Shirakawa-go's Snow Falls at Exactly 0°C. That Is Why the Roofs Are 60 Degrees.

Japan has thatched villages in a lot of places. Miyama outside Kyoto, Ouchi-juku in Fukushima, and dozens of smaller ones.

Only two places built the roofs like praying hands: Shirakawa-go in Gifu and Gokayama in Toyama, where the pitch runs to 45–60 degrees and the frame is lashed with rope rather than nailed (Shirakawa village office).

Steep roofs shed snow. Everyone knows that much. The question worth asking is why these villages needed it and Miyama did not, and that turns out to be measurable.

The snow at four thatched villages

Deepest snow on the ground each winter, median of the last three, from the same reanalysis we use for every ski area on this site.

Village	Style	Deepest snow (median)	Range over 3 winters	Temp when snowing
Shirakawa-go (Ogimachi)	gasshō	162 cm	93–223 cm	−0.0°C
Gokayama (Suganuma)	gasshō	133 cm	67–212 cm	−0.9°C
Gokayama (Ainokura)	gasshō	76 cm	59–157 cm	−0.9°C
Ouchi-juku	ordinary thatch	63 cm	49–90 cm	−2.1°C
Miyama Kayabuki-no-Sato	ordinary thatch	13 cm	7–24 cm	−0.7°C

Shirakawa-go gets twelve times the snow Miyama does. Two villages, the same building material, the same country — and one of them has to survive a load the other never sees.

The second number is the one that matters

Look at the last column. Shirakawa-go’s snow falls at 0.0°C.

That is not a rounding artefact of a cold place. It is the specific temperature at which snow is heaviest. Snow falling at −15°C is dry, low-density and light; the Daisetsuzan volcanoes in Hokkaido average −15.8°C on snowing days and produce the powder people fly for. Snow falling at 0°C is wet, dense, and closer to slush than to powder.

Shirakawa-go has the deepest snow of the four and the heaviest snow of the four, at the same time. That combination is the design brief.

Some arithmetic makes the scale plain. A flat roof under 162 cm of wet snow at roughly 300 kg/m³ is carrying about 490 kilograms per square metre — half a tonne standing on every square metre of roof, refreshed all winter, and considerably more in a 223 cm year.

What 60 degrees is actually for

The pitch is not there to hold that load. It is there so the roof never has to.

At 45–60 degrees, wet snow slides before it can build to full depth. The roof sheds continuously through the winter instead of accumulating to February and then bearing everything at once. A shallow roof — Miyama’s, say — would need to carry it, and thatch on a light timber frame cannot.

Two other features follow from the same problem:

No nails. The frame is bound with rope and neso withes, which lets the structure flex under wind and shifting load rather than splitting at fixed joints.

The snow improves the thatch. Weight presses the straw down and compacts it, and a compacted thatch roof sheds water better than a loose one. The load the building is designed to survive is also, up to a point, doing it a favour.

And the gables face the same way. The houses in Ogimachi are aligned so their gable ends point along the valley — into the prevailing wind, and presenting the least surface to it, while the long roof faces catch enough winter sun to help the snow release.

Why not just build steep everywhere

Because a 60-degree roof is expensive. It uses far more thatch than a shallow one covering the same floor, it takes far more labour to re-thatch, and re-thatching a gasshō house is a village-scale operation that happens every few decades.

Miyama, at 13 cm, would be paying that cost for nothing. Ouchi-juku at 63 cm sits in between and builds in between.

The roof pitch across these villages is a reading of the local snow. Not a style, not a regional taste — an answer to a number, and the number is measurable three centuries later.

If you go

Both villages are deep in the mountains and both are surrounded by ski areas that see the same snow. Hakusan Ichirino, 18 km from Shirakawa-go, is 65th in the world for snowfall; Hakusan Seymour is 81st; IOX-Arosa is 16 km from Suganuma.

January and February are when the roofs look like the photographs. Our figures are the seasonal maximum, which in these valleys is usually reached in late January or February — see when there is actually snow at Japan’s festivals and hot spring towns for how much that timing varies year to year.

Both villages are UNESCO World Heritage sites and both are lived in. They are not open-air museums.

Where this is weak

The roof pitch figure is not ours. 45–60 degrees comes from Shirakawa village’s own description, cited above. We measured the snow, not the buildings.

Snow depth is modelled. ERA5 works on a 10–30 km grid that flattens terrain, so these figures represent the valley rather than a specific roof. The comparison between villages is the sound part; a single number is approximate.

Three winters. The 93–223 cm range at Shirakawa-go shows how much one winter differs from another, and three is not enough to describe the tail. A century of records would show worse years than any of these.

The load estimate is illustrative. Snow density varies from about 150 kg/m³ freshly fallen to over 500 kg/m³ once compacted and wet. We used 300 as a middle figure; the true load at any moment could be half or twice that.

Correlation, not proof. We have shown that the villages with the steep roofs have far more and far heavier snow than the villages without them. That is consistent with every account of why they are built this way, but four villages is an illustration rather than a study.

How this was measured

Daily maximum snow depth and mean temperature from Open-Meteo historical data on the ERA5 reanalysis, 1 December to 31 March, for the last three completed winters. “Deepest snow” is the maximum depth reached in a season, taken as the median across the three. “Temp when snowing” is the mean daily temperature across days that recorded 1 cm or more of snowfall — 190 such days at Shirakawa-go over the three winters.