Author |
Message |
09/02/2007 23:45:34
|
dpc_map
Joined: 07/02/2007 04:02:37
Messages: 16
Offline
|
If I enter "Buffalo, NY" into geomanes, I get
----------------------------
Buffalo Bufalo,Buffalo,Бафало,バッファロー,水牛城 United States, New York park
population 279,557 N 42° 53' 11'' W 78° 52' 43''
------------------------------
I looked this record up, geonameid=5110629, and could see that the Feature Class is L and the Feature Code is Park. I was surprised that there isn't a record for Buffalo with a Feature Class of P (ie. some populated place). I assume that most of the US data is coming from from the US GNIS references. I looked up in GNIS and found this record for Buffalo, NY (in GNIS):
973345 Buffalo Populated Place NY 36 Erie 029 425311N 0785242W 42.8864468 -78.8783689 183 Buffalo NW
I imagine it was quite of bit of work mapping GNIS (US) data into the geonames data model, which is more closely related to the NGA data (world-wide), and you had to apply some heuristics in dealing with the feature designations. (I never really understood how my US tax dollars produced these very useful but incompatible data models, GNIS and NGA.)
Can you explain what is happening with this Buffalo example? I've come across some other cities in the US where I was surprised with the representation in geonames. For example, Staten Island comes up as an Island. However, there is a GNIS record for a populated place.
I've been tinkering with GNIS and NGA data for some time and recently come across geonames. org. You've done an incredible job in putting together geonames. Thank you for the great resource.
Dan
|
|
|
10/02/2007 08:24:43
|
marc
Joined: 08/12/2005 07:39:47
Messages: 4416
Offline
|
Hi Dan
Buffalo was updated by an anonymous user. I have changed it back and I have also found and corrected three other places with the same incorrect update. Two out of three were updated by an anonymous with IP-Address 24.205.148.225 and 24.205.161.64 (Charter Communications).
I will block updates by the anonymous users for all US records. The data quality of gnis is good enough and anonymous users are probably not increasing it.
You find an overview of the geonames to gnis feature type mapping here :
http://forum.geonames.org/gforum/posts/list/5.page
For Staten Island I find this record in the newest gnis data set :
942078 Staten Island Island NY 36 Richmond 085 403500N 0740859W 40.5834379 -74.1495875 49 Arthur Kill
It is definitely an Island in gnis.
Marc
|
|
|
|
12/02/2007 16:01:41
|
dpc_map
Joined: 07/02/2007 04:02:37
Messages: 16
Offline
|
This record from GNIS indicating that Staten Island is a populated place doesn't show up in geonames when I search for Staten Island:
1802959 Staten Island Populated Place NY 36 Richmond 085 403500N 0740859W 40.5834379 -74.1495875 49 Arthur Kill
Dan
|
|
|
12/02/2007 16:12:34
|
marc
Joined: 08/12/2005 07:39:47
Messages: 4416
Offline
|
Hi Dan
The record is here :
http://www.geonames.org/5139560/staten-island-subdivision.html
We have not yet updated our database with the newest gnis release. This might explain why the name of the record is not exactly the same.
Marc
|
|
|
|
12/02/2007 22:55:30
|
dpc_map
Joined: 07/02/2007 04:02:37
Messages: 16
Offline
|
Thanks Marc. I think that subdivision parenthetical remark has gone away in the updated GNIS data (I am glad to see that go away).
|
|
|
|