How city area is measured in Japan

What "city" means in Japan

Japan's national area statistics are published by GSI in a table of prefecture and municipality areas (全国都道府県市区町村別面積調), updated every quarter. The table has two levels: prefectures (都道府県) and, under each prefecture, municipalities (市区町村). Tokyo is a prefecture, the Tokyo Metropolis (東京都). There is no municipality row named Tokyo in the table. Instead, GSI gives three subtotals inside the Metropolis: the 23 special wards (区部), the cities (市部), and the towns and villages grouped by subprefecture and district (支庁・郡部). The Tokyo Metropolitan Government's own table uses a similar breakdown, and lists the islands (島部) as a group of their own.

So "Tokyo" can mean two quite different areas. The Metropolis, at 2,200 km²^1,2, includes cities, towns, villages and the outlying islands. The 23 special wards, at 627.5 km²^3,4, are one part of it. The Tokyo Metropolitan Government's population estimate for 1 January 2026 puts 14,270,748 people in the Metropolis and 9,944,992 in the wards.

Outside Tokyo, a city (市) is a single municipality, and the name alone says little about its size. Designated cities such as Hamamatsu are divided into wards, which GSI lists under the city total, written as (浜松市)中央区; for Hamamatsu GSI lists three, Chūō, Hamana and Tenryū. In GSI's figures, the whole city of Hamamatsu covers 1,558 km²^5,6, more than twice the area of Tokyo's 23 wards combined. The city itself states that it ranks second in Japan by area, after Takayama in Gifu Prefecture, and gives GSI as the source of its figure.

Why our map may show a different number

When we have no stored boundary for Tokyo, the app sends OpenStreetMap's Nominatim service the search "Tokyo, Japan", restricted to cities. On 17 October 2026 that search returned OpenStreetMap relation 1543125, displayed as "東京都, 日本". It is an administrative boundary at prefecture level, and Nominatim labelled its address type "province". In other words, the search returned the whole Metropolis, not the 23 wards. When no boundary is stored and the result is a relation with its own polygon, the app uses that polygon directly and does not mark it approximate.

When we have no stored area for a boundary, the map calculates the area from the polygon, not from GSI. For Tokyo the two are far apart. When we fetched relation 1543125 on 17 October 2026, its polygon took in wide areas of open sea around the outlying islands, and the area we measured from it was about 19 times the official 2,200 km²^1,2. GSI measures land only up to the high-water shoreline. An area calculated from that polygon should therefore not be read as the size of the Metropolis.

We found no single OpenStreetMap boundary for the 23 special wards, so we do not draw them as one shape. Each ward is its own relation. For Chiyoda ward, relation 1761742, the area we measured from the OpenStreetMap polygon on 17 October 2026 is within three percent of GSI's 11.7 km²^7,8.

A search for "Hamamatsu, Japan" returned relation 4674741 ("浜松市, 静岡県, 日本"), with address type "city". When no boundary is stored, it too is used directly and not marked approximate. When we fetched it on 17 October 2026, its polygon extended offshore past the coast, and the area we measured from it was about a third larger than GSI's 1,558 km²^5,6, so we do not draw it on this page.

Problems getting boundaries in Japan

The first problem is undetermined boundaries (境界未定). GSI marks the area of any municipality whose boundary is not fully settled with "（参考値）", a reference value, and says the area is then an approximate figure given for convenience. In the 1 July 2026 edition, five of Tokyo's wards carry the mark: Chiyoda, Chūō, Minato, Katsushika and Edogawa. So do the 23-ward total, the Metropolis total and the city of Hamamatsu. The table does not say which boundary sections are undetermined, and we have not sourced that.

The second problem is reclaimed land in Tokyo Bay. The 23-ward total includes three areas that belong to no ward: the Arakawa river mouth (荒川河口部), the outer Central Breakwater landfill (中央防波堤外側埋立地), and the New Sea Surface Disposal Site (新海面処分場), which covers 2.4 km²^9,10. GSI's notes say the river mouth is excluded from the Kōtō and Edogawa ward areas but counted in the special-ward total, that the breakwater figure leaves out the part already covered by a landfill notice, and, in the notes for earlier editions in the same file, that the disposal site's affiliation is undecided (所属未定). The Tokyo Metropolitan Government also states that the ward total includes all three. Adding up the 23 ward areas therefore does not reproduce the ward total, and a total built only from the ward areas leaves these areas out.

The third problem is where land ends. GSI measures the area enclosed by the coastline and municipal boundaries on its national digital basemap (電子国土基本図). It takes the coastline at high water, counts rivers and lakes as land, and closes river mouths with a line joining the tips of the two banks. OpenStreetMap boundaries can follow other lines, as the Tokyo and Hamamatsu polygons show.

Finally, figures change. GSI rounds each published unit to two decimal places, so a prefecture total may not equal the sum of its municipalities. Hamamatsu reorganised its wards on 1 January 2024, according to the file header, and its city total changed between the 1 January and 1 April 2026 editions. The Tokyo Metropolitan Government's table is dated 1 October 2025, and gives the same values as GSI's 1 July 2026 edition for all four Tokyo figures on this page.