On 3 Dec 2009, at 9:08 , Ian Dees wrote: > On Thu, Dec 3, 2009 at 11:01 AM, Thea Clay <t...@cloudmade.com> wrote: > Hi guys, > I am so excited that more land use imports are in the works. They make such a > huge visual difference. Check out the border between a state with the import > complete and one without: > http://www.openstreetmap.org/?lat=34.896&lon=-85.408&zoom=9&layers=B000FTF > > Although the lower-detail zooms might look marginally better (I would say > they're way too cluttered right now...), the import was not done very well > for two main reasons: > > 1. All of the areas were imported with overlapping edges. This means there is > *tons* of duplicate information in the database. I learned my lesson with the > counties import: overlapping edge imports like this should be broken apart > and use relations for the borders. > > 2. The resolution of the landuse information is very low. If you zoom in and > use Potlatch to see what the aerial images look like, you can see that in > most cases the polygons don't come close to matching the actual landuse. in > the future, we should make sure that imports are high-enough resolution to be > useful in our datasets. 1:24k is the minimum and even that is not useful in > some cases. > > I'm not trying to belittle the effort, I just want to make sure we don't > repeat the same mistakes on other huge imports like this. >
These are very important points and I don't want to repeat it. It's like tiger import. It was the best approach at that time but from all we learned imports should be done better in future. a bit planning upfront can save all the time spent to cleanup. But it's also possible to overdo the high res requirement. The MassGIS import didn't remove redundant points and the data is just huge without any benefit. _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
