Quick Facts

The 30-year baseline: when typhoons actually form

The most useful question is not “is there a typhoon coming,” but “is this the part of the calendar where one is likely.” JMA publishes the 1991–2020 normal — the average count of tropical cyclones forming in the Northwest Pacific each month over a 30-year window. The pattern is heavily skewed toward late summer.

July through October account for roughly 18 of the 25 cyclones formed in a typical year. Winter and early spring are quiet enough that most travel guides simply ignore typhoons as a January-through-April risk; the average for those four months combined is 1.5 cyclones, and very few of them ever turn toward Japan.

Formation peaks in August, but damage peaks in September

August produces more storms (5.7 on average) than September (5.0), but the damage record looks almost the opposite. JMA’s list of typhoons that caused major damage in Japan is dominated by September strikes. The reason is meteorological: by September, the autumn rain front (akisame zensen) is hovering over the Japanese archipelago, and an incoming typhoon’s outer rainbands ride along the front. The result is heavy, prolonged rainfall — often well inland, well away from the storm centre — and the worst flooding events tend to be September events. If your trip is flexible and you want the lowest weather risk, August is usually the busier-but-shorter-event month and September is when the bigger systems hit.

Where the storms hit hardest

Typhoon risk is not distributed evenly across Japan. Okinawa and the Amami Islands take the brunt of approaches every year, while the main island regions see roughly equal exposure to one another but at much lower frequency.

Average approaches per year, 1991–2020. Source: JMA.

Two practical takeaways. First: a Tokyo or Kyoto trip in August or September faces three to four typhoon approaches across the whole season — a real but manageable risk, not the constant threat it sometimes sounds like. Second: an Okinawa trip in the same window is meaningfully riskier, with about one approach every six weeks on average. If you are travel-insurance-minded, that’s where to focus the hedging.

What 2025 looked like, and what 2026 has shown so far

Recent seasons help calibrate expectations. The 2025 Pacific typhoon season produced 27 named storms, 13 of which reached typhoon strength. By accumulated cyclone energy (a standard measure that combines storm count, intensity and duration), 2025 was below average, and only one storm — Typhoon Ragasa — reached super typhoon strength. The first named storm did not form until June 11, the fifth-latest start to a season on record.

2026 has done the opposite. The first named storm, Nokaen, formed on January 15 — the first January named storm in the Pacific basin since Pabuk in 2019 — and the season’s first typhoon, Sinlaku, reached typhoon status on April 10 and then super typhoon strength on April 12. An early start does not mechanically mean a busier total season; the relationship between early-season activity and late-season activity in the Northwest Pacific is weak. But it is a season worth watching attentively.

Live tracking: the sources Japanese forecasters watch

If you have a trip in the August–October window, the question is not “what does the climatology say” but “is anything moving right now.” Four sources give you 90% of what a domestic Japanese forecaster watches:

What actually cancels (and when)

One of the most useful things to know in advance: Japan’s transportation operators are conservative on flights, less conservative on shinkansen, and most conservative on ferries. The cancellation order is fairly consistent storm to storm.

Two recurring patterns are worth flagging because they trip up first-time visitors. First: a typhoon with its centre well offshore can still suspend the Tokaido Shinkansen if the wind sensors along the track exceed thresholds. The trains stop because of the side wind on the embankment, not because the storm is overhead. Second: ferry cancellations cascade. If the Naha–Ishigaki run cancels on Tuesday, the Wednesday and Thursday departures are often also cancelled or sold out, because passengers are bumped forward. If your itinerary depends on a Yaeyama or Ogasawara ferry, build in a one-day buffer at minimum.

An exact cancellation threshold in km/h does not exist as a single published figure. Each carrier publishes its own internal criteria — wind speed at the relevant airport or track segment, sustained vs gust, plus precipitation thresholds. The public-facing decision arrives as “this flight is cancelled” rather than “wind exceeded X” — which is why the carrier status page is the only authoritative source within 48 hours of a storm.

Pre-trip checklist if your dates fall in typhoon season

FAQ