I am looking for sources of geo-referenced data on villages and cities (or all other things there are) in The Gambia (and Senegal if available). One issue with geocoding in rural West Africa is the varying spelling of village names.

I own a (possibly incomplete) list of some 2000 Gambian village names, I was able to acquire some 1000 coordinates from HERE maps, Google maps, and MapQuest, through the respective APIs. But I estimate that this is missing roughly halve of the villages in The Gambia. What is more. Some of these one thousand are rather bad/uncertain matches, so it would be good to compare with additional sources.

Are there any other sources, with a decent level of detail and fairly up-to-date?

Here is a list of villages that I was not able to find with any of the above sources. Since I know for sure where they are located I use those as a benchmark:

(I am aware that some of these actually have labels with different spellings/entirely different names in google maps. However, this is so far of no help to me though, as these spellings mostly nowhere close to how the village names are usually spelled/pronounced)

Clarification in response to some comments/answers: geonames for Gambia is in most cases rounded to minutes, which in the case of Gambia gives a grid in which places are as far as ~1.3kms, away from where they actually are. In the case of Gambian villages which are densely and rather uniformly distributed, this is not very helpful. If there are ways to obtain unrounded geonames data I'd be happy.

  • geonames.org/search.html?q=gambia&country= would be the obvious answer, so I'm probably missing something. The data appears to be rounded to the nearest second. – Barry Carter Aug 7 '15 at 13:25
  • 1
    By the way, is your dataset of 2000 Gambian village names available somewhere? It could help test possible solutions. Cheers! – Nicolas Raoul Aug 17 '15 at 5:29
  • It stems mostly rom the Gambia Economic Census 2006. The "village"-variable there had ~1900 labelled values. When you Google it you should find it, I don't have the link at hand right now – sheß Aug 17 '15 at 7:40
  • Now I've found the link. You find it e.g. on gbos.gov.gm this contains village name and the standard village ID used by GBOS. The first two digits of the ID also indicate the district. – sheß Aug 18 '15 at 2:06

You can find the information you are looking for (cities, villages, populated places) with their local and english name and coordinates at the US National Geospatial Agency's (NGA) Geographic Name Server (GNS) for foreign countries. They have a dataset per country, in some countries the numbers are however rounded and have lower precision than in others:

http://geonames.nga.mil/gns/html/namefiles.html

Also, I converted all the datasets from last year into a linked CSV format. You might find this format easier to digest into a database (e.g., ETL).

http://www.opengeocode.org/cude1.2/NGA/GNS/index.php

The Linked Vocabulary definitions are here:

http://www.opengeocode.org/cude1.2/LinkedCSV-Vocab.php

  • This is great data, but if I'm not mistaken it's rounded to minutes. Which can correspond top more than one kilometer in the case of The Gambia. The distance between two villages in The Gambia is often much smaller – sheß Aug 6 '15 at 16:35
  • Is that not the case? Please clarify @Andrew – sheß Aug 18 '15 at 2:15
  • I checked the NGA site for clarification. The only reference I see to precision is on the web page describing the format layout. I've included a link to it. It appears from the description, that the field precision supports seconds. The geographic centroid is the geographic center for the location. I would think it would be unlikely that they would map two different locations to the same centroid. – Andrew - OpenGeoCode Aug 18 '15 at 16:14
  • They might not mention it there, or you might find it unlikely that they would map two different locations to the same centroid... but taking a brief look at your linked data shows: (1) most coordinates look like: 13.566667,13.466667, 13.5, 13.766667, which is clearly indicative of some rounding being applied at some level. (2) places like Wessedung & Bulengart or Mayork & Kanjending are mapped to the same point.The latter is likely a consequence of the former, as both pairs of places are very close together. – sheß Aug 19 '15 at 7:54

Have you taken a look at Open Street Map (OSM)? The data are downloadable and it is always being updated. If you happen to have more detail than OSM, you could always contribute to the data set.

  • The data geocoded from mapquest should be OSM data. So I'm not expecting much more by downloading the dataset in full. Not sure whether I should share the data gathered from Google and HERE with osm, I doubt they would approve. – sheß Aug 6 '15 at 16:25

I find Genomes (geonames.org) as currently the most comprehensive source still for town/village coordinates in that region of Africa. A lot of work has been added as part of the International Aid Transparency Initiative (IATI)’s geocoding element that has contributed to geonames.org’s more comprehensive coverage.

  • Thank you for your answer. Read however what I wrote under Clarification in response to some comments/answers. This applies to what you write, or am I getting you somehow wrong? – sheß Sep 1 '15 at 12:44

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.