How accurate is the snow temperature?
raceday.ski calculates the snow temperature and checks itself against real measuring stations in Switzerland and Tyrol. How that works — and how accurate it is.
We check ourselves against real measurements
Most wax recommendations look up the air temperature. raceday.ski instead calculates how warm the snow surface is — for your slope, your altitude and your start time. That is the number that matters for waxing.
But a calculated number is only worth as much as the check behind it. So a simple comparison runs in the background: every two hours, the model calculates the value it expects at each measuring station — and compares it with what the station actually measured. Every deviation is recorded. This happens automatically, all winter long, with nobody helping it along.
So where the model is off, we see it — before you notice it on the slope.
How accurate is that?
The honest answer has two parts.
First: we publish the deviation this comparison produces — openly and continuously updated on the accuracy page. It shows how far the model sits from the measurement on average, broken down by time of day and weather situation. Including when the numbers do not flatter us.
Second: accuracy is not a single value. Under clear skies and little wind, the model hits the snow temperature considerably better than during foehn, rain or close to freezing — there every forecast becomes uncertain, ours included. That is why every recommendation comes with an uncertainty band (the “± X °C” next to the temperature). A wide band does not mean we calculated badly — it means conditions are hard to calculate that day, and that you should plan a little more margin when waxing.
What we do not do: claim an accuracy we have not measured.
How we develop the model
Before a model version goes into service, it re-runs 18 past winters at the Weissfluhjoch measurement field above Davos — hour by hour, against the snow surface temperature actually measured there. That is not an operational figure but a test bench under controlled conditions: a flat measurement field, not a prepared race piste.
The test bench exists above all to reject approaches. One example: a physically elegant turbulence model was meant to damp the night-time heat exchange more strongly in wind. On the test bench it then cooled the snow far too much — real exchange does not collapse in wind the way the theory suggested. The approach was restricted to the calmest case and otherwise discarded.
That is how trust is built: not by everything working, but by what does not work failing against the measurements and then being thrown out.
Where the measurements come from
We do not measure ourselves — we calculate against those who have done it for decades. In Switzerland these are the automatic stations of the WSL Institute for Snow and Avalanche Research SLF, in Austria those of the Tyrol Avalanche Warning Service. Together over 270 stations in high alpine terrain, delivering snow and weather data around the clock through the winter half-year.
Both networks make their data openly available. That is not a given, and it is the reason this page can exist at all.
Messdaten: WSL-Institut für Schnee- und Lawinenforschung SLF (www.slf.ch), CC BY 4.0 · Datenquelle: Land Tirol - data.tirol.gv.at (LWD Tirol), CC BY 4.0.
The raceday.ski accuracy page shows the current numbers for the season — openly and broken down by conditions.
→ Go to the Wax Advisor