January 7, 2024, marked my first OpenStreetMap edit in years, and the beginning of a journey that I was not aware I was embarking upon.

After about a month of mostly adding driveways and doing minor road corrections around Oregon’s Lincoln County, my home county, I started a systematic review process for United States Forest Service roads in the Siuslaw National Forest. From early February to early June, I reviewed, added, and standardized the tagging of 1,055 such roads, and corrected the alignment of 963 of those. This amounts to a little over 1,145 miles of road geometry manually corrected. There are still many USFS roads left in the SNF to review, as my target review area was south of Neskowin Creek to north of the Siuslaw River. I intend to get to those eventually, but I decided to pause this and move on to features more local to myself.

I noticed that accurate address data is virtually non-existent in Lincoln County. Google Maps has good address coverage, but is littered with duplicates, errors in data and placement (sometimes quite significant), and the occasional missing address. Bing Maps suffers from much of the same. The county’s own assessor map only has address data for taxlots, not for individual structures, and some of this data is incorrect (a 5### address near the middle of the 4000 block, just to name one example that I can recall). No one seems to have consistently reliable structure-level address data, aside from the state of Oregon address geocode service, which only offers lookup of a single address, not any kind of bulk data request or export, and also has occasional placement errors. I wanted to change that.

I exported taxlot address data from the county map and online property search tool, made corrections and filled in gaps with my own research, and put this data into a spreadsheet. I was able to use a function to automatically query and collect the coordinates returned by the aforementioned geocode service for each address, and then convert this into a shapefile. I used Oregon Transportation Network data to verify and correct the address data and placement relative to roads, then used county taxlot polygons, open building footprint data, and imagery such as Bing and NAIP to correct placement errors. After all this, as a final quality check, I randomly selected and verified various addresses around the county, and used my personal knowledge to review what I knew as well. Through open street level imagery providers like Bing, as well as physical visits to areas like campground and mobile home parks, I was able to manually add over 6,000 sub-addresses, with no help from the state geocode service, as it does not return sub-addresses. I also developed a workflow to identify both new and removed addresses from the county assessor map. At this time, it is likely that I possess the most accurate address point data for Lincoln County. I feel that it is important to put this data somewhere where it can be both helpful and easy to access.

I had originally considered going through the OSM import process for the address data, but what stopped me from doing this was the aforementioned building footprint data. I noted during the address placement review that every building footprint dataset covering Lincoln County is full of all kinds of errors, and I also noticed that OSM is severely lacking in building footprints locally. What better way to ensure the data I am contributing is free from error than by manually verifying every single addition? Why not add and correct building footprints at the same time? Though incredibly time consuming, doing it this way has produced valuable results.

Starting on June 15, I began the work of adding these addresses and building footprints, and making fixes to other OSM data along the way. Today, November 23, 2024, I have added every address in the 97498 ZIP code to OSM, except for maybe 5 that were already present. This amounts to about 2,350 addresses. I am not sure of an easy way to pull this kind of information, so I will roughly estimate having added 1,800 buildings, and modified 700 pre-existing buildings.

I intend to continue this level of work north through Seal Rock. As of now, that means 6,546 addresses and probably 5,000 or so buildings waiting for me to add. I have been monitoring the area for other active users, but find none outside of the rare editor making spot edits as they travel through or remotely participating in MapRoulette challenges. I doubt I will be committed to this level of work throughout the rest of the county, and will probably move towards an import of the rest of the address data when that time comes.

I doubt anyone will read this screed, but if you have, I hope that means you are interested in helping out. Maybe it means you want the address point data. Send me a message either way.