Japan’s Meteorological Agency sets itself a target of about three minutes to publish a tsunami warning after an earthquake, and about two minutes for a subset of them. If the quake is bigger than magnitude 8, that first bulletin does not give a height in metres at all. It gives a word.
The target is about three minutes, and about two minutes for quakes near Japan
The agency describes the sequence in fixed order. When an earthquake occurs it estimates the size and position straight away, derives the tsunami height expected on the coast from those two values, and then issues one of three products for each tsunami forecast zone: a major tsunami warning, a tsunami warning, or a tsunami advisory. The stated goal is about three minutes from the earthquake.
A footnote narrows the faster case. About two minutes applies to earthquakes that occur in the seas near Japan and for which the earthquake early warning technique can fix an accurate epicentre and magnitude quickly. The page does not extend the two minute figure to distant earthquakes.
Whatever the timing, the content is the same pair of items per zone: the height of the largest expected wave, and the predicted arrival time of the earliest first wave anywhere inside that zone. The agency attaches a warning to the second of those. The arrival time is the earliest point in the zone, and the page states that elsewhere in the same zone the water can arrive tens of minutes later, in some cases more than an hour later.
Three products split at 1 metre and 3 metres, with five numbers inside them
The thresholds are printed as inequalities rather than as descriptions. A major tsunami warning is issued when the largest expected wave exceeds 3 metres at the highest point. A tsunami warning covers the band above 1 metre and up to 3 metres. An advisory covers 0.2 metres up to 1 metre, and only where there is a risk of damage from the tsunami. The agency also records that it places the major tsunami warning in the emergency warning category.
| Product | Expected maximum wave | Numbers used | Wording for a giant quake |
|---|---|---|---|
| Major tsunami warning | Above 3 m | Over 10 m (10 m < h); 10 m (5 m < h ≤ 10 m); 5 m (3 m < h ≤ 5 m) | Giant |
| Tsunami warning | Above 1 m and up to 3 m | 3 m (1 m < h ≤ 3 m) | High |
| Tsunami advisory | 0.2 m to 1 m, with a risk of damage | 1 m (0.2 m ≤ h ≤ 1 m) | Not stated |
Five numbers therefore sit across the three products, and the advisory row carries no word substitute at all. The agency publishes the upper value of whichever band the forecast falls into. Its own example: a tsunami predicted at 3 to 5 metres produces a major tsunami warning with the height given as 5 metres.
Above magnitude 8 the first bulletin is a word, and numbers follow in about fifteen minutes
The reason given is mechanical rather than editorial. For an earthquake above magnitude 8 the agency cannot fix an accurate magnitude immediately, so the first warning cannot rest on one. Instead it evaluates the possibility of a giant earthquake from the amplitude of the long period component of the seismic waves and from how far the shaking is spread, and if the magnitude decided immediately looks too small, it assumes the largest tsunami envisaged for that sea area.
The first major tsunami warning or tsunami warning then gives the expected height as giant or high. The agency’s stated purpose for those two words is to communicate that the situation is an emergency. It also states that even for a giant earthquake an accurate magnitude is obtained about fifteen minutes after it happens, and at that point the warning is updated and the wording is replaced by the five step numbers.
These arrangements are dated. The agency labels this set of changes as the tsunami warning improvement of 7 March 2013.
The coast is cut into 66 forecast zones, and one of them spans four prefectures
Warnings, tsunami information and tsunami forecasts are all issued by zone, and the agency divides the whole country into 66 of them. The boundary table names each zone and the area it covers, and the zones do not map onto prefectures one for one. Counting the rows of that table gives 66 zone names.
Hokkaido alone is split into six zones, separating the Sea of Okhotsk coast, three Pacific stretches and two Sea of Japan stretches. In the other direction, one zone in the table lists municipalities in four different prefectures: the Ariake and Yatsushiro Sea zone, which covers coastal parts of Fukuoka, Saga, Nagasaki and Kumamoto. One further zone, Inner Tokyo Bay, lists three, covering part of Chiba, the special wards of Tokyo and part of Kanagawa. Those counts are ours, taken by scanning all 66 rows. Okinawa Prefecture is handled by three zones, with the Daito Islands and the Miyako and Yaeyama area kept separate from the main island group.
The companion table that maps zones to individual municipalities carries its own date on the agency’s page: 26 March 2026.
The forecast is read out of a database of about 1,500 fault positions
A tsunami from a nearby earthquake can reach the coast before any simulation started at the moment of the quake could finish. The agency’s answer is to do the computing in advance. Faults capable of producing a tsunami are set up beforehand, numerical simulations are run for each, and the results are stored as a tsunami forecast database. When a real earthquake happens, the matching case is retrieved.
The dimensions of that database are published. Faults are placed at about 1,500 horizontal positions, at six depths between 0 and 100 kilometres, and at four magnitudes. Two fault parameters are not varied at all: the dip and the slip direction are fixed at a pure reverse fault dipping 45 degrees, which the agency describes as the setting that produces the largest tsunami.
Coastal heights are not taken straight from the simulation grid. The agency uses the offshore result, where it considers the error still small, and applies Green’s law to obtain the height at a water depth of 1 metre, which it then treats as the height at the coast. For arrival times it adds the travel time from an offshore prediction point, roughly 15 kilometres out, using the rule that a tsunami travels at the square root of gravity times water depth.
Zone figures ran about 1.8 times the observed height across 139 recorded cases
The agency has compared the database against past events, and publishes two different ratios from that comparison. Across 139 cases in which a tsunami was observed, the average ratio of the height measured at tide stations to the height predicted by the database was about 1.2.
The zone figure behaves differently, and the page explains why. A zone warning takes the highest of several prediction points inside that zone, so the published zone value averaged about 1.8 times the observed height. The agency sets this against the opposite risk in the next paragraph: because local topography changes tsunami height sharply, the water can also come in higher than the figure given.
While a warning is running, a small measured wave is published as words
Observed heights are held back in a defined way. Under a major tsunami warning, a measured maximum wave above 1 metre is published as a number, and 1 metre or below is published as under observation. Under a tsunami warning the switch sits at 0.2 metres. Under an advisory every value is published as a number, with a very small one expressed as weak. The agency’s stated reason is that later waves can be higher than the first, so a small early number invites people to stop evacuating.
The page defines several more pieces of fixed wording. Where the first wave cannot be timed, the bulletin says the first wave could not be identified. Where a station stops returning data, the bulletin says the observation is missing, wording the agency introduced on 24 July 2025.
Offshore stations have their own rule. For stations more than about 100 kilometres from the coast the agency does not publish a coastal estimate at all, because tying them to a forecast zone is difficult, and the measured maximum is given as words rather than a number. Where no tsunami disaster is expected, the product issued is a tsunami forecast instead, one case of which is a predicted sea level change of less than 0.2 metres.
The police count for the 2011 earthquake stands at 15,901 dead and 2,519 missing
The National Police Agency publishes a one page table for the 2011 Tohoku district Pacific coast earthquake. The version dated 9 March 2026 gives the position at the end of February 2026: 15,901 deaths and 2,519 people still missing.
The table is broken down by prefecture, and twelve prefectures are listed with deaths while six are listed with missing persons. Miyagi accounts for 9,545 deaths and 1,213 missing, Iwate for 4,675 and 1,106, Fukushima for 1,614 and 196. The remaining entries are much smaller, and they reach well beyond the coast facing the epicentre: Tokyo 7, Ibaraki 24 with 1 missing, Chiba 21 with 2 missing, Kanagawa 4, Tochigi 4, Yamagata 2, Gunma 1, Hokkaido 1, Aomori 3 with 1 missing. Adding the prefecture rows ourselves reproduces both printed totals exactly.
One limit of that table needs stating plainly. The sheet prints deaths and missing persons by prefecture and nothing else. It carries no column separating deaths caused by the tsunami from deaths caused by the shaking, so the figures above are the count for the earthquake as a whole.
For how one city on that coast records the same 2011 figures, together with its own climate normals, see Sendai’s coldest month and the 59 cm its snow adds up to.
Sources Checked
Japan Meteorological Agency — page explaining tsunami warnings, tsunami information and tsunami forecasts
Confirmed the target of about three minutes and about two minutes for earthquakes near Japan resolved by the earthquake early warning technique; the three products and their 3 m, 1 m and 0.2 m thresholds; the five numeric bands; the wording giant and high and the blank entry for an advisory; the major tsunami warning being placed in the emergency warning category; the number of forecast zones as 66; the zone arrival time being the earliest in the zone and the statement that water can arrive more than an hour later elsewhere; the thresholds of 1 m and 0.2 m for publishing a measured maximum wave as words; the term weak for a very small value; the first wave could not be identified; the missing observation wording introduced on 24 July 2025; the 100 kilometre rule for offshore stations; and the tsunami forecast issued for a predicted sea level change under 0.2 m.
https://www.jma.go.jp/jma/kishou/know/jishin/joho/tsunamiinfo.html
Japan Meteorological Agency — page on the points of the tsunami warning improvement
Confirmed that a magnitude above 8 cannot be fixed immediately; that the possibility of a giant earthquake is judged from the long period component of the seismic waves and the spread of the shaking; that the largest tsunami envisaged for the sea area is then assumed; that an accurate magnitude arrives about fifteen minutes after a giant earthquake and the wording is replaced by the five numbers; that the published height is the upper value of the band, with 3 to 5 metres published as 5 metres.
https://www.jma.go.jp/jma/kishou/know/jishin/tsunami/kaizen/about_kaizen_gaiyou.html
Japan Meteorological Agency — page on how tsunamis are predicted
Confirmed the tsunami forecast database and the reason for precomputing it; about 1,500 horizontal fault positions, six depths between 0 and 100 kilometres and four magnitudes; the fixed 45 degree pure reverse fault; Green’s law applied at a water depth of 1 metre; prediction points about 15 kilometres offshore; the square root of gravity times depth for travel speed; the 66 zones and the rule that a zone takes its highest prediction point; and the comparison with 139 observed cases giving about 1.2 at tide stations and about 1.8 for zone values; and the date of the tsunami warning improvement, 7 March 2013.
https://www.jma.go.jp/jma/kishou/know/jishin/tsunami/ryoteki.html
Japan Meteorological Agency — boundary table for the tsunami forecast zones
Confirmed the 66 zone names and the areas they cover, including the six Hokkaido zones, the Ariake and Yatsushiro Sea zone covering parts of Fukuoka, Saga, Nagasaki and Kumamoto as the only zone naming four prefectures, the Inner Tokyo Bay zone covering parts of Chiba, Tokyo and Kanagawa as the only one naming three, the three zones covering Okinawa Prefecture, and the date of the municipality mapping table, 26 March 2026.
https://www.jma.go.jp/jma/kishou/know/jishin/joho/tsunami-yohoku.html
National Police Agency — damage table for the 2011 Tohoku district Pacific coast earthquake
Confirmed the publication date of 9 March 2026 and the position as at the end of February 2026; the totals of 15,901 deaths and 2,519 missing; the prefecture rows including Miyagi 9,545 and 1,213, Iwate 4,675 and 1,106, Fukushima 1,614 and 196, Aomori 3 and 1, Ibaraki 24 and 1, Chiba 21 and 2, Tokyo 7, Kanagawa 4, Tochigi 4, Yamagata 2, Gunma 1 and Hokkaido 1; and that the sheet carries no column separating causes.
https://www.npa.go.jp/news/other/earthquake2011/pdf/2026higaijyoukyou.pdf
Last updated: October 2026
