GeoTIFF DEM converted from 10m / 5m / 1m resolution DEM by Geospatial Information Authority of Japan (GSI). Approval for Reproduction pursuant to the Survey Act, granted by the Director-General of the Geospatial Information Authority of Japan (R8JHf51).
When reproducing this product in order to make it available to the general public, approval from the Director-General of the Geospatial Information Authority of Japan must be obtained.
Digital Elevation Model (DEM) data for Japan, converted to GeoTIFF from the Geospatial Information Authority of Japan's (GSI, 国土地理院) 基盤地図情報 (Fundamental Geospatial Data) DEM distribution. Available at three resolutions: 10m, 5m, and 1m.
Published by the UN Smart Maps Group. Generated by
optgeo/japan-geotiff-dem,
an open pipeline that converts GSI's zipped GML DEM tiles into GeoTIFF
— see that repository for the code, and for exactly how each file was
produced.
One GeoTIFF per GSI source mesh, grouped by resolution:
10/ — 10m resolution DEM5/ — 5m resolution DEM1/ — 1m resolution DEMFormat: GeoTIFF, single-band FLOAT32 elevation (meters), EPSG:6668 (JGD2011 geographic), ZSTD-max compressed. Filenames follow GSI's own mesh-code naming directly, so any given file's coverage and survey vintage is traceable back to GSI's own 基盤地図情報 catalog.
Coverage grows incrementally and unevenly across resolutions and prefectures — this file won't always describe the exact current extent; browse this product's file listing for what's actually present right now. In particular, the 1m tier is being updated prefecture by prefecture as GSI adds newer aerial-laser survey coverage, so different meshes can carry different survey dates even within the same prefecture.
GSI republishes an updated survey for a mesh as a new file rather
than overwriting the old one — so a given grid cell can have more than
one file in this bucket, at different survey dates (see the filename's
trailing YYYYMMDD). To avoid guessing which one is current, each
resolution prefix carries two manifests:
{res}/latest_file_list.csv.gz — one row per grid cell, the current
file for every cell in that resolution tier.{res}/obsolete_file_list.csv.gz — one row per file, every
superseded file (still published, just no longer current).Both are gzipped CSV with header url,size,md5 — size in bytes,
md5 as a lowercase hex digest (each object's own S3 ETag, which is a
true MD5 for these files). This lets a downstream client check whether
it already has the current file for a cell without any network request
beyond fetching this one manifest — compare local file size (and hash,
if you want to be sure) against the row, no per-file HEAD/GET needed.
Regenerated whenever new data lands; safe to re-fetch any time you want
the current split.
Direct URLs (Source Cooperative's own browse UI makes these tedious to find by clicking through, so they're spelled out here):
SYNOPSIS (decompress and read line-by-line without saving to disk):
Source data: Geospatial Information Authority of Japan (国土地理院), 基盤地図情報 DEM.
測量法に基づく国土地理院長承認（複製）R8JHf51
不特定多数の者が提供を受けることができる状態に置く措置をとるために本製品を複製する場合には、国土地理院の長の承認を得なければなりません。
Approval for Reproduction pursuant to the Survey Act, granted by the Director-General of the Geospatial Information Authority of Japan (R8JHf51).
When reproducing this product in order to make it available to the general public, approval from the Director-General of the Geospatial Information Authority of Japan must be obtained.
4929/
4930 area (109 mesh cells, 7,485 files) found 3 more 1m files with
the same silently-wrong-data signature, all within mesh 492963
(previously unchecked) -- re-converted correctly and re-uploaded;
independently re-verified against the raw source.
latest_file_list.csv.gz regenerated. The remaining 7,482 files
checked in this sweep came back clean.4929/4930 area
flagged below (silently published as 0% valid data despite real
elevation content), re-converted correctly and re-uploaded;
independently re-verified against the raw source.
latest_file_list.csv.gz regenerated to reflect the corrected
checksums. This closes out the correction work for the originally-
suspect 10 mesh cells; a broader sweep of the surrounding area for
any further, previously-unsuspected corruption is still ongoing.4929/4930 (Nagasaki area),
previously published with silently or overtly corrupted pixel data
due to a bug in the conversion tool, re-converted correctly and
re-uploaded; independently re-verified against the raw source.
latest_file_list.csv.gz/obsolete_file_list.csv.gz regenerated to
reflect the corrected checksums. A small number of additional files
in the same narrow area may still need similar correction; this will
be recorded here once resolved.latest_file_list.csv.gz/obsolete_file_list.csv.gz
regenerated: 5m now 471,062 files (422,119 latest / 48,943 obsolete),
up from 378,618 (all latest) before this cycle. 10m unchanged (no
relevant GSI update since 2024).latest_file_list.csv.gz/
obsolete_file_list.csv.gz regenerated for all three resolution
tiers: 1m 308,855 files (291,779 latest / 17,076 obsolete), 5m
378,618 files (all latest), 10m 4,981 files (all latest).latest_file_list.txt.gz/obsolete_file_list.txt.gz
manifests renamed to .csv.gz, adding / columns (see
"Finding the current file per grid cell" above) — the old
objects were removed from the bucket, so any link to a
manifest from before this date is stale.File an issue at
https://github.com/optgeo/japan-geotiff-dem/issues, or see that
repository's README.md for other contact paths.
sizemd5.txt.gz.txt.gzlatest_file_list/
obsolete_file_list manifests (then .txt.gz, since renamed
.csv.gz the next day — see above) added for all three resolution
tiers.