areas of the resort because of its snowmaking and grooming capabilities.

Since warmer winters with less snow are likely to become more common, we asked resort managers what the effect would be if seasons such as 2017–2018 became more common. Two managers thought there would only be a slight effect, while 5 others stated it would have an extreme effect on their resorts. However, one manager of a resort with extensive snowmaking abilities mentioned that the resort benefited that winter because it drew from other resorts that had no snow.

Snow quality and density

When asked about high-quality snow, 6 managers replied that quality snow is the cold, dry powder for which Utah is renowned; one replied “cold and copious.” One resort manager responded that quality snow depends on the skier: “Beginners may prefer different snow than a ski racer or a powder hound.” Two resort managers stated that unlike before, the focus was on coverage and long-lasting snow instead of cold, copious powder. One manager noted, “It's now kind of like, well we've got enough snow to cover the rocks and the trees and stumps and it may be icy, but at least there's enough coverage. So that's kind of the direction now that we're looking at.” With coverage, more of the mountain can be open and exposure of rocks and other obstacles becomes less of a risk; the season can continue for a longer period.

The density of snowfall events in Utah from 1999–2017 seems to have increased slightly (Figure 4), possibly indicating less of the cold, dry powder in recent years. However, this trend is not statistically significant for any of the percentiles shown. The trend of slightly denser snow events in the last decade is consistent across all 7 SNOTEL stations (see  Appendix S7 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). Overall, resort managers felt that seasons with less high-quality snow would reduce the number of visitors and decrease user experience. A potential implication of lower-quality snow mentioned by resort managers was that fewer skiers may purchase season passes, instead buying day tickets to seek the best conditions. This approach may make resorts with high grooming/snowmaking capacity more attractive.

FIGURE 4

Snowfall event density in Utah from 1999–2017, by percentile. The 25th percentile indicates that 25% of the snowfall events for that year are less dense (below the point), while 75% are denser (above the point). Lines display linear trends; the trends are not statistically significant at the α ≤ 0.05 level.

Climate projections

The majority of resort managers indicated an increase of 3°C would have moderate to extreme effects at their resort. Though large variability exists in the year-to-year predicted values, temperatures during winter (December–March) in Utah are likely to increase by the end of the century. Figure 5 shows the increase in minimum temperature from the 2021–2022 to the 2099–2100 winter seasons. Under the RCP 2.6 scenario, there is not a statistically significant change in winter temperature from 2021–2099 in either northern or southern Utah (Mann–Kendall trend test results in  Appendix S8 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). Under RCP 4.5, the minimum temperature in northern Utah is projected to increase by 2.0°C in the winter from 2021–2099 (0.026°C per year), while southern Utah is expected to warm by 2.3°C (0.030°C per year). Under the RCP 8.5 scenario, winter temperatures for northern Utah are projected to increase by about 6.0°C by the end of the century (0.076°C per year), and 6.6°C in southern Utah (0.084°C per year). Given there is always uncertainty in climate projections, confidence intervals for all Sen's slopes can be found in  Appendix S8 (mred-41-03-05_s01.pdf) (Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1).

FIGURE 5

Projections for average daily winter minimum temperature for December–March, by winter season, for northern and southern Utah. Background lines represent the projections for each winter season, whereas the straight lines represent linear trends from 2021–2099.

Similar trends are projected for maximum temperatures in the winter (see  Appendixes S8 (mred-41-03-05_s01.pdf),  S9 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). Specifically, there is not a significant change in maximum temperature from 2021–2099 under RCP 2.6. Under the RCP 4.5 scenario, both northern and southern Utah are expected to warm 2.4°C by the end of the century. Under the RCP 8.5 scenario, projections show an increase in maximum temperatures of 7.3°C in northern Utah and 6.1°C in southern Utah.

It is important to note that NA-CORDEX projection data are colder than Daymet data from 2006–2018 winters, indicating actual future temperature projections may be higher (projections are about 0.9–3.0°C colder for minimum temperature and 5.1–6.5°C colder for maximum temperature; see  Appendix S10 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). This does not affect the change-over-time projections (as described above), but does affect the raw projected values.

Precipitation projections indicate no substantial and consistent change in total winter (December–March) precipitation from 2021–2099 (see  Appendix S11 (mred-41-03-05_s01.pdf), Supplemental material,  https://doi.org/10.1659/MRD-JOURNAL-D-20-00065.1.S1). There is high variability in seasonal precipitation projections. However, with winter temperatures rising, we would expect that in the future, more precipitation will fall as rain, and less precipitation will fall as snow.

Limitations and future research

We interviewed managers at 8 Utah resorts; the data from interviews may not be representative of all ski resorts within the state. Resort managers who agreed to participate may have been more likely to believe that climate change was real and impacted their resort. Additionally, since we were able to interview only 1 individual at most resorts, their views are not necessarily representative of the entire resort. Future research could aim to interview more employees at each resort and see how perceptions differ based on job position.

Many resorts in Utah focused on snowmaking as an important adaptation strategy. Additional research is needed to investigate future water availability, use, and rights across Utah for snowmaking. For analysis of historical water use, snowpack, and hydrological changes in Utah, see Khatri and Strong (2020). Additionally, more research on projected changes to snowfall and snowpack at Utah resorts is needed to better understand future water and snowmaking needs (eg Lackner et al 2021). Future studies could also examine how the characteristics of a resort (eg size, location, visitor profile) relate to the ability of a resort to adapt and their perceptions of barriers to adaptation. The climate and physiography of Utah are similar to those of other states in the western United States; thus, similar trends may be expected in nearby states. However, additional research is required to explore whether similar patterns would be observed in the other locations.