Mapper since: June 07, 2014

I am learning the format for completeness and using rules based on geometry to help provide visual consistency. Following this is what I have observed to be the most relevant way to map the area in which I live, as each area may vary by feature type. If you see a mistake, please, let me know.

  • highway=stop points at the stopping line (for conveying how far from the intersection a vehicle must stop) with direction=*.
  • highway=traffic_signals also points at the stopping line and before any crosswalk with traffic_signals:direction=*.
  • traffic_sign=stop represents the sign itself (for 3D modeling) with direction=*.
  • parking:lane=* excludes amenity=parking (usually for street parking).
  • service=parking_aisle crosses but must not connect amenity=parking.
  • amenity=parking extends to barriers, gardens, landuses or other areas.
  • Gardens and landuses divide into multiple features if separated by pavement.
  • barrier=kerb may implicitly define the width of the road along any given point.
  • footway=sidewalk often has a curb as it is then counted as a separate way.
  • barrier=fence or barrier=hedge often mark the back of a plot.
  • Road intersections have a single vertex (regardless of crossings present).
  • 90 degree quarter turns in roads reasonably have 3 or more vertices.
  • 180 degree U-turns in roads reasonably have 6 or more vertices.
  • 360 degree rotary turns reasonably have 12 or more vertices.
  • Arclength precision is limited by coordinate precision (currently 7 decimals):
  • At 0° latitude, two decimals (e.g. width=0.99) are distinguishable (i.e. ±5 mm).
  • At 56° latitude, three decimals (e.g. width=0.999) are distinguishable (i.e. ±500 µm).
  • For LiDAR, the precision you get is typically two or three decimals (both are reasonable).
  • But our coordinate system is believed to be centered with an accuracy of only ±1 cm. 🇬🇧🔀🌐

Armchair workflow (i.e. editors that I have found to be most useful)

  • iD is for hand tracing aerial/satellite imagery (by laser mouse or stylus), for getting good suggestions on how and what to map, and for verifying features detected in street level photography.
  • JOSM is for editing with advanced geometry tools and plugins (e.g. building, improve way accuracy, changeset reverter, auto_tools, SimplifyArea, terracer, todo, utilsplugin2).
  • MapComplete makes the task of adding and improving POIs by category very easy.
  • Overpass-turbo is for finding missing tags on features or missing features by query.
  • MapRoulette is for organizing mapping tasks. It works by launching the editor in a new tab, based on features that a vigilant user provided from an Overpass query or an ugly GeoJSON file (that must certainly be cleaned up).
  • RapiD is for importing freely licensed Esri ArcGIS Online datasets. It also allows clicking on individual lines to import them as roads, buildings, etc.
  • OSMBuildings is mostly for checking building relations and roofs.
  • F4Map, Kendzi3d, and Cesium Stories are for getting good interpretations of the data in 3D.
  • QGIS is for analyzing data. It converts between formats, selects features by query, and performs batch GIS operations. It supports plugins (e.g. QuickOSM, QuickMapServices, qgis2web).
  • Frameworks are for when you want to improve the workflow or use a programming language as part of your workflow.

Field/Survey workflow (i.e. editors targeting “mobile” platforms)

  • OsmAnd is for perusing the Map, for recording GPX traces loaded in iD, for offline and live 2D hillshade vector maps, for navigation of all vehicle types, and analyzing details about a route. Editing and Field Papers (e.g.) are also possible through plugins.
  • StreetComplete is for answering a guided set of questions (of your choosing) about features near you that have missing tags. This makes contributing really fun and easy.
  • Vespucci is for having an offline editor on-the-go. This can be really useful for remote, indoor or woody areas, and for quickly adding POIs on the go.

Organized edits and imports (et. al.)

  • Some addresses for Santa Barbara County (source:
  • Some buildings for Santa Barbara County (source: Microsoft building footprints).
  • Buildings realigned and merged with respective address points (‘Shift+Click’ node and way, then ‘C’ in iD editor).
  • Remodeling of Isla Vista.

Other OSM related work

  • Uploading street level imagery to OpenStreetCam.
  • Uploading artwork to Wikimedia Commons (planned: logos in an SVG format, information in TeX format).


Most collaboration is done through notes, messages, and changeset comments. I keep tabs on friends and local edits. I follow up on the user diaries, Reddit, Mastodon, and Telegram. The ideas you see here are unoriginal and can be found in some form from these sources and from the wiki.


I believe that all FOSS advocates are a pocketful of ideas. I credit OpenStreetMap as one of my main reasons for living. There just are no words to explain that sentiment. I am increasingly convinced at the moment that what the community needs is more editors that streamline specific tasks.


A user once told me that our data are beautiful. It appeared absurd, until I realized they’d been designed that way all along. I try not to gaze too long at the map because there are many simulators and many, many other applications of OSM data, including more important ones. If we want to save the trees, we can do so by tracing forests, which help to train a computer to detect deforestation and wildfires. Preventing disaster is possible, too. We can also reduce costs by combining our mapping efforts. So, we are mapping unique places, places less mapped, rather than just the most popular places. These are just examples. Once mapped, the data are actually invaluable. 🇬🇧🔀🌐