GeoNames Forum

How long is your column 'name' ?

The dump files are in utf8 format, this may explain why you see strange names like the ones you mention.
You have to configure your database to use utf8. In the NL file there are only dutch names and therefore no exotic characters.

Marc

Question :

I have edited and corrected a place name on geonames.org. What happens with my update if a new nga release is loaded? Will my update be overwritten and disappear?

Answer :
Your update will stay untouched.

First we only update nga records if they have a timestamp newer than our last update, and second even if it has a newer timestamp we compare the change history for both update branches and the two branches are merged whenever possible. If it is not possible to reconcile the two conflicting branches the modification by the geonames user takes precedence.
The concept is similar to the 'progressive fork' concept of the wikipedia fork Citizendium.

We are also in touch with nga and send them periodically our updates. It is therefore possible that your change on www.geonames.org makes it into the official nga release.

Question :
We are using geonames data on our website and we would like to have a backlink to a geonames record where our users can move and update a wrong entry. Is this possible?

Answer :
Sure. Geonames is supporting links of the following pattern :

http://www.geonames.org/<geonameId>/<name>.html

where <geonameId> is the integer id of the geonames database entry. You find the id in the database download and in the webservice.
There is also an optional parameter "zoom" if you want the google map to be displayed in an other zoom level (default is 3).

Example : http://www.geonames.org/3123709/eibar.html

If you don't know the name a default page name like '/geonameId/index.html' will also work.

The previous link is deprecated but still supported :
www.geonames.org/maps/geonameId=<geonameid>
http://www.geonames.org/maps/geonameId=3123709

We have updated the geonames database with the latest nga release 2006-02-22.

3355 records have been modified
7584 new records

With this release nga has changed its language codes and is now using 3-letter iso language code instead of using its own codes. Geonames.org has already been using iso 2-letters iso language code. Geonames.org will continue using 2-letters iso language code if available and only use 3-letter language code where no 2-letter code is available.

Code:

 
 countries updated with this data load :
 
  count | countrycode
 -------+-------------
   1295 | CO
    762 | KP
    263 | KR
    212 | RU
    182 | AF
    131 | YE
     63 | IQ
     50 | ID
     43 | PE
     43 | MN
     27 | HT
     15 | CN
     12 | UA
      8 | OM
      4 | NG
      3 | TN
      2 | SL
      2 | IR
      2 | LR
      2 | LB
      2 | MA
      2 | DO
      1 | BG
      1 | QA
      1 | GE
      1 | BZ
      1 | DZ
      1 | RE
      1 | GR
      1 | TW
      1 | MR
      1 | BF
 
 
 countries with new records :
 
  count | countrycode
 -------+-------------
   3697 | CO
   2379 | ID
    756 | KR
    391 | RU
    133 | CN
     58 | IQ
     45 | PE
     35 | AF
     28 | HT
     21 | KP
     17 |
      7 | YE
      3 | UA
      3 | IR
      3 | TW
      2 | LB
      2 | DO
      2 | CF
      1 | QA
      1 | TN
 (20 rows)

We have parsed and extracted geo location from the wikipedia dumps in english and german.

We have thus been able to extract the geolocation of 155'000 Wikipedia entries in 189 languages:
http://www.geonames.org/wikipedia.html

A google maps mashup is here :
http://www.geonames.org/maps/wikipedia.html

Fulltext search over geo referenced wikipedia articles :
http://www.geonames.org/wikipedia-search.html?

The data is also available as webservice : http://www.geonames.org/export/wikipedia-webservice.html

For more information about geo location in wikipedia :
http://en.wikipedia.org/wiki/Wikipedia:WikiProject_Geographical_coordinates

Or in German :
http://de.wikipedia.org/wiki/Wikipedia:WikiProjekt_Georeferenzierung

(edit 2006 March 5)

More generally, the connection between city or village name and position can be flawed. Spatial image browsers such as Google Earth make use of several databases that link place names with latitude and longitude, and unfortunately all such databases suffer from inaccuracies, such as placing Islamabad 100 miles from its true location. All mapping efforts, whether community based or UN supported, would benefit from a stronger feedback mechanism, such as a community filtering system that enables authenticated individuals to update shared geographical data online. For example, we hope that database providers will support initiatives, such as the Geonames Forum (http://www.geonames.org/about.html).

http://www.nature.com/news/2006/060213/full/439787a.html

We have just added a couple of new countries for the postalcode based geocoding service :

# Canada, FSA only (CA)
# Czech Republic (CZ)
# Finland (FI)
# India (IN)
# Croatia (HR)
# Hungary (HU)
# Liechtenstein (LI)
# The Netherlands (NL)
# Pakistan (PK)
# Slovenia (SI)
# Slovakia (SK)
# Thailand (TH)
# South Africa (ZA)

Ich habe heute wiedereinmal den Wikipedia Export runtergeladen und damit rumgespielt.

Die 23'000 (Stand 5 Februar) georeferenzierten Einträge in der Deutschen Wikipedia sehen auf googlemaps so aus :
http://www.geonames.org/maps/wikipedia.html

Die drei Fehler (!) auf der Einstiegsseite habe ich im Wikipedia soeben korrigiert. soviel zum Thema Wikipediaqualität ...

Falls jemand noch weitere Fehler findet, hier die Erläuterungen zur Wikipedia Georeferenzierung :
http://de.wikipedia.org/wiki/Wikipedia:WikiProjekt_Georeferenzierung

Beim nächsten Wikipedia-Export werden diese Fehler dann auch bei geonames korrigiert sein.

Eine Volltext-Suche über die georeferenzierten Wikipediaeinträge ist hier :
http://www.geonames.org/wikipedia-search.html?

Aufgefallen ist mir, dass die Georeferenzierung im Deutschen Export viel homogener und damit einfacher zu parsen ist, als im Englischen Export. Im Englischen Export gibt es unzählige Formatvorlagen die man beim parsen berücksichtigen müsste.

PS, Hier die erwähnten Fehler :
- Einwohnerzahl von "Amiens" stimmt nicht, bzw ist doppelt im wikipedia (einmal richtig und einmal falsch).
- Einwohnerzahl von "Beaucaire (Gard)" stimmt nicht.
- Das "Metropolis-Haus" in Madrid ist keine Stadt und hat schon gar nicht 3 Millionen Einwohner.

Mit dem Update vom 5 Feb. sind die Fehler korrigiert. Es sind nun 24'000 Einträge. (editiert am 7 Feb 2006)

Hallo Sascha

Die Daten von geonames sind in WGS84 gespeichert. Sowohl der Datenbank-Export wie auch die Webservices verwenden dieses Format.

Für die Webservices ist natürlich grundsätzlich auch ein anderes Format denkbar. Beispielsweise, könnte über einen Parameter beim Aufruf festgelegt werden, in welchem Format die Ausgabe erfolgen soll. Wobei die meisten Formate nur lokal beschränkt sinnvoll sind, CH1903 würde nur für Koordinaten innherhalb der Schweiz Sinn machen.

Welches Format wäre dir am liebsten?

Marc

Oooops. Sorry.

Die Koordinaten für diesen Service waren für Deutschland und Belgien vertauscht. Ich habe den Fehler soeben korrigiert, die Koordinaten sollten nun korrekt sein.

Danke für den Hinweis.

Marc

The downloadable country files are a simplyfied version of the database. In particular Alternatenames are normalized.

Hi Marius

First of all: great service!

thanks.

When I use the search option on this page, I get results in a hierarchy like 1. adminstrative division -> 2. adm. div. -> location. (For Norway). When I download the db dump I cannot figure out how to recreate such an hierarchy. I suppose it is not possible...

How do you build this hierarchy and why isn't there a linking field (table self reference) supplied in the db dump?

Cheers!

Marius,
Norway

You find the first administrative dision in the column 'admin1 code' in the dump. It currrently contains the fips code, but we are working on replacing it with the more meaningful iso 3166-2 code. For the time being you find fips and iso codes here :
http://www.geonames.org/NO/administrative-division-norway.html

For Norway the feature distribution is :
geonames_db=> select count(*),admincode1 from geoname where countrycode = 'NO' group by 2;
count | admincode1
-------+------------
613 |
560 | 13
951 | 00
2357 | 06
3168 | 08
7448 | 09
3053 | 15
2961 | 16
131 | 12
741 | 20
2329 | 11
1362 | 02
870 | 01
2038 | 14
5291 | 07
6929 | 18
3329 | 10
2995 | 17
5205 | 05
1647 | 04
1027 | 19
(21 rows)

Hope this helps

Marc

Hallo Daniel

- wer betreibt GeoNames? Vielen Dank an die Betreffenden!

Geonames wir von mir als Hobby betrieben. Ich war für ein anderes Projekt auf der Suche nach 'freien' Webservices. Und weil ich nichts gefunden habe, habe ich selbst damit angefangen.

- welche Technik steckt dahinter? Die Anfragen werden wirklich extrem schnell beantwortet. Habe mir vorhin die komplette Datenbank gezogen und muss schwer arbeiten, um "lokal" überhaupt Vergleichbares hinzubekommen!

Als Datenbank wird Postgres 8 verwendet. Auf Postgis habe ich bislang verzichtet.
Die Volltextsuche wird mit Lucene erledigt. (http://lucene.apache.org/).
Auf der Client-Seite hat es viel Ajax und Googlemaps.

- seit wann gibt's GeoNames -- und wie stehen die Chancen, dass es GeoNames auch in 1 Jahr noch gibt, wenn 10.000x so viele Anfragen am Tag kommen wie jetzt?

Geonames gibt es seit Ende 2005. Wie lange es bestehen wird, ist wie bei allen Projekten, schwer zu sagen. Dies gilt natürlich nicht nur für kostenlose Projekte, auch kommerzielle Projekte kommen und gehen.

Marc

Canadian Geographical Names are available on www.geobase.ca, unfortunately the download is only available in 13'500 files. Theses files are generated on the fly and you will receive an email containing the link to these files (no more than 50 for one download order).

We have gone through this painful process and have downloaded all files in xml format.

All files in a single tar ball are here : http://download.geonames.org/export/geobase.tar.gz (30MB)
The tar-file has a size of 30M and contains around 13'500 files with a total of 300'000 distinct names (around 70'000 are duplicates). The datum used in these files is NAD27.

What is the difference to the normal geonames download for Canada ?
* The datum in the normal download is wgs84. (geobase.ca is nad27).
* duplicates are eliminated
* the feature codes have been converted. (http://forum.geonames.org/gforum/posts/list/9.page)
* the geonames download is a cvs file (geobase.ca is xml)

If you are working with wgs84 the normal geonames download might be better for you.

O' Reilly has an example on how to use amazon's REST webservices with php.

http://www.onlamp.com/pub/a/php/2003/10/30/amazon_rest.html

Let us know if this helps you.

Calling the geonames webservice from javascript gives the Error: uncaught exception: Permission denied to call method XMLHttpRequest.open

Most browsers enforce a cross-domain security restriction. This means your javascript code can only access the server from which the javascript has been loaded. As geonames.org is a different server you cannot directly call the geonames webservice using XMLHttpRequest.

Workarounds :

* use one of our JSON webservices. With JSON you don't have this problem : http://www.geonames.org/export/JSON-webservices.html
* Use a software proxy on your server forwarding the requests to geonames.org.
* use apache modrewrite

Links :

http://nothing-more.blogspot.com/2005/09/xmlhttprequest-security.html
http://spaces.msn.com/members/siteexperts/Blog/cns%211pNcL8JwTfkkjv4gg6LkVCpw%212776.entry

We are pleased to announce the availability of postal codes and postal codes based geocoding for Australia and New Zealand.

http://www.geonames.org/postal-codes/

http://www.geonames.org/export/free-geocoding.html

Geonames.org now includes the geobase.ca database with more than 300'000 Canadian names.

The following feature conversion has been used :

concise term :

Bay = H.BAY
Beach = T.BCH
Cape = T.CAPE
Cave = S.CAVE
Channel = H.CHN
City = P.PPL
Cliff = T.CLF
Conservation area = L.RES
Crater = T.CRTR
District municipality = P.PPL
Falls = H.OVF
Forest = V.FRST
Geographical area = L.AREA
Glacier = H.GLCR
Hamlet = P.PPLL
Indian Reserve = L.RESV
Island = T.ISL
Lake = H.LK
Low vegetation = T.PLN
Military area = L.MILB
Miscellaneous
Mountain = T.MT
Other municipal/district area - major agglomeration = A.ADMD
Other municipal/district area - miscellaneous = A.ADMD
Plain = T.PLN
Province = A.ADM1
Rapids = H.RPDS
River = H.STM
River feature = H.STM
Sea = H.SEA
Sea feature = H.SEA
Shoal = U.SHSU
Spring = H.SPNG
Territory = A.TERR
Town = P.PPL
Unincorporated area = L.AREA
Valley = T.VAL
Village = P.PPL

generic term :
=========================

Point.Cape = T.PT
Cove (2).Bay =H.COVE
Pond (2).Lake = H.PND
Locality.Unincorporated area =P.PPLL
Hill.Mountain =T.HLL
Head (1).Cape = T.HDLD
Pointe.Cape = T.PT
Peak (2).Mountain = T.PK
Islands.Island =T.ISLS
Lakes.Lake = H.LKS
Hills.Mountain = T.HLLS
Bureau de poste.Unincorporated area = S.PO
Colline.Mountain = T.HLL
Lacs.Lake = H.LKS
Inlet (3).Bay = H.INLT
Reef.Shoal = H.RF
Pass (2).Valley = T.PASS
Ledge (1).Shoal = U.LDGU
Marsh.Low vegetation = H.MRSH
Marais.Low vegetation = H.MRSH
Pond (1).Lake = H.PND
Étang.Lake = H.PND
Fiord.Bay = H.FJD
Peninsula.Cape = T.PEN
Harbour (1).Bay = H.HBR
Bight.Bay = H.BGHT
Gully (2).Lake = U.GLYU
Ravin.Valley = H.RVN
Gullies (2).Lake = U.GLYU
Mountains.Mountain = T.MTS
Dam.Miscellaneous = S.DAM
Reservoir.Lake = H.RSV
Peaks.Mountain = T.PKS
Parc public.Conservation area = L.PRK
Pont.Miscellaneous = S.BDG
Ridge (2).Mountain = T.RDGE
Provincial Park.Conservation area = L.PRK
Ponds (2).Lake = H.PNDS
Bog.Low vegetation = H.BOG
Canyon (2).Valley = T.CNYN
Arrêt ferroviaire.Unincorporated area = S.RSTN
Ponds (1).Lake = H.PND
Barrage.Miscellaneous =S.DAM
Portage.Miscellaneous =R.PTGE
Drain (2).Miscellaneous =H.CNLD
Camp saisonnier.Miscellaneous =S.CMP

Question :
The geocoding webservice does not work with special character like ö or blanks.

Answer :
The parameters for the webservices have to be utf8 url encoded. If you call the webservice with "New York" you have to url encode "New York", otherwise our service will search for "New" and all other parameters afterwards get lost.

"New York" becomes "New%20York" and instead of ö use %C3%B6.

URL Encoding replaces spaces with "+" signs, and unsafe ASCII characters with "%" followed by their hex equivalent. Safe characters are defined in RFC2396. They are the 7-bit ASCII alphanumerics and the mark characters "-_.!~*'()".

Check the documentation of your programming language to see how url encoding is done in your programming language.

Java example :
java.net.UrlEncoder("ö","UTF8");

Perl :
use URI::Escape;
$escapedParameter = uri_escape_utf8("new york");

javascript example :
encodeURIComponent("ö")

(Note that the standard JavaScript escape and unescape functions operate slightly differently: they encode space as "%20", and treat "+" as a safe character.)

Pyton :
Nicolas Laurance has written a client library for geonames :
http://www.zindep.com/blog-zindep/Geoname-python/

Hi

We have separated the postal codes from the geographical names as for many names (mountains, lakes, etc) its impossible and it does not make sense to use a postal code.
It is easier for us and for our users to separate general names from postal code names. Postal codes are therefore in a separate dump file, which is currently not available for general download as we don't want to feed search engine spammers. (The ones creating thousand of pages with place names and having google ads as only content.)
We believe for 'serious' applications the use of a webservice makes more sense. Drop an email to info@geonames.org if you need the postal code dump file. (And if you are not a spammer

)