Guides to snow and snow forecasts
Short answers to the questions that come up when you are reading a forecast before a trip.
Reading the weather
What are the ECMWF, GFS and ICON weather models, and which is best for snow?Three of the world's main global weather models, run by Europe, the United States and Germany. Each has strengths, none is always right, and comparing them is the most useful thing a snow forecast can do.What is the freezing level, and why does it matter for snow?The freezing level is the height where the air reaches 0 °C. Above it precipitation falls as snow, below it as rain, give or take a few hundred metres, which makes it the single most useful number for judging whether a storm will snow on your resort.How many days ahead is a snow forecast reliable?Snow forecasts are good for about three days, useful for five, and a guide to the pattern beyond that. Amounts are the least reliable part. Here is how to read a 14-day forecast without being misled by it.Why does Japan get so much snow, and what is lake-effect snow?When very cold air crosses relatively warm water it picks up moisture and dumps it as snow on the first land it meets. This is why Hokkaido, the Sea of Japan coast, Buffalo and Syracuse get more snow than places at the same latitude a few hundred kilometres away.Why is it raining in the resort but snowing on the mountain?Air cools by about 6 °C for every 1,000 m of height, so a storm that is a few degrees too warm for snow in the village is often a few degrees too cold for rain at the top lift. The difference is normal, and it is why snowingplaces forecasts both.What will the 2026/27 ski season be like? A very strong El Niño is comingNOAA says there is a greater than 90% chance of a very strong El Niño through the 2026/27 northern winter, and a 75% chance of the strongest on record. Here is what that has meant for snow in past El Niño winters, region by region, and how much to trust it.How much is 10 cm of snow, and what does it do for skiing?Ten centimetres of fresh snow is about four inches, enough to refresh the pistes and soften the off-piste, but not a powder day. Here is what the numbers in a snow forecast mean on the ground.How we work
The choices behind the forecasts and reports on this site, and what they can and cannot tell you.
How do snowingplaces forecasts work?The models, the thresholds behind Likely and Possible, and what the ten-year snow record can and cannot say.How do we grade a ski day, and pick the best day to ski?Each day of a resort's upper-mountain forecast gets a score out of ten from four things, fresh snow, the sky, wind and rain, and the highest scoring day is named the best day. Here is one real week, day by day, to show how the arithmetic goes and what it leaves out.What do Likely and Possible mean on a snow forecast?Likely means all three weather models forecast snow. Possible means two do, or they disagree. It is a vote, not a percentage, because three models are too few for a probability and a vote is honest about what we know.Why does our snow depth differ from the resort's snow report?Our figure is a weather model's estimate of snow on the ground across a grid cell several kilometres wide. The resort's is measured on its pistes, and resorts do not all measure the same thing. Neither is wrong, and they are for different jobs.Why does the snow record show snowy days but not centimetres?The monthly snow record on every page comes from a global reanalysis on a grid about 30 km wide. It is trustworthy on when it snows and how one winter compared with another, and badly wrong on how much falls in the mountains. So we publish the first and not the second.What can a snow forecast not tell you?Global weather models are good at the broad pattern and weak on the exact amount at one spot in steep terrain. This guide sets out the limits of what the site knows, one by one, and says what to check instead.Why do we show a snowfall range instead of one number?A total such as "10–20 cm" runs from the lowest of the three models to the highest. The width of the range is information: a narrow range means the models agree on the amount, a wide one means the storm is real but the amount is not settled.