On Thursday 28 April 2011 20:11:39 Ben Supnik wrote: > Hi Y'all, > > This is one sub-basin (01090001 - go Sox :-), isolated from the latest > NHD data, then converted via the latest rule set off of the wiki (with > shapefile key capitalizations fixed). > > http://dev.x-plane.com/download/01090001.zip > > If anyone has done NHD imports before, please take a quick look and let > me know if this looks useful. If so, I can batch the rest of the HUC8s. > > This particular archive contains every layer from the source in both shp > and osm format for comparison. > > cheers > Ben
Hi Ben, I've done a fair bit of importing of NHD so I'll have a fair bit of critique given my prior mistakes that I've had to correct. Here are my opinions: 1. The medium resolution is far too coarse. Use high resolution. 2. You should use one file per attribute. If you are using Ian's shp-to-osm tool, crank up the changeset size to some very large value to accomplish this. 3. Duplicate nodes need to be merged, especially in the flowline files. The data currently have duplicate nodes a every point where one reach intersects another. This creates many duplicate nodes. The Corine data import folks wrote a java tool to remove these (http://wiki.openstreetmap.org/wiki/WikiProject_Corine_Land_Cover/Corine_Data_Import) which I use to remove the duplicate nodes. 4. The conversion rules you are using seem to have a lot of issues. I see some of these on the default rules for shp-to-osm wiki page and I have corrected them where I see them, but others I can't identify their source: a I didn't see any names. If there is a GNIS_name this should be the source for the name. b. The 55800 connectors should have a waterway tag. I typically default them to waterway:stream and then change them to waterway:river if their name has a "River" in it. c. 55600 should be tagged natural:coastline d. 36400 is foreshore and should either be additionally tagged as water:tidal or ignored. e. 46000/46006 should be tagged as waterway:riverbank if it is a polygon f. It would probably be good to have a discussion of how much of the additional NHD/GNIS tags we want to import. I think we should import the FCODE and g. In points and lines files there were many things that had no osm tags including gauging stations and nonearthen shores. We should not import things that have no osm tags. 5. Once we set this up, we'll need to give a fair bit of guidance on what sorts of things to look for. One thing that is quite tricky is importing coastline. One has to make sure that you delete all the previously existing coastline in the area and to connect the new coastline at the edges of its extent. Given the poor resolution of the PGS and the general disagreement between it and USGS on whether tidal riverbanks are coastline or not this can be quite a bit of work. Anyway, good work so far! If you have any technical questions, feel free to email me offlist. James _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us