Ongoing and projected climate change is accelerating the warming of the cryosphere throughout Earth's mountain regions (Huss et al., 2017). Reductions in winter season snow, ice, and permafrost cover and volume primarily result from rising air temperatures (Brown and Mote, 2009) and shifts in precipitation from snow to rain (McCabe et al., 2018). These changes have cascading effects from mountains to lowlands with wide-ranging socioeconomic and ecologic impacts (Huss et al., 2017). In mountain regions of the United States, Europe, and Canada, winter recreation and tourism are central to economic activity. The economic benefits from winter recreation are projected to decline as a result of continued climate change that reduces season length and makes access to reliable snow more difficult (McBoyle et al., 2007; Scott et al., 2008; Wobus et al., 2017; Steiger et al., 2017).

Most winter tourism-based climate change impact studies have focused on ski-resort-related activity (Steiger et al., 2017), although research has begun to address how other recreation-based components of the winter economy may be affected (McBoyle et al., 2007; Scott et al., 2008; Tercek and Rodman, 2016; Wobus et al., 2017, Hagenstad et al. 2018). Skier visits are positively correlated to snowfall (Hagenstad et al., 2018) and we assume that such a correlation is consistent across winter recreation activities. Due to the dependence on natural snowfall and reduced adaptive capacity compared to the ski community, which can use cost-effective snowmaking to augment the natural snowpack, oversnow vehicle (OSV) recreation is highly vulnerable to climate variability and change (McBoyle et al., 2007; Scott et al., 2008). Climate change projections for Canada and the northeastern US under an aggressive greenhouse gas emissions scenario suggest that by the mid-21st century, OSV season lengths will be reduced by 50 %–100 % in most areas (McBoyle et al., 2007; Scott et al., 2008). A survey of the OSV community in Vermont found that reductions in the length of the winter season with sufficient snow coverage for OSV use were observed by 45 % of respondents, with 74 % of respondents decreasing their OSV use in response to low snow conditions (Perry et al., 2018). This survey also found that encounters with other recreationists, including OSV users, detracted from a high-quality recreation experience. The net effects of reduced season length, more congestion, and lower-quality experiences result in lower economic benefits from consumer surplus, or the amount a person is willing to pay over the amount actually spent. For OSVs, consumer surplus is estimated to be approximately USD 61 person-1 day-1 (Hagenstad et al., 2018).

In the Lake Tahoe region of California (Fig. 1a), and many other rural mountain areas of the western US, OSV use is a regionally significant component of winter season recreation. Estimates of annual economic impact from OSV recreation in the United States range between USD 7 and 26 billion (Fassnacht et al., 2018). As a result, OSV recreation has an appreciable economic impact on rural counties within the northern Sierra Nevada, many of which have a greater dependence on tourism-related employment than elsewhere in California (United States Census, 2016).

The proximity of the Lake Tahoe region to large population centers creates demand for OSV recreation over a limited and ecologically sensitive area. In order to limit potential negative impacts on natural resources (e.g., Keddy et al., 1979) during OSV operation, a minimum snow depth must be present. Minimum snow depth restrictions have been proposed by several forests undergoing winter travel management planning across the Sierra Nevada. This restriction is usually proposed as a minimum depth of 30 cm of un-compacted snow (USFS, 2013). Few forests have such a requirement at this time, but several are currently engaging in the process of winter travel management planning in response to a 2015 US federal court ruling (Federal Register, 2015). The Eldorado National Forest in northern California (located in the southwestern quadrant of the study area) currently requires a minimum snow depth of approximately 30 cm for off-trail OSV use.

To our knowledge, no precise value of this minimum depth has been established through comprehensive studies quantifying OSV use and impacts or disturbance. Nonetheless, evidence indicates that OSV use can alter the landscape when a shallow snowpack is present. Keddy et al. (1979) observed that OSV use on very shallow snow (10–20 cm deep) doubled snow density and compressed underlying vegetation. When OSV use began under a deeper snowpack, less difference in snow density and hardness was observed compared to a control (no-OSV use) snowpack (Fassnacht et al., 2018). Further complicating the minimum depth requirement is the dependence of snow depth on the density of the snow, which varies seasonally and as a function of weather conditions that drive snowpack metamorphism processes (Sturm et al., 2010).

Resource managers tasked with day-to-day operations such as opening and closing OSV trailheads over large, diverse areas may not have the resources to visit trailheads to obtain snow depth and density measurements. Instead, they often rely on subjectively based qualitative assessments of what is deemed sufficient snow. Managers often do not set a specific OSV season, leaving it to user discretion to determine when OSV use is appropriate. This can potentially cause conflict with other uses during the start and end to the winter season and can allow opportunities for inadvertent damage to natural resources due to insufficient snow depth. Here, we estimate the median timing of achieving sufficient snow depths for OSV operation and their trends during the past 34 years using observations of snow water equivalent (SWE) and a reasonable assumption of snow density. We focus on the initial timing of sufficient snow depth since the greatest demands for OSV recreation and potential ecological impacts occur between early and middle winter. The proximal causes of the identified increasingly later onset of achieving a minimum SWE value are further investigated. Because the trend towards later onset is not expected to reverse under continued regional warming, we provide adaptation strategies to cope with diminishing early season snowpack resources that can be included in forest travel management plans. The techniques can be extended to other regions where OSV recreation is an important component of economic activity and where early winter snowpack losses may be impacting seasonal recreation.

The study area is the Lake Tahoe region of the western US, a coastal moderate-elevation snow-dominated mountain range (Fig. 1a). Daily maximum and minimum temperature, SWE, and precipitation were acquired for 16 SNOTEL stations from the Natural Resources Conservation Service (https://wcc.sc.egov.usda.gov/nwcc/tabget, last access: 1 June 2018). Daily, gridded estimates of SWE at 100 m horizontal resolution were provided by a satellite-era SWE reanalysis product (Margulis et al., 2015, 2016). The SWE reanalysis utilizes a Bayesian data assimilation framework to condition a priori snow model estimates with Landsat fractional snow-covered area images (Margulis et al., 2015). It verifies the posterior estimates against in situ daily snow pillow and monthly snow course data, which were found to compare favorably to previous studies (Margulis et al., 2016). The limitations of the remote-sensing approach include lower temporal frequency of Landsat passes (approximately every 16 days) and potential obscuration of the land surface by clouds and vegetation, which can reduce usable imagery. Challenges with the in situ verification data include representativeness or the discrepancies resulting from point-based snow pillow versus transect-based snow course SWE measurements, undersampling of forested and sloped terrain, and the bias of sites towards the intermediate elevations of the Sierra Nevada (50 % of the stations are between 1500 and 2500 m; Margulis et al., 2016). The period studied encompasses 1 October 1984 to 31 March 2018 (2016 for the SWE reanalysis), which corresponds to the winter seasons of 1985–2018.

No accepted value of a minimum snow depth exists for OSV operation. Anecdotal values used by managers vary between 15 and 45 cm depending on compaction (USFS, 2013), but these do not take into account variability in snow density. To provide a conservative and reasonable estimate of sufficient snow depth for what is assumed to be required for non-intrusive OSV operation, we specified 90 mm SWE (hereafter SWEmin) as the required minimum SWE corresponding to a minimum uncompacted depth of 30 cm for approval of OSV use. This value was obtained with Eq. (1): SWE[mm]=d[mm]×ρs/ρw, where d is depth, ρs is the density of the snow, and ρw is the density of water. We assume that in a coastal snowpack with marginal compaction, ρs is typically 0.3 g cm-3 (Sturm et al., 2010). This value appears reasonable to approximate a depth of 30 cm for early season conditions and is consistent with values used by the USFS (2013). Our SWEmin value is close to that of Patterson (2016) and Tercek and Rodman (2017), who both chose 100 mm SWE as a threshold value for winter recreation in the Rocky Mountain National Park and Yellowstone National Park, respectively. We report the median timing of when each SNOTEL station and reanalysis grid point achieves SWEmin and the annual timing as the median of the 16 SNOTEL stations.

To explore possible processes controlling the onset date of SWEmin, snow fractions (Sf) between 1 October and 31 January were calculated using the empirical hyperbolic tangent function formula developed by Dai (2008) with Sierra Nevada ecoregion parameter values estimated by Rajagopal and Harpold (2016). In contrast to Rajagopal and Harpold (2016), who used maximum temperature to estimate snow fraction, we selected average temperature because it gave a closer approximation to the median snow level (∼1750 m) based upon independent estimates from observations (Hatchett et al., 2017). Dry days were days when less than the minimum measurable amount of precipitation (2.54 mm) was measured at SNOTEL stations. Mean minimum temperatures on dry days were calculated over the 16 stations for each year, as minimum temperature influences both snowpack dynamics and ecological processes (Oyler et al., 2015).

For all data, linear fits were estimated using a Theil–Sen slope and we report Spearman rank correlations. Statistical significance was tested using a modified Mann–Kendall test that accounts for serial correlation (see Hatchett et al., 2017, and references therein).

Using snow water equivalent and a density assumption as a proxy for depth, we have presented a pilot study aimed at a better understanding of when the Lake Tahoe region attains sufficient snowpack depth to allow safe oversnow vehicle (OSV) usage. A station-based analysis of 16 remote weather stations in the region and a spatially distributed SWE reanalysis product indicated that the median timing of achieving sufficient depth varies with elevation from early November to late December. The median timing of sufficient depth has increased by approximately 2 weeks during the past 3 decades with significant changes on the order of 3 weeks. The proximal causes for this shift towards later onset appear to be due to both a shift from snowfall to rainfall and increases in dry day frequency and temperature during the early winter season. However, further research is needed to estimate specific contributions from each cause and constrain the role of surface-albedo and/or humidity feedbacks at various elevations throughout the region (Patterson, 2016; Walton et al., 2017).

A primary limitation of our study is the lack of an established snow depth to avoid negative impacts of OSV operation as a function of land cover type and snow density. The work of Fassnacht et al. (2018) represents an important advance towards achieving this value, which can be used to guide winter travel management planning, although the US Forest Service has begun to recommend a snow depth (USFS, 2013). Additional studies on achieving regionally relevant minimum snow depths and better quantification of economic and ecological impacts from reduced-snow-cover area and duration will guide more robust travel management plans in national forests. They can also help prioritize pragmatic adaptation strategies for specific regions. Given the economic impact of OSV recreation and likely reduction in land available for OSV or other human-powered recreation uses (McBoyle et al., 2007; Scott et al., 2008; Tercek and Rodman, 2016; Hagenstad et al., 2018), combined with increasing numbers of winter recreation participants (Fassnacht et al., 2018), achieving winter travel management plans that are adaptive to varying snowpack conditions while minimizing user conflicts will be a key step towards sustainable mountain recreation.