Over the last couple months I’ve been busy fixing mapping in North Andover, MA, where I currently live. Issues with the mapping here in Massachusetts have largely been because of a lack of addresses, street data, business data, and stagnant notes left on the map for upwards of 4 years or so.
I’ve been very active with a couple projects as of late:
This project is progressively going through all parcels within the town and manually adding each address to the building, plus adding driveways, and assigning building=garage or shed as necessary. Plus, drawing in sidewalks as separate ways and correcting intersections and turn restrictions as necessary. It’s a slow process, because there’s so much to verify and add, and I’m doing this process largely on a block-by-block process, alone. This has gotten massively easier lately, however, as I’ve learned how to use JOSM to some extent, and discussion started about a Massachusetts wide address import, of which I provided details on how to do such an import using JOSM and the Conflation plugin. (Details of the current import effort are documented here: https://wiki.openstreetmap.org/wiki/Import/Catalogue/MassGIS_Addresses) Let’s delve deeper into this…
The Massachusetts Basic Address Points is a database divided by town with POI / point data encoded at the center of building polygons for addresses, however the data for streetnames and housenumbers have tags that need renaming, and all the streetnames are in upper case. We had lots of talk on the talk-us-massachusetts list and discovered several issues:
1. Is the title() function appropriate considering the possibilities for false positives, and how many such positives would there be (McCoy Street, YMCA Way).
2. How would you handle data conflicts with multiple addresses to the same building.
3. How would you split up the data import so that it doesn’t cause havoc on the server?
My opinion on the title() function was that the number of double capital items that would be made in error would be small enough to still warrant importing all address data, reliant on persons to manually or semi-manually edit such features at a later date. I believe we came to an agreeing consensus, but discussion is still taking place on this.
I suggested that data conflicts could be manually reviewed. If you have 36 Howe Street, and the import wants to put in 38 Howe Street next to it, you could pull up the MassGIS parcel map, check which neighboring parcels are officially using which addresses, and in this case, merge the two into addr:housenumber=36;38. Note, I wouldn’t use the hyphen symbol, as it has the connotation that every number between the two are in the same building, e.g. 36, 37, and 38 Howe Street, which may or may not be the case. Discretely listing each one as it’s own number separated by a semicolon is much cleaner.
While the community continues to stir on the subject of how to execute a massive import of the data, I realized that I could use this address data to simplify my own small scale edits. By loading the address import database for my town in with my edits, I can conflate the import data, do a manual verify for each address using the L3 Parcel Map, move buildings, draw addresses, and complete it all in one fell swoop. I thoroughly reviewed the OSM Wiki on whether doing small scale imports with manual verification are an issue, and apparently it doesn’t qualify as an automated edit, and not necessary of a full import documentation, since it’s all verified personally for each and every edit being made. It’s still a slow process, but verifying matching data is a lot faster than clicking and entering numbers ad nauseum. It’s progress.
As such, the North Andover address and general fix-up project is about 20% complete, with all addresses west of MA Route 125 being completed and verified. I really wish I knew how to implement a tasking engine to make this easier…
I’ve also being doing detail/micromapping of very specific locations within the town. Specifically, Merrimack College, large shopping areas, or public schoolgrounds. I read through the information contained on the wiki, and basically came to the conclusion that micromapping is fine, but if done in specific circumstances that could warrant it. I feel as if it’s warranted for locations where the land features are managed, such as planted trees, parking lots, and public areas. It helps to create detailed views of these public areas so people know exactly what it will be like there.
I’ve been drawing grass, adding trees, walkways, and parking lots to make such locations as detailed as possible.
I’ve been listening to the SOTM talks while at work, and discovered one talk about turn based and directional assistance mapping within OSM. Apparently some of the modern tools being developed are using front facing camera data to assign turn lanes, and using these tools you can easily add such data into OSM. I plan on exploring this some more, but for those curious, you can see the talk here: https://www.youtube.com/watch?v=MEml0vO3qvM
As always, I’m learning new things about OpenStreetMap every day, and every little bit of mapping I do helps to learn new tips and tricks about mapping more effectively and in a way to make a better map for others. JOSM is great for power mapping and toolset use, but iD still provides a dirt-easy way to contribute.