What our figure is
The “snow on the ground” line on every forecast page is the middle value of three weather models’ estimates of snow depth at the elevation of that forecast. It is not a measurement. Each model keeps a running account of the snow on its own grid, adding what it thinks fell and taking away what it thinks melted, over a cell several kilometres across. The number is the model’s best guess for open, level ground at that height, smoothed over the whole cell. The methodology page has the details.
That makes it good for two things: the trend, because the same estimate is made the same way every three hours, and comparison between places, because every place is measured with the same ruler. It is a poor guide to what is under your skis.
What the resort’s figure is
A resort measures snow at a stake, usually at one or two spots it has chosen, at a fixed time each morning. The piste it stands on has been groomed, which packs the snow down, and at many resorts topped up with snowmaking, which the models know nothing about. It is also a single point, on a slope that may catch or lose snow to the wind, chosen partly because it holds snow well.
Resorts do not all report the same quantity, either. Some give a depth at the base and the summit. Some give only the season’s accumulated snowfall. Some give the last 24 hours. The label matters more than the number.
A worked example: Portillo, 21 September 2026
On the morning of 21 September 2026 our forecast for Portillo in Chile said about 1.2 m of snow on the ground, both for the top lift at 3,310 m and for the base at 2,800 m.
Portillo’s own report that morning gave three depths: 269 cm at Plateau, 245 cm at the hotel and 150 cm at Juncalillo. It had also measured 32 cm of new snow in the previous 24 hours.
So the resort’s measured depths ran from 150 cm to 269 cm across the mountain, and our estimate sat below all of them. That is the usual direction. A model smooths a steep bowl and the flat ground around it into one cell, so it misses the way terrain like Portillo’s collects snow, and it never sees grooming or snowmaking. The spread within the resort’s own figures, a difference of well over a metre between two of its stakes, is the other lesson: there is no single true depth on a mountain.
Valle Nevado, a short drive away, shows the other problem. Its report for 20 September 2026 gave 20 cm in the previous 24 hours and a season total of 567 cm at 3,000 m, but no depth on the ground at all. Our page that morning said about 1.6 m at the top lift and 92 cm at the base. The two sets of numbers cannot be compared, because they measure different things.
Which to use
- Planning a trip or comparing resorts: ours. It is consistent from place to place and from day to day, and the monthly snow record on each page is built the same way.
- Deciding what to ski tomorrow: the resort’s. Every resort page here links to the resort’s own snow report and lift status where we have them.
- Watching a storm: both, for different reasons. Our forecast says what is coming and how sure the models are. The resort’s report, the morning after, says what actually arrived on the pistes.
If our depth reads higher than the resort’s, the likeliest reason is that the resort measures on a slope that has lost snow to wind or sun, or that its stake is lower than our forecast point. If ours reads lower, which is more common, the resort’s slopes are catching more than the model’s flat cell, or the snowmaking is doing its job.