Talk:New Jersey

From OpenStreetMap Wiki
Jump to: navigation, search

NJ LULC data

Some of the pieces of the NJ LULC data do not show on OSM. I tested one failed polygon and found that a node number range, somewhere in the 492s, is not being accepted. CrystalWalrein 00:35, 19 September 2009 (UTC)

Due to the size of the changesets, the data will take a while to show up in Mapnik and Osmarender. If you switch between the two, you will see where Osmarender has rendered and Mapnik not, and vice versa. The polygons will show up when editing in Potlach or JOSM, as they have been uploading to the DB server, they just have not yet been rendered. If you can post a link to a polygon that is not showing up in either, that would be helpful. - Johnjreiser 14:42, 20 September 2009 (UTC)
Galloway Township. Pomona Golf Club does not show completely or at all, as do many forest parcels. CrystalWalrein 16:56, 20 September 2009 (UTC)
I've also found many shapes missing from the NJ LULC dataset import. For example: Part of the reservoir in Harrington Park, NJ was imported but the much larger part (to the southwest) is missing. The missing area is present in the LULC dataset and, other than the "acres" tag, is tagged identically to the part that was imported. In the same area, other water bodies coming off the reservoir are also missing. Olek Lorenc 12:06, 21 September 2009 (UTC)
Another example: Compare this area to LULC in ArcExplorer. Many of the lakes are missing (the waterbodies layer is helpful for highlighting lakes). Oleklorenc 02:00, 21 September 2009 (UTC)
The LULC layer for the entire state is over 800,000 polygons and the line work is incredibly complex. The API limits ways to 2,000 nodes, so I simplified the layer using ArcInfo. Only the most very large and complex polygons were split into multiple ways and made into multipolygons. The layer also had to be split into 75 chunks and uploaded individually. Where the LULC layer does not appear on the other side of a major roadway is intentional. The layer was split geographically, which is why the LULC chunks that have been rendered stop at major roadways. Those geographic chunks are not necessarily close to one another - I had a script group all of the pieces into 75 files by looking at adjacency, limited by the number of nodes, ways, and relations that could be uploaded at one time (50k).
I still have the GIS data and 75 OSM pieces. I'll check out the areas both of you highlighted and find out which of the 75 files they were in and then double check to make sure it was properly uploaded. FYI - When you look at NJDataUploads' changesets, you'll see a two digit number in parentheses denoting which of the 75 pieces that changeset is. (Those changesets with (JOSM) in the comments were once uploaded using JOSM instead of bulk-upload.py.) Both of the pieces you mentioned could be in the same file that I might have missed. I'll let you know later today.
Looks like chunk 12 was not properly uploaded, reverted, and then never reuploaded after being fixed. I'll upload it later today. Both areas Olek identified are in chunk 12. I still need to look into Pomona, the link posted earlier takes me to Florida. If you come across any others, please let me know. - Johnjreiser 14:57, 21 September 2009 (UTC)
Thanks John. Looks good. I'll make may way through NJ comparing the LULC data to see if there are any other areas that appear to be missing geometry. -Oleklorenc 05:45, 22 September 2009 (UTC)