Winter does not begin on the day the first lift turns. It begins several weeks earlier, in the nights when the snow guns start up. Those nights are not happening.
Updated 27 September 2026. The figures have been fetched again. The first version, of 22 September, covered the sixteen days to 7 October; this one covers those to the 12th.
Two things have moved, and both are more interesting than the update itself. The cold window sits five days later: it fell on 4 to 6 October and now falls on the 9th and 10th. And the cold has changed region. In the first version Davos led with 19 hours, ahead of Arosa Lenzerheide on 18. Today Davos is at zero and Arosa at one hour, and Adelboden leads.
Anyone wanting to know how much a sixteen-day forecast is worth has the answer here: usable for the first days, not for the last.
We pulled the sixteen-day outlook for ten Swiss ski areas, calculated at 2,000 metres, the altitude where the early pistes are built. The result in one sentence: 134.6 millimetres of precipitation, of which 8.1 centimetres of snow, and almost all of that falls in a single area.
When it would be cold enough
Show the figures as a table
| Entry | Value |
|---|---|
| 27 Sep | 0 h |
| 28 Sep | 0 h |
| 29 Sep | 0 h |
| 30 Sep | 0 h |
| 1 Oct (24 mm rain) | 0 h |
| 2 Oct (23 mm rain) | 0 h |
| 3 Oct | 0 h |
| 4 Oct (20 mm rain) | 0 h |
| 5 Oct | 0 h |
| 6 Oct (6 mm rain) | 0 h |
| 7 Oct (33 mm, 3 cm snow) | 0 h |
| 8 Oct (15 mm, 4 cm snow) | 0 h |
| 9 Oct | 41 h |
| 10 Oct | 47 h |
| 11 Oct | 7 h |
| 12 Oct (10 mm rain) | 0 h |
Three things sit in that chart.
First, for the first time some of it falls as snow. 8.1 centimetres across ten areas and sixteen days, on 7 and 8 October. That sounds like a turn and is not one: 6.1 centimetres of it fall in Zermatt, 1.3 in Andermatt, 0.8 in Airolo. Seven of the ten areas get not a single centimetre.
Second, the rest stays rain. The wettest day is 7 October with 33 millimetres. Grindelwald is the clearest case: 36.9 millimetres of precipitation, the highest figure in the field, none of it as snow, and not one cold hour. Rain at 2,000 metres takes old snow away, it does not bring any.
Third, the cold is not spread out but clumped. It comes on 9 and 10 October, with 41 and 47 hours, then briefly again on the 11th. Before that lie twelve consecutive days on which none of the ten areas has a single hour below the threshold. Snowmaking needs continuous nights, not scattered hours: a system that starts at two in the morning and shuts down at five has the cost and not the snow.
Why air temperature is the wrong number here
Weather reports give the air temperature, and anyone who uses it gets this wrong on a regular basis. What matters for a snow gun is the wet-bulb temperature: water travelling through dry air as droplets cools itself by evaporation. At twenty per cent humidity snow can still be made at plus two degrees of air temperature; at ninety per cent it cannot.
The thresholds, as manufacturers and lift companies give them: minus two degrees wet-bulb is the marginal band where modern lances just about run, minus four degrees is reliable operation, and only from minus six degrees does the volume add up that actually matters. Minus six degrees is not reached in these sixteen days in any of the ten areas, in any single hour. The coldest value in the whole field is minus 5.6 degrees, in Andermatt.
Where it would be most likely
Five days ago this order was almost upside down. Davos and Arosa Lenzerheide led the field then; today they sit at the bottom. Calculated at a fixed 2,000 metres, what counts is not how high an area reaches overall but how cold the air sits above the particular valley, and in two weeks that air is different air.
Where it would work, it is forbidden
Adelboden tops the field with 23 hours and still may not. The canton of Bern runs a fixed calendar: snowmaking is permitted from 15 October to 15 March inclusive, and a cold night on 9 October does not change that. The same goes for Grindelwald, which has no cold hour anyway.
Grisons dropped its dates in 2007 and relies on conditions instead of the calendar, but has the fewest cold hours in this outlook: Davos none, Arosa one, St. Moritz three. Five days ago it was exactly the other way round, with Grisons in front and allowed to run. The rules in detail are in our piece on when a ski area may make snow at all.
What the figures do not say
This outlook ends on 12 October. A weather model does not sensibly reach further, and what happens after that we do not know. How little the later days are worth is shown by the update above: the cold window that sat on 4 October on 22 September sits on the 9th today.
It also applies to 2,000 metres. The glacier areas sit a thousand metres higher and have their own figures; those are in the piece on glacier opening dates. Up there the picture is mixed: Kaunertal called off its 26 September start and has named no new date since, Pitztal opened that same day, and Saas-Fee resumes glacier skiing on 3 October, at first only on the two upper drag lifts.
And it says nothing about whether an area opens on time. An area starting in December has ten weeks left, and in a normal year the first usable snowmaking nights come only in late October anyway. What these figures show is a delay, not a cancellation. Which areas have named a date is in the opening dates overview.
Read on: When may a ski area make snow? · Six glaciers have a date · Where can you ski in October? · What will winter 2026/27 be like?
Sources and method
- Weather data: Open-Meteo, which combines the MeteoSwiss and DWD models, retrieved 27 September 2026. Queried hourly were wet-bulb temperature and relative humidity, and daily precipitation and snowfall, for sixteen days from the retrieval date. The first version of this piece used the 22 September retrieval.
- Altitude: fixed at 2,000 metres for every area. Without that setting the model uses the terrain height of its grid point, which in high mountains is regularly several hundred metres off. The fixed altitude makes the ten areas comparable; it does not match the actual snowmaking terrain of each one.
- Thresholds: minus two degrees wet-bulb as the marginal band, minus four degrees as reliable operation, minus six degrees for high output. These are the figures manufacturers and lift companies give; there is no exact limit, because it depends on the system, the water pressure and the wind.
- Snowmaking calendars: canton of Bern, snowmaking ordinance, 15 October to 15 March. Grisons, guidance of the office for spatial development, 2007 revision, “no fixed earliest snowmaking date”. Both quoted in full in the linked piece.
- Glacier dates: notices on kaunertaler-gletscher.at, pitztaler-gletscher.at and saas-fee.ch, all retrieved 27 September 2026.
- Checking the figures: the script that fetches them is in the repository at
scripts/beschneiung-pruefen.mjs. Running it again gives today’s state, not that of 27 September.
