Philly OpenStreetMap Project: the first four months

#mapgoals

Happy 2017 everyone. Another year, another set of #mapgoals.

In August, I set out on a mission to full map the City of Philadelphia on OpenStreetMap with high quality, high resolution data. This is a supremely tall order, but progress has been steady. Frankly, the OSM data of the City has been severely lacking, and open data sources provided by the city are not of very high resolution. My motivation has been to help increase the geospatial capabilities of the city while remaining committed to open source data and software. OSM is an obvious platform for this, as anyone with or without a free OSM account can download the data to their own web maps or GIS and run with it. This has also led to the creation of a new project, the Map Jawn, which will deserve its own introductory post when it’s ready to launch.

I was partly inspired by geo-famous mapper Jonah Adkins and his Paid Mappers project. He wrote a blog post detailing his year of OSM edits back in August.

My project is divided into three distinct phases, each with progressively more useful data attached. While there may be some overlap when parts of these phases are completed, they represent a general outline of the plan going forward.

  • Phase One: Feature delineation and extraction (buildings, land use, waterways, roads, etc.)
  • Phase Two: Points and some attribution (businesses, points of interest, naming, etc.)
  • Phase Three: Advanced attribution (addressing, business hours, contact information, etc.)

I am decidedly in Phase One. There’s a tremendous amount of grunt work here, particularly with proper building delineation. There are open data sources for Philadelphia building footprints, but multiple attached buildings in a row (meaning, a huge majority of buildings in Philadelphia) are considered a single feature and are not broken into their individual building elements. The alternative is parcels, but they don’t really represent actual building footprints. For the sake of high quality, high resolution data, I’ve decided to map each building. Luckily, I discovered the Buildings plugin for JOSM early on, which makes mapping buildings very efficient.

I’m currently using Bing aerial imagery to do the majority of the mapping. It’s several years old now, so it is missing some new changes throughout the city. The City of Philadelphia had high resolution imagery flown in 2015 which is available as a layer in JOSM. I’ve used that imagery to supplement the Bing. I cannot reliably use it as a primary source because it is a traditional orthomosaic rectified to a bare earth elevation model, while the Bing imagery is a “true ortho.” This means that there is a significant amount of building lean in the city’s imagery, making it difficult to use for precise delineation. Building lean is largely removed from the Bing imagery, making it more suitable for top-down mapping.

2016 OSM Updates

That is, only the last four-or-so-months of 2016. I think I began this project on August 17th. I’m not positive about that, but on August 31st I wrote an OSM Diary Entry claiming the 17th sounded about right. Below is a map showing my updates through December 31st, 2016.

The data was retrieved using an OverpassQL query with the QuickOSM plugin in QGIS. The query pulled all features attributed to my username/userID. I then filtered these to only the Philadelphia area in QGIS. In the future, I may specify a specific datestamp range and bounding box to reduce the actual amount of work that I need to do. I then brought these features into TileMill, styled them appropriately, exported them to mbtiles, converted them to traditional PNG tiles using mbutil, and slapped them on the web map above. Note: For some reason I can’t yet explain, the OverpassQL query skips over random vertices here and there, causing errors of alignment and geometry in some of the features. Forgive me and my map for that. I believe that because it’s pulling only my user data, shapes that attach to vertices attributed to other mappers get cut out, but I haven’t tested this. You may notice some odd shapes or bad lines here and there, but the data is fine on the actual OSM database.

The vast majority of my edits were building footprints, accounting for the large number of polygons, thanks in large part to JOSM. As you can probably tell, I did a lot of work in South Philly and a lot of work on Northeast Philly Airport. I live in South Philly and work at Northeast Philly Airport, so forgive me for my bias. My workflow is fairly heuristic so I don’t get bored working in a single area for too long. I’ll go map something new for a while before returning to a sea of row homes. South Philly has become my first checkpoint. I’d like to map everything below Washington Avenue before moving north into Center City and beyond. West Philly will eventually get some love after South Philly is complete too. Point Breeze is the next target neighborhood currently underway. The last days of 2016 were spent mapping up Point Breeze Ave and filling in between there and Broad Street.

A warning to anyone who may be thinking about following this closely: this will probably take years. If you’re interested in helping out, drop me a line and keep me up to date on your progress. I’ll be formalizing contributions through the Map Jawn when it launches.

Highlights

Hey, feel free to take it for a spin on the OpenStreetMap site. The highlights might include Northeast Philly Airport, Passyunk East, and Girard Estate.

Northeast Philadelphia Airport

East Passyunk Ave

Girard Estate

Leave a Reply

1 Comment on "Philly OpenStreetMap Project: the first four months"

Notify of
avatar
Sort by:   newest | oldest | most voted
Nem
Guest

Great work in Philly mapping… it needs to get done, and it’s gotta start somewhere. Thanks for taking a huge initiative!

wpDiscuz