Now that the building and address import is complete, I am moving on to some of the planimetrics that make OSM look good but also contain useful information. Parts of the county have a very good, dense network of streams, traced by a few users, but it can be difficult to spot some streams amidst tree cover on air photos. In these areas where hydrology (streams, rivers, ponds, storm water management, piers, etc) exist, I am proposing to import public domain Baltimore County hydrology, while taking care not to overwrite anyones hard work.
Before I write up a wiki, here are the basic details: 1. I made a model that simplifies the polyline hydrology feature class with a simplification factor of two feet to prevent "overnoding". In experimenting, two feet between nodes keeps the data simple on the map, but typically doesn't result in it looking too jagged. The source data is drawn with a digital pen and then converted to polyline, so any given 10 foot segment might have 500 nodes in the source. 2. I ran Paul Norman's ogr2osm on the result from step one, with a custom translation to convert the various feature types in the source data to something compatible with the OSM key/value data model. Here is that translation: https://gist.github.com/talllguy/52066d13d323f1535f3a 3. Here is the resulting OSM file of that translation: http://1drv.ms/1lLHSIm 4. In JOSM I look for a stream system that is missing. I disconnect the last segment that would intersect an existing water feature already on the map, like a river or coastline. I use the 'select all connecting ways' tool to select the entire system and then copy that to a new layer and download existing OSM data to that layer. (some tags are purposefully translated with no type, so I have to go back and correct them individually and check the data) 5. I run the validator to look for issues. Null names, and nodes that don't quite connect are easy to find this way. 6. Upon solving all validator issues I import the system. Notes: - I'm not going to mess with importing coastline. Our coast is pretty good and there are too many ways to introduce bugs when importing coastline. - Some of the features like boat docks, ramps, and piers may be too complicated for import. Those I'd address individually and may just use a node for leisure=slipway. - For larger rivers where there is a riverbank in the data, I'll use the approved water=* tagging. Questions: We have data on where the coastline is comprised of bulkhead, like where the coastline is unnatural. Is that relevant to the project? I haven't seen a bulkhead type tag but it could be considered a seawall. As a test, I uploaded a small stream system that had not been traced at all, the Sawmill Branch, so you all can see how it'd look on the map. That system is here: http://www.openstreetmap.org/#map=15/39.5217/-76.5440 This is will be a very manual process and as a steward of open data, I will take great care not to undo other mappers hard work. -- Elliott Plack http://about.me/elliottp
_______________________________________________ Imports mailing list Imports@openstreetmap.org https://lists.openstreetmap.org/listinfo/imports