Data and scoring

Winter snowfall for 3,069 ski areas worldwide. 397 of them are in Japan, and for every Japanese one we also measure not just how much falls but what falls — the temperature when it snows, how often it snows, how often it rains instead, and the wind.

The 79 onsen towns designated by Japan's Ministry of the Environment are measured exactly the same way. Lists of onsen are easy to find; a list measured on the same ruler as the ski areas is not.

The point is not that any one place has been studied carefully. It is that everywhere has been measured identically. Comparability matters more here than the precision of any single value.

What counts as one ski area varies by place. Hakuba Valley's ten ski areas are counted separately, because each sells its own day ticket even though a joint pass exists. Niseko United (Hirafu, Village, Annupuri and HANAZONO) is counted as one place and measured at a representative point. Shiga Kogen is counted as nine separately run areas, such as Yakebitaiyama and Yokoteyama, plus a single entry covering the smaller slopes around Hasuike. This is one place where we have not managed to apply a single rule everywhere, and we are fixing it as we find it.

Snowfall is the sum of daily snowfall from December to March, averaged across the twelve most recent completed seasons. Southern-hemisphere resorts are calculated over June to September. We widened this from three seasons because a single big winter could lift a resort far above where its long-run average belongs.

The source is Open-Meteo historical data on the ERA5 reanalysis — one calculation covering the whole world, which is what makes cross-border comparison possible at all.

Snow quality is built from four things: the temperature when it snows, how often more than 5 cm falls, how often it rains, and the wind on snowing days. How much snow falls is deliberately excluded — including it would make the quality ranking a restatement of the snowfall ranking.

The weights and thresholds are our judgement, not a measurement. Someone else would choose differently and the order would shift. What the ranking can carry is that every place went through one identical calculation.

A single resort's "typical" figure is an average; a prefecture's or country's is a median. A resort's typical snowfall is the mean of its seasons. For a prefecture or a country we take the middle value of the resorts in it, so that one extreme place cannot drag the whole figure.

Share and amount can point in opposite directions. Hokkaido loses less of its snow between January and March than Japan as a whole, yet gets less snow in March than the national median. Niigata is the reverse: the steepest fall, and still among the most March snow in the country. Where the answer depends on which you look at, we give both.

Read the ranks and the comparisons, not the centimetres. The estimate runs below what is actually measured, so these figures cannot be set beside a resort's published snow depth or another site's numbers. They exist to compare places measured the same way.

The snowfall figures come from a reanalysis — an estimate, not a record of what fell. We checked how far off they are against 63 Japan Meteorological Agency weather stations.

They come to about six tenths of what was actually observed (mean ×0.60). They run low, and they run low by different amounts in different places: some read barely a third of the real figure, others match almost exactly.

We tried scaling everything by a fixed factor to correct for it. It barely helped, because the error differs from place to place. Having established that it cannot be fixed, we do not correct the figures — we say so instead.

The one exception is the regional map on the Japan page. Comparing the 156 ski areas that have a Japan Meteorological Agency station nearby, the error turned out to cluster by region: Hokkaido comes out at about 52% of what was observed, Hokuriku at about 88%. The estimate converts precipitation into snow at a fixed ratio, so Hokkaido's dry, fluffy snow comes out lowest. The error at any single place cannot be fixed, but the middle value for a whole region can, so the map alone adds that difference back. Resort pages, prefecture pages and the rankings all use the uncorrected estimate.

The stations used for the check were chosen mechanically: within 15 km of a ski area, at most four per prefecture, nearest first. That rule exists so it can be verified that we did not pick convenient stations.

The numbers on this site are built by finding mistakes and correcting them. When we fix something, what was wrong and how it was fixed is recorded here. Articles whose conclusions changed have been rewritten and carry a note saying so.

We do not forecast. What is published here is past observation and calculation, plus same-day information supplied by our weather data provider. We do not produce forecasts of our own.

Resort-published snow depths are a different thing entirely. Each measures in its own place in its own way, so those figures are not comparable to each other. Ours are comparable only because every location went through one identical process.

Questions about how these figures are produced, enquiries about using the data, and corrections to anything published here are all welcome through the contact page.