China datums

Datums in China

Please see e.g., http://www.asprs.org/a/resources/grids/05-2000-china.pdf and https://www.iag-aig.org/doc/5d8dd34ba63c6.pdf (reports #1, #3). On this page we aim to just supplement the more widely known information.

Transformation for PRC datum are only legally acquired from the local government, which should normally be your source for maps made in old datums too.

• CGCS2000 (EPSG:4490) is the modern datum. For casual use, assume it's close enough to WGS 84. For more precise use, it is referenced to ITRF97 at epoch 2000.0 and you can use that for conversion. There are later, better parameters such as Wang et al. (2019), Table 1 (frame) and the official recommendation of CPM-CGCS2000 (plate) + usual ITRF97 frame, defined in CH/T 2014-2016.
• Xian 1980 is still commonly used. Papers dealing with its conversion are heavily redacted.
• For Beijing 1954 (EPSG:4214), the EPSG has a few local transforms to try, centered on some oil-producing areas.

Possible Xian 1980 sources

This section is for those who really want to put a TOWGS84 on Xian 1980 once and for all.

• The publication doi:10.32629/gmsm.v2i4.291 has three pairs of XYZ points in Xian 1980 and CGCS2000. The redaction is sloppy enough to have the top pixels show through. If anyone wants to try, use the PDF archived at [1] (click "Download Zip").
• In the GCJ-02 tradition, open-source code often inadvertently leak classified information. GitHub: GNSSAMS (forked for preservation) provides six pairs of XY points in Xian 1980 and Beijing 1953.
• SuperMap/iDesktop-MainDocs (also forked) provides a dozen pairs of tightly clustered XYZ in Xian 1980 and CGCS 2000, 3° zone 36.

Coordinate systems used in China

Web map examples

Some common coordinate systems observed, 觀測中國電子地圖諸網站常見座標系統，

Here we give examples of how a single longitude latitude pair is located differently depending on which web mapping site one uses!

This is necessitated by the state mandated distortion... 其政府所規定之偏移…

1. http://maps.google.com/maps?q=loc:40,116.5&t=k
2. http://maps.google.com/maps?q=loc:40,116.5&t=m
   • http://ditu.google.com/maps?q=loc:40,116.5&t=k
   • http://ditu.google.com/maps?q=loc:40,116.5&t=m
3. http://api.map.baidu.com/staticimage?markers=116.5,40&zoom=14

We observe we can obtain the exact same three points,

• http://map.sogou.com/#hb=1,1&c=12968861,4838218,15&tip=1,12968861,4838218
• http://map.sogou.com/#hb=1,1&c=12968189,4838039,15&tip=2,12968189,4838039
• http://map.sogou.com/#hb=1,1&c=12967462,4837158,15&tip=3,12967462,4837158

via 搜狗 Sogou API's type=1, 2, 3. We also notice type=1 is a plain Mercator projection, Clarke 1866 ellipsoid, a=6378206.4 rf=294.978698213898 .

$ proj +proj=merc +ellps=clrk66

(Note not all sites are actually reachable from inside China.)

See the Wikipedia page on GCJ-02 for how and why the map distortion is done. On top of messing with the lat/lon, each site uses their own variation of Mercator for tiling, which is less than convenient. cntms is a Python map tile proxy with support for some Chinese sites: you will see that Baidu also uses Clarke 1866 Mercator.

See also

• Taiwan datums