Haiyan is on record at 170 knots and at 125 knots. Both numbers are correct, and they are not the same measurement. What is left once you allow for that is a disagreement that reverses sign, grows with the storm, and in forty years has not closed.
IBTrACS is not one archive. It is every agency's own best track laid side by side in 174 columns, and NOAA's column documentation says plainly of the summary column: “NO adjustment is made for differences in wind speed averaging periods.”
A maximum sustained wind is an average, and the agencies average over different windows. JTWC uses a 1-minute mean. JMA uses a 10-minute one. Hong Kong and Korea also use ten minutes; China uses two. The same gust of wind, averaged over ten minutes instead of one, is a smaller number — so part of the gap between any two agencies is definitional before anyone has disagreed about anything.
Which makes the interesting question not whether they differ, but whether the difference behaves like a unit conversion. A unit conversion is a constant. This is not.
Across 17,244 observations where both agencies put a number on the same storm at the same hour, JMA's wind as a share of JTWC's is 1.2121 for the weakest systems, exactly 1.0 at tropical-storm strength, and 0.7407 for category-5 winds.
For systems JTWC puts below tropical-storm strength, JMA's median wind is 7 knots higher, not lower. A ten-minute average cannot exceed a one-minute average of the same wind, so this is not the averaging window.
At tropical-storm strength the median ratio is exactly one and the median difference exactly zero. Two independent estimates do not land on the same number that often.
On the 1,055 observations JTWC puts at category-5 strength, its median wind is 35 knots above JMA's.
Of 840 paired observations recorded over Philippine land, the two agencies put the storm on opposite sides of the typhoon threshold 19.88% of the time — and the disagreement runs almost entirely one way.
At the typhoon line, JTWC alone calls it a typhoon 158 times against JMA's 9. At the major-typhoon line the split is 92 to nothing. Go the other way, down to tropical-storm strength, and it reverses completely: there JMA alone is above the line 57 times and JTWC never is.
This is the part with consequences. Warnings, evacuation orders and pre-positioned relief are keyed to categories, not to knots, and the category a storm is in depends on which agency you read.
If the gap were a fixable measurement problem you would expect it to shrink as satellites improved. The median difference on storms JTWC puts at typhoon strength was 12 knots in the 1980s, 20 knots in the 1990s, and 17 knots in the 2020s.
The sibling page to this one ends a section by saying that whether recent record intensities reflect stronger storms or better measurement “is not something this archive can settle”. This is why. The four storms it lists as sharing the record at 170 knots are JTWC's numbers. JMA has Haiyan at 125 and Surigae at 120. The archive cannot settle the question because the two agencies inside it do not agree about the storms the question is about.
Three limits, and the first one is the one that would break the page if it were ignored.
JMA only assigns a wind once a system reaches tropical-storm grade. Every figure here is computed on paired observations only; 13,247 of 30,491 observations in the box were dropped because one agency said nothing. Treating those as agreement would have made the gap look far smaller than it is.
Nothing here says which is right. There is no independent measurement of a typhoon's peak wind to check either against; both are estimates from satellite imagery, and the page compares them rather than scoring them.
“Philippine waters” here is 5–25°N, 115–135°E — the bounding rectangle of the official Area of Responsibility, not its polygon. It catches some storms the PAR would not.
Every figure on this page traces to one of these, through a CSV in
data/ph-typhoon/. Each is checked against its source query on every
build.
243,281 six-hourly observations of 4,105 western Pacific storms from 1884: position, one-minute sustained wind, Saffir-Simpson category, and distance to the nearest land — A track archive with no impact data of any kind — no deaths, damage or people affected. Intensity is missing for 100% of pre-1945 observations and 43.3% of 1945-69, so every rate and trend here starts at 1980.
The Philippine authority for the Area of Responsibility and for local storm names — Cited for provenance and for the PAR definition. pagasa.dost.gov.ph is not reachable from a script here, and IBTrACS carries international names — Haiyan rather than Yolanda.
Disaster response reports carrying people affected, houses damaged and barangays served — The source the previous version of this page cited for its four impact figures. It is named here because impact is the obvious question this page cannot answer; none of its numbers are used, and IBTrACS contains none of them.
The wind averaging period each agency reports on: JTWC 1-minute, JMA 10-minute, CMA 2-minute, Hong Kong and Korea 10-minute — The whole agency comparison rests on these five lines, so they are quoted from NOAA's own document rather than recalled. The same document warns of the summary wind column that "NO adjustment is made for differences in wind speed averaging periods" -- which is why a page that reads one agency's column and calls it the wind is making a choice it has not declared.
I wrote a companion post — “Two Weather Offices Watched The Same Storm. They Wrote Down Different Numbers.” — in simple, everyday words.