[NUUG kart] Importing GNS data for Svalbard

Dag Bjerkeli dag at jodalin.org
Tue Oct 7 22:13:52 CEST 2008

Gustav Foseid said the following on 07. okt. 2008 12:24:
> 2008/10/6 Mike Collinson <mike at ayeltd.biz <mailto:mike at ayeltd.biz>>
>     Gustav, thank you for your kind feedback.  All gone now except
>     island names.  May I ask about your experiences with quality?  Was
>     the problem just positional accuracy or did you find substantial
>     errors in the names themselves?
> The main problem was positional accuracy, combined with the large 
> number of small (close to irrelevant) place names. For many of the 
> names we had knowledge, or could easily check, most names were misplaced.
> When we combined misplaced names, with the fact that many names were 
> for very small places like farms, rocks and houses, they were of 
> little or no use. You would need very detailed knowledge to be able to 
> correct these names. I looked at the farm names around where I grew 
> up, and had not heard about half of the names. Trying to correct the 
> position, or even verify that the name was in fact still in use, would 
> be impossible. Another example was a rather small area off the coast 
> of Troms, where around 60 named rocks would have to be verified.
> For administrative names (kommuner), the names in GNS was old and was 
> no longer correct. Other names also had archaic spelling.
> So, we basically ended up with a map where large areas were filled 
> with names we knew were wrong, but with no way of correcting. Keep in 
> mind, that we are acouple of handfulls of mappers, and the GNS file 
> for Norway contains 84.000 names. That is approximately 10.000 names 
> for each mapper to verify.
> Our conclusion is that importing the GNS data would ruin OpenStreetMap 
> as a source of reliable geographic information in Norway.
> For Svalbard, the data looks more accurate. At least there are a 
> number of digits in the latitude and longtitude, but also here the 
> data suffers from unknown quality. There are also a few examples of 
> the same name being given in different languages or spellings (the 
> worst example I found is South Cape Islands, Syd Cap Æerna, 
> Sydkapholman, Sydkapøerne, Sørholmen, Sørkapholmane, Sørkappholmane 
> and Sørkapøyane, which seem to be the same islands). I wish we had 
> known more about what sources are used to collect these data.

I had a quick look at the data for Bjørnøya, south of Svalbard. Most of 
the names there were placed in a clear grid.
With the disclaimer of not knowing the locality of Bjørnøya well, this 
did not look correct.



More information about the kart mailing list