Less reliable snow. Warmer seas. Changing swells. Explore what scientists project, and what it could mean for the places and sports we love.
Global warming relative to 1850–1900 · °C
2.1–3.5°C very likely range
An intermediate emissions pathway. The shaded band shows uncertainty within this pathway.
These are 20-year averages. Connecting lines guide the eye; they are not annual forecasts.
| Pathway | 2041–2060 °C | 2081–2100 °C |
|---|---|---|
| SSP1-2.6 | 1.7 (1.3–2.2) | 1.8 (1.3–2.4) |
| SSP2-4.5 | 2 (1.6–2.5) | 2.7 (2.1–3.5) |
| SSP3-7.0 | 2.1 (1.7–2.6) | 3.6 (2.8–4.6) |
Lower and higher emissions are conditional futures, not guaranteed best and worst cases. No probability is assigned to these pathways.
UNEP’s 2025 assessment of current policies over this century. A useful reality check, assessed separately from the IPCC pathways above.
What changes the outcome? ↓Yes, with growing differences between regions and elevations. At +2°C global warming, one study projects very high snow-supply risk at 53% of European resorts without snowmaking, or 27% with snowmaking on half their slopes.
François et al., 2023Explore the future of skiingThere is no global yes or no. Under high emissions, mean offshore wave heights are projected to decrease in the North Atlantic and increase in the Southern Ocean. Wind, swell direction, water depth and coastal change shape the waves at each break.
IPCC SROCC, 2019Explore the future of surfingEmissions cuts change the scale of warming and its impacts. Adaptation can reduce some losses, but cannot replace cold winters or healthy oceans. The pathways on this page describe different assumptions, with no assigned probability of occurring.
IPCC AR6, 2021See what changes the outcomeReading the evidence: observations describe measured changes; projections depend on specified scenarios; regional interpretations explain possible sporting consequences. Each keeps its own baseline.
Put three regions side by side. See where the science draws a difference, then explore what it means on the ground.
Groomed natural snow, across RCP2.6 / 4.5 / 8.5
Regional reliability elevation
2,000 m is within this study’s range.
Higher terrain retains an advantage, but low return runs and valley links can still be exposed.
Read the regional evidenceExplore Morzine’s recent seasonsRegional reliability elevation
2,000 m is below this study’s range.
The same natural-snow criterion is reached higher here than in the northern French Alps.
Read the regional evidenceRegional reliability elevation
2,000 m is below this study’s range.
The model points to pressure on lower terrain; altitude, aspect and management all affect individual slopes.
Read the regional evidenceThe study’s criterion is at least 100 days with 100 kg/m² of snow between 15 December and 15 April in seven winters out of ten. Being above a regional band does not guarantee snow on an individual slope. These are scenario spreads, not uncertainty percentiles. No snowmaking is included here.
Direction of mean significant wave-height change under RCP8.5, a high emissions pathway
Regional context for Hossegor and Newquay
A fall in average offshore wave height can alter swell supply. Storms, local winds and moving sandbanks still shape each session.
Read the surfing evidenceSwell source for exposed southern coasts
Increasing offshore waves can change swell reaching distant coasts. Break orientation and the swell's path determine local exposure.
Read the surfing evidenceRegional offshore projection
More mean offshore wave height does not establish more rideable days. Reef shape, water depth and local wind still matter.
Read the surfing evidenceArrows show the direction of regional offshore change, not its magnitude or a future surf score. Local bathymetry, swell direction, tides, wind and coastal change must be assessed separately. These studies do not establish which beach will have the best surf in 2050.
Higher terrain keeps an advantage, but the height needed for dependable snow changes with place and emissions. Explore eight published Alpine and Pyrenean case studies.
At 1,800 m: your chosen terrain sits below the modelled 2030–2050 reliability threshold in Chablais under RCP4.5. Compare the height of the runs and connecting slopes, as well as the resort village.
+480m
Rise in the modelled reliability elevation by 2080–2100 under RCP4.5, relative to 1986–2005.
What “reliable” means here: at least 100 days with 100 kg/m² of groomed natural snow between 15 December and 15 April, reached in 7 winters out of 10 in the model. No snowmaking. This is a massif-scale benchmark, not a prediction for a particular run or resort’s opening days.
| Pathway | 1986–2005 HIST | 2030–2050 | 2080–2100 |
|---|---|---|---|
| Lower · RCP2.6 | 1,580 m | 1,920 m | 1,880 m |
| Intermediate · RCP4.5 | 1,580 m | 1,860 m | 2,060 m |
| Very high · RCP8.5 | 1,580 m | 1,930 m | 2,640 m |
The reference is the historical climate-model ensemble (HIST), not direct station observations. The 70% criterion describes frequency across modelled winters, not confidence in a projection. RCP pathways are separate experiments from the SSP pathways elsewhere on this page.
Swiss observations show snow arriving later and disappearing earlier. Projections point to fewer snow-covered days, with large differences between places. Explore the duration evidence and the timing evidence separately.
Select a future period below to compare it with the modelled reference.
−108days
49 snow-covered days compared with 157 in the modelled 1999–2012 reference. About 31% of that reference count remains.
A 5 cm snow cover is not a skiable piste. These counts do not include snowmaking or predict resort operating dates.
| Location | 1999–2012 | 2020–2049 | 2045–2074 | 2070–2099 |
|---|---|---|---|---|
| Bern | 40 | 13 | 5 | 0 |
| Interlaken | 41 | 14 | 6 | 0 |
| Grindelwald | 101 | 72 | 37 | 14 |
| Adelboden | 139 | 107 | 75 | 35 |
| Mürren | 173 | 148 | 116 | 77 |
| Chur | 39 | 14 | 5 | 0 |
| Disentis | 98 | 68 | 40 | 18 |
| Davos | 157 | 123 | 86 | 49 |
| Samedan | 165 | 139 | 101 | 61 |
| Weissfluhjoch | 254 | 228 | 197 | 163 |
The chart follows Table S2’s label, “mean number of snow days”. The article’s main text calls these medians; the published numbers are reproduced unchanged.
This 2017 study uses the older SRES A2 scenario, a high-emissions experiment. It is not a current-policy forecast and cannot be relabelled as one of the SSP pathways above. Its reference snow cover is modelled from 1999–2012 weather. Individual years vary; the table does not supply uncertainty intervals.
Average shifts across 11 Swiss Alpine stations between 1,139 and 2,540 m.
Snow-cover duration shortened at every study station, averaging 8.9 days per decade. The later-onset trend was statistically significant at about 30% of stations.
Klein et al. (2016) ↗Approximate shifts from the modelled 1999–2012 reference under SRES A2. Timing refers to the longest continuous cover of at least 1 cm. This catchment result does not change with the location selected above.
Marty et al. (2017), section 3.5 ↗The timing cards use different populations, periods and units: observed days on the left, projected weeks on the right. They are not a single time series. No future opening or closing date has been inferred from a snow-day count.
Published regional projections make the global numbers tangible. Explore the differences that matter on the mountain and at the coast.
Yes, but where, how long and at what cost changes with warming. Europe’s snow-supply risk rises sharply.
For skiing, the pressure falls on dependable snow and the length of the season. Resort closures also depend on local terrain, adaptation and business decisions.
Share of 2,234 European resorts at very high snow-supply risk. Global warming levels relative to pre-industrial conditions, not particular calendar years.
| Warming | No snowmaking | 50% snowmaking |
|---|---|---|
| +2°C | 53% | 27% |
| +4°C | 98% | 71% |
These percentages describe snow risk, not how many resorts will close. Business models, adaptation and local terrain also matter.
Very high snow-supply risk by region · same adaptation selection
| Region at +2°C | No snowmaking | 50% snowmaking |
|---|---|---|
| French Alps | ~33% | 7% |
| Pyrenees | 89% | 9% |
| French / Swiss mid-mountains | 80% | 56% |
The French Alps estimate without snowmaking is approximately one third. Regional averages conceal differences between resorts.
2030–2050 · modelled regional elevations, metres above sea level
Northern French Alps: 1,850–2,000 m
Groomed natural snow meeting a 100-day criterion in seven winters out of ten. Ranges span regional medians across three emissions scenarios, not uncertainty percentiles. They are not universal viability cut-offs.
| Region | Elevation range (m) |
|---|---|
| Northern French Alps | 1850–2000 |
| Southern French Alps | 2500–2650 |
| French Pyrenees | 2250–2300 |
| Spanish / Andorran Pyrenees | 2550–2650 |
Conceptual Alpine bands. Select an elevation to explore the pressures.
A viable core winter can coexist with less dependable early and late snow. Aspect, local climate and slope management matter alongside altitude.
Lower slopes lose snow reliability first. The southern Alps and Pyrenees need more elevation than the northern Alps to meet the same modelled natural-snow criterion. More dependable high terrain does not protect lower links and resort return runs.
Regional direction of mean wave-height change · late-century high emissions, RCP8.5
Illustration of direction only, not a wave-height forecast.
A projected decline in average significant wave height under high emissions. Storms still occur, and season, swell direction and local wind decide whether a session works.
A multi-model study finds regional mean height and period changes of 5–15% and direction shifts of 5–15° under high emissions. These are regional changes of either sign, not a 15% improvement in surfing.
Projected further decline in warm-water coral reefs
For reef breaks: coral decline threatens the living ecosystem and coastal protection. It does not translate directly into the percentage of surf breaks lost.
Global warming relative to pre-industrial levels; reef decline relative to the 2018 assessment’s present-day baseline. The grid represents projected reef decline, not a map of individual reefs.
A regional climate signal passes through local conditions before it reaches your board.
Height, period and direction change the energy reaching a coast.
Local wind shapes the surface. Tide and sea level set the water depth.
Reefs, sandbanks and coastal shape determine how waves break and where you can enter.
For a trip or session, check the local swell, wind, tide and access forecast. Higher average offshore waves alone do not establish better surfing conditions.
Warming and oxygen loss alter marine habitats. Fish populations respond to temperature, fishing pressure and management, so temperature alone cannot forecast a local catch.
Coral loss affects biodiversity and coastal protection. Reef structure and sea-level rise can change how waves break, alongside impacts on diving and local livelihoods.
Rainfall, runoff and wastewater infrastructure affect bathing water. A warmer sea is not a direct measure of sewage or bacterial pollution; check local water-quality evidence before entering.
Wave quality is only part of the picture. Higher water levels and longer spells of unusual heat change coastlines and the life beneath the surface.
Global mean rise in centimetres · relative to 1995–2014
56 cm median rise by 2100
SSP2-4.5: likely range 44–76 cm. Bars show ranges; dark marks show medians. Regional sea level can differ substantially.
| Pathway | Median (cm) | Likely range (cm) |
|---|---|---|
| SSP1-2.6 | 44 | 32–62 |
| SSP2-4.5 | 56 | 44–76 |
| SSP3-7.0 | 68 | 55–90 |
For surf and coastal access: a higher baseline changes water depth over reefs and sandbanks and raises flooding exposure. It does not mean every tidal range grows.
Global frequency in 2081–2100 · multiplier versus 1995–2014
2–9 times as frequent
Reported projection range. This comparison uses SSP5-8.5, a very high emissions pathway, rather than the SSP3-7.0 pathway in the global chart.
| Pathway | Frequency multiplier |
|---|---|
| Lower · SSP1-2.6 | 2–9× |
| Very high · SSP5-8.5 | 4–18× |
For diving, swimming and marine life: warmer averages also bring damaging heat extremes. Habitat loss and shifting species can change familiar reefs and fisheries.
Global mean sea-surface warming by 2081–2100 · change relative to 1995–2014, in °C
+1.51°C model mean · SSP2-4.5
Bars span the 5–95% model range; dark marks show means. This is additional warming from a recent baseline, rather than the pre-industrial baseline in the opening chart. Local sea temperatures can change differently.
| Pathway | Mean warming (°C) | 5–95% range (°C) |
|---|---|---|
| SSP1-2.6 | 0.86 | 0.43–1.47 |
| SSP2-4.5 | 1.51 | 1.02–2.19 |
| SSP3-7.0 | 2.19 | 1.56–3.3 |
Explore the processes that connect a warming planet to the snow, seas and ecosystems around us.
Permafrost is ground that stays frozen for at least two consecutive years. In carbon-rich Arctic soils it locks away large stores of organic material.
They matter, but they are different systems. Changes in atmospheric circulation can shift regional weather; a “jet stream breakdown” is not an established local forecast. Atlantic overturning is expected to weaken, while the timing and risk of a collapse remain uncertain. Neither supports a guaranteed future ski season for a particular resort.
Explore regional implications beside real recent history. The projections above come from climate studies; the charts below come from PeakWave’s stored conditions archive.
Lower return runs face greater pressure than higher terrain. The regional study places the 2030–2050 groomed natural-snow reliability elevation at 1,850–2,000 m; individual slopes depend on aspect, weather and management.
Open local conditions, season heatmap and evidence →Loading stored season history…
Archive coverage is currently up to two years for these weather charts and one rolling year for the score heatmap. Four or five recent seasons would still be too short to isolate climate change from natural variability. We show available coverage rather than filling gaps with invented data.
The future is constrained by physics, but it is still affected by choices. Cutting emissions limits the scale of losses; adaptation helps communities manage the changes already under way.
Annual global greenhouse-gas emissions in 2035 versus 2019, for the selected pathway in UNEP’s 2025 assessment.
Expand low-carbon electricity, grids and storage. Electrify transport and heating while improving efficiency, so rising energy demand need not mean rising fossil-fuel emissions.
Reduce fossil-fuel leaks and other methane sources, and protect forests, wetlands and soils. Faster emissions reductions limit the warming that drives feedbacks.
Diversify mountain livelihoods, reduce water and energy pressure, and plan for shorter or less reliable seasons. Snowmaking has physical and environmental limits.
Protect habitats, improve wastewater systems and plan for changing shorelines. Local resilience matters alongside global emissions cuts.
Explore the questions behind the charts. Every guide brings together published evidence, practical implications and the limits of what we can infer.
Understand European ski snow risk, snowmaking limits and seasonal planning, with clear warming levels, study definitions and scientific sources.
Explore changing wave climate, sea level and coastal access, with sourced science explaining why an offshore projection cannot predict a surf break.
Compare northern and southern French Alpine snow reliability, understand elevation thresholds, and separate climate evidence from resort forecasts.
Understand Pyrenean snow-supply risk, snowmaking assumptions and regional differences, with research and practical questions for seasonal planning.
Research edition
What the latest published assessments tell us about the year behind us and the seasons ahead. This edition includes observations from 2025, with their original baselines and uncertainty.
Read the 2026 research editionEach chart page includes a permanent link, source notes, a data table and reusable PNG, SVG and CSV downloads. Embeds keep attribution alongside the visual. Original source rights and the reuse terms travel with each download.
This observatory brings together published assessments and studies, reviewed on 11 October 2026. Projections remain conditional on emissions and model assumptions. Each chart names its time period, baseline and range; those definitions differ between studies.
Compiled and interpreted by PeakWave. Published assessments and peer-reviewed studies are cited directly. The sporting explanations are our editorial interpretation, not an endorsement by the study authors.
Evidence checked: . Recent weather records have their own coverage dates. New evidence is reviewed before projections change.
11 October 2026: first observatory edition. Added regional comparisons, published snow-reliability and snow-duration tables, reusable chart pages, research guides and the 2026 State of the Seasons research edition.
For a correction, include the chart name, selected place and source in your feedback. All numeric projections retain their study’s scenario and baseline.
Section 9.2.1.1 and Figure 9.3. Global mean sea-surface temperature change in 2081–2100 versus 1995–2014; CMIP6 means and 5–95% model ranges from the chapter's detailed assessment. This is neither absolute water temperature nor a beach-level forecast.
Read the original source ↗Table SPM.1. Global surface temperature relative to 1850–1900; best estimates and very likely ranges. SSPs are conditional pathways, not probability-ranked forecasts.
Read the original source ↗Final Table 9.9, medium-confidence processes. Global mean sea-level rise relative to 1995–2014; medians and likely ranges. Local land movement, ocean circulation and low-likelihood ice-sheet outcomes are not represented.
Read the original source ↗Section 12.4.7 and Box 9.2: global frequency in 2081–2100 relative to 1995–2014. Reported model ranges, not a prediction for a particular beach.
Read the original source ↗2,234 resorts across 28 European countries at +2°C and +4°C global warming. Very high snow-supply risk is a study-defined hazard, not a probability of resort closure. Adaptation assumes snowmaking on 50% of slopes.
Read the original source ↗Research team's regional explanation of François et al. Percentages at +2°C; the French Alps value without snowmaking is approximately one third.
Read the original source ↗Section 3.1.2. Regional median elevations during 2030–2050 across RCP2.6, 4.5 and 8.5. Groomed natural snow: at least 100 days with 100 kg/m² between 15 December and 15 April in 7 winters out of 10. These are scenario spreads, not uncertainty percentiles or universal snowlines.
Read the original source ↗Cross-Chapter Paper 5. Mountain snow, glacier and permafrost impacts, with differences by elevation and latitude.
Read the original source ↗High mountain assessment. Regional ski-season risks depend on emissions, altitude, latitude and adaptation. High-emissions late-century outcomes are not near-term forecasts.
Read the original source ↗Peer-reviewed early publication, 3 September 2026. Regional simulations find greater relative resilience at high, snowy resorts; Colorado resorts generally lose less snow security than resorts nearer the Pacific coast.
Read the original source ↗Multi-model wave projections under RCP8.5, 2081–2100 versus 1979–2004. Regional mean height and period changes of 5–15%, direction changes of 5–15°. These are not changes in surf quality at individual breaks.
Read the original source ↗Projected significant wave-height increases in the Southern Ocean and tropical eastern Pacific, and decreases in the North Atlantic and Mediterranean under RCP8.5. Mean changes do not describe every storm or season.
Read the original source ↗Further warm-water coral reef decline of 70–90% at 1.5°C (high confidence) and more than 99% at 2°C (very high confidence), compared with the report's present-day baseline. Not a measurement of reef loss since 2026.
Read the original source ↗Process explanation: thawed organic matter can release carbon dioxide and methane. Page accessed October 2026; access year is not a new projection publication.
Read the original source ↗Ocean warming, acidification, habitat loss and changing marine species distributions. Fish stocks also depend on fishing pressure and management.
Read the original source ↗The Atlantic overturning circulation is expected to weaken. A dated collapse forecast or a claim that the jet stream simply breaks down would overstate the evidence.
Read the original source ↗Current-policy warming around 2.8°C over this century; full national pledges 2.3–2.5°C. Annual greenhouse-gas emissions in 2035 need to be 35% / 55% below 2019 for 2°C / 1.5°C-aligned pathways. This policy assessment is separate from the IPCC SSP experiments.
Read the original source ↗Higher background sea level increases high-tide flooding. It does not imply that every location's astronomical tidal range increases. Accessed October 2026.
Read the original source ↗The evidence points to an uneven future. Lower and warmer terrain loses snow reliability sooner; colder, higher locations retain an advantage. Snowmaking can reduce risk within its limits. The European risk percentages above describe snow supply, not a list of resorts predicted to close.
Regional wave projections describe offshore climate. Surf quality also depends on bathymetry, swell direction, tides, local wind, erosion and reef condition. PeakWave does not convert a global warming pathway into an unsupported local surf score.
No. They show stored modelled weather and conditions scores with explicit coverage. Long-term climate analysis normally needs decades of comparable data. A short archive is useful context for recent seasons, including natural variability, but cannot establish a trend by itself.
El Niño and La Niña influence seasonal probabilities, with regional differences. They do not determine every local winter. Use the dated 2026–27 ski-season outlook for the shorter-term evidence; the climate charts here address longer horizons.
The warming chart uses IPCC very likely ranges; sea level uses likely ranges for medium-confidence processes. Elevation bars span regional medians from different scenarios, and marine-heatwave bars show the reported projection range. None is a probability that an emissions pathway will happen.