Plan with facts
When to Go · Calendar

Japan Fall Foliage Forecast: How Koyo Timing Works, and When to Book

Kyoto's autumn colour now books out almost as far ahead as cherry blossom week, but the koyo forecast works differently: a three-week window instead of a ten-day gamble, moving south from Hokkaido's mountains to Kyushu.

Red maple leaves framing a wooden pagoda at Daigo-ji temple in Kyoto during peak koyo season

Kyoto’s autumn colour books out almost as far ahead as cherry blossom week, and for the same reason: everyone wants the same ten days and nobody can say in July which ten days those will be. The koyo forecast exists to close that gap, and it works differently from the sakura forecast in one important way: the window it gives you is two to three weeks wide, not ten days, because a whole hillside of mixed trees does not turn colour on the same schedule that one cloned variety of cherry does.

This guide covers who produces the koyo forecast, how it is calculated, how the colour actually moves down the country from Hokkaido to Kyushu, and the booking strategy that works when “peak” is a three-week range rather than a single date.

The short answer

First forecast
early-to-mid September, refreshed roughly every two weeks through early December.
Producers
the Japan Meteorological Corporation and Weathernews, the same two private forecasters behind the cherry blossom forecast.
The window is wider than sakura’s.
Best-viewing (*mikoro*) periods usually run 2-3 weeks per location, not 7-10 days.
Colour moves south and downhill
Hokkaido’s mountains from mid-September, the Japan Alps and Tohoku in late October, Tokyo and Kyoto in late November, Kyushu into mid-December.
Book Kyoto and Tokyo hotels 4-6 months out
for the last week of November; everywhere else, 6-8 weeks is usually enough.

Why the koyo forecast isn’t built like the sakura forecast

Cherry blossom forecasting works because Japan’s Somei-yoshino trees are almost all genetic clones, so an entire city’s trees open within days of each other and a forecaster can name a single flowering date with real confidence. Autumn colour has no such shortcut. A koyo view is made up of maples, ginkgo, oaks and dozens of other species, planted at different elevations, all colouring at their own pace in response to temperature and daylight rather than a single dormancy trigger.

Forecasters deal with this by publishing a best-viewing window, mikoro, for each location rather than a first-colour and full-colour date. It is a less precise-looking number, but it is arguably more honest: a mountainside does not hit “full bloom” the way a single tree does, and a two-to-three-week range reflects that better than a false-precision single date would.

Who actually produces the forecast

The Japan Meteorological Agency once tracked far more of this than it does now. Its phenological observation network, which recorded the seasonal timing of dozens of plants and animals at stations across the country, was scaled back in 2021 to a short list of species. Only cherry and Japanese maple survive at a reduced set of sites now, and mostly for long-run climate records rather than travel planning.

The forecasts visitors actually use come from the same two private companies behind the sakura forecast: the Japan Meteorological Corporation and Weathernews. Both publish free colour maps in Japanese, both are widely mirrored in English by travel and news sites, and both update on a rolling basis as autumn progresses rather than issuing one fixed forecast and leaving it.

How the colour moves down the country

Two things drive the timing: latitude and elevation. Cold arrives in the north and in the mountains first, so colour starts in Hokkaido’s interior peaks in mid-September and works its way south and downhill over the following ten weeks. A rough rule of thumb is that every 1,000 metres of elevation shifts the colour calendar by three to four weeks, which is why you can find peak colour in the Japan Alps in mid-October while the valley towns below are still green, on the same day.

RegionTypical best-viewing windowVerdict
Daisetsuzan & Hokkaido’s interior mountainsmid-Sep – early OctEarliest colour in the country
Sapporo & Hokkaido lowlandsmid – late OctReliable, fewer crowds
Japan Alps, Nikko & Tohoku highlandsmid – late OctDramatic, book lodges early
Tokyo & the Kanto lowlandsmid Nov – early DecLong window, heavy crowds
Kyotolate Nov – early DecPrices rival sakura week
Fukuoka & Kyushulate Nov – mid DecLatest colour, milder weather

So “when is koyo season” has no single answer worth memorising. A trip almost any time from mid-September to mid-December can hit good colour somewhere in the country; the only real question is whether your itinerary and the calendar land in the same place.

Mount Fuji viewed across Lake Kawaguchiko with autumn foliage in the foreground, a region that colors weeks before Kyoto
The Fuji Five Lakes region typically peaks two to three weeks before Kyoto does.

What actually decides how good the colour is

Timing and vividness are two different questions, and the forecast is much better at the first than the second. Maples need cool nights to trigger the red pigments and enough daytime sun to let them develop fully; a run of warm, overcast autumn weather can leave the leaves a dull brown-yellow even if they change colour on schedule. A hot, dry September also tends to push the whole calendar later and can mute the display when it finally arrives, and an early hard frost can do the opposite, cutting the season short by forcing leaves down before they finish turning.

None of this is forecastable more than a couple of weeks out, which is the honest limit of what any koyo forecast can tell you. It can tell you when; it is far less reliable on how good.

The booking strategy that works

Treat the koyo window the same way as sakura, but with more slack to work with. Book accommodation for the Tokyo-Kyoto corridor 4-6 months out for the last week of November, since that week now prices close to cherry blossom week. Everywhere else, 6-8 weeks is genuinely enough. And use the latitude-and-elevation gradient as insurance: if your dates land early, head to the mountains or Hokkaido; if they land late, head south to Kyushu.

How much to trust the forecast at each stage

  • Early-to-mid September (first forecast): useful only for the broad shape of the season: running early, average, or late. Treat any date as accurate to within a week or two.
  • Early-to-mid October: the picture sharpens for the northern and mountain regions, which are now only weeks away. Treat Kyoto and Tokyo dates as still provisional.
  • Two to three weeks out: reliable to within a few days for that specific location, and this is the point to lock in train reservations and finalise which city you’ll be in.
  • Inside the stated window: basically a current-conditions report rather than a forecast. Check it the week you travel, not the month before.

The overlap with typhoon season

Koyo season opens in the north while the country’s typhoon season is still technically live. A storm crossing Hokkaido or Tohoku in late September can strip leaves off the trees before they finish colouring, which is a real risk for anyone chasing the earliest colour in the calendar. It fades as a concern through October and is essentially gone by the time Kyoto and Tokyo peak in late November. See the rainy season and typhoon season guide for how the regional risk actually breaks down by month.

Where this fits against the rest of the year

November already comes out as one of the best months in the calendar on temperature, rainfall and daylight (see the month-by-month breakdown for how it compares against every other month on those measures), and koyo timing is the reason it now prices like a second cherry blossom season in Kyoto specifically. If your dates are flexible, the Alps and Tohoku in mid-October get you comparable colour at spring-shoulder prices, three to four weeks before the Kyoto crowds arrive.

If you are choosing between the two seasons entirely, the practical difference is patience versus a gamble: the cherry blossom window is genuinely ten days and can be wiped out by one storm, while koyo gives you a three-week runway and, because it stretches across the whole country for ten weeks, room to simply move to wherever the colour currently is. The cherry blossom forecast guide covers how that shorter, more fragile forecast is produced.

Sources

Data checked: 2026-09-27