How long does winter have left?
Åre's winter does not end everywhere at once. The village sits at 380 m and Åreskutan tops out at 1420 m, and a thousand metres is most of a climate zone — so the honest question is not whether Åre keeps its winter, but how far up the mountain you have to go to find it. This page works through seven climate models to answer that, and is explicit about where the numbers stop being trustworthy.
The short answer, by height
Modelled days per winter with at least 30 cm of snow on the ground — the ski industry's usual test of whether a season exists at all. Compared between the current standard climate period and the last twenty years the models cover.
The snow line, walking uphill
The same measure at every 200 m, for three periods. The gap between the top row and the bottom row is the whole story.
A century, one winter at a time
Observed record in blue, the model median in violet, and the full spread across the seven models behind it. Pick the height you care about.
The snowmaking window
Nights between November and March cold enough to run a snow gun — a wet-bulb temperature of −2 °C or below. Whether a resort opens for Christmas is decided by this number long before it is decided by snowfall.
The other half of the year
A shorter winter is a longer everything else. This is not a consolation, but it is real, and it is what the same models say about the seasons this site scores for running, hiking and riding.
What these numbers can and cannot tell you
- One scenario, not a range of futures. The high-resolution models here were only run under SSP5-8.5, a high-emissions pathway. Out to 2050 that matters less than it sounds: the pathways barely separate before mid-century, because the warming to then is largely already committed. For 2090 it would matter enormously — which is why nothing on this page is extrapolated past the end of the data.
- The data stops in 2050. Every figure here ends there. If you want a date for the last skiable winter, this page will not give you one, because the honest answer depends on emissions between now and then.
- A 25 km model cannot see a mountain. These grid cells are wider than the distance from Åre to Storlien. The whole massif is read at one point and moved to each elevation with the same lapse rate the forecast uses. That captures the height dependence, which is the dominant effect, and nothing about aspect, shading, wind scouring or which side of the ridge you are on.
- The snowpack is modelled here, not by the climate models. The climate data carries snowfall but not snow depth, so season length and the 30 cm figures come from a degree-day model written for this page: accumulate what falls as snow, melt it back at a fixed rate per degree above freezing. It has no wind redistribution, no rain percolating and refreezing, and no snowmaking. Treat those numbers as an index that moves the right way, not as a measurement.
- Snowmaking is not in any of this. Every modern resort in this region makes a large part of its base. The snowmaking panel counts whether the weather allows it; it says nothing about whether SkiStar chooses to, what it costs, or whether the water is there. A resort will keep a season open long after the natural snowpack stops delivering one.
- Model spread is not the same as uncertainty. The band on the charts is the disagreement between seven models. Real uncertainty is wider: it also includes what the models all get wrong together, and the year-to-year variability that means any individual winter can land far outside the trend. A bad winter inside a warming trend is not evidence against it, and a deep one is not evidence for the opposite.
Thresholds, the melt factor and the assumed snowpack density are all listed on the method page and can be argued with there.