Automated edits/ckronenburg

Airport surface markings from FAA charts and imagery

Cyriel Kronenburg, VP Airports & ATM, uAvionix Corporation

OSM account: Ckronenburg Contact: cyriel.kronenburg@uavionix.com

Motivation

uAvionix makes avionics and software for airports, ATM, drone operators and manned aircraft. We use OpenStreetMap airport data as guidance for some products to drive surveys, and in doing so found inconsistencies between our own FAA and survey derived maps and OpenStreetMap: holding position markings missing at most airports checked, including towered fields, and taxiway designators and physical pavement years out of date relative to the current FAA airport diagram.

These edits correct that. The work is done on my employer's time and in my employer's interest, and the corrections are equally useful to anyone else consuming aviation data from OSM.

What is changed

• Missing aeroway=holding_position added as ways drawn along the

 painted marking, tagged holding_position:type=runway or =ILS, named
 <taxiway>@<runway>. It now is a line instead of a single marker to
 better reflect the actual line on the field.

• Taxiway ref added or corrected where the FAA airport diagram and the

 FAA Airport Mapping Database disagree with OSM.

• Missing taxiway connectors present on the FAA diagram and in AMDB.
• aeroway=threshold with per-end refs on existing runway end nodes.
• Stale holding positions for decommissioned runways are untagged, never

 deleted — the node is left in place.

Nothing is deleted. Taxiways absent from current published documentation are retagged disused: or abandoned: where the pavement is still visible in imagery, and deleted only where the pavement is physically gone. No object is changed that is not covered by one of the cases above.

Sources

• FAA airport diagram for the current 28-day cycle (public domain).
• FAA Airport Mapping Database (AMDB) via the FAA ADDS ArcGIS service(public domain). Not available at every field.
• Esri World Imagery, used under the OSM community's existing permission.
• On-site surveys conducted by uAvionix together with the airport authority. (This is our own first-hand survey data, contributed to OSM with permission).

No third-party proprietary data is used.

Method

This is machine-assisted, not unattended.

Candidate positions are computed; every single object is then reviewed by a human against the imagery before it is written, and the reviewer's answer overrides the algorithm.

1. Candidate holding positions are found by walking OSM taxiway centerlines outward from the runway and detecting the painted marking in imagery.
2. For each candidate the bar angle is fitted by maximising the marking's cross-section over a ±5° sweep, then its lateral position is measured from a continuous intensity profile across the marking. The way is placed on the outermost solid line — the solid furthest from the runway, which is the furthest an aircraft may be while holding.
3. Where the imagery resolves the four individual painted lines, the reviewer is shown four numbered candidate lines one line-pitch apart and picks the correct one. Where it does not, the marking band is located by its outer edge and the line placed half a line width inside it, and the reviewer confirms alignment and angle instead.
4. A holding position marking is not always straight. Where a connector meets a parallel taxiway at a wide throat the marking follows the pavement and bends; those are drawn as a single way with a corner node, not as two ways.
5. For taxiway naming, FAA airport charts (public) are used as the authoritative source of naming
6. For airport pavement in use, FAA airport charts (not geo referenced) and survey data prior to deployment of surface awareness systems at the airport are used and maintained upon changes.
7. Human review gate. Each candidate is rendered on imagery at high magnification and reviewed one at a time. Anything not confirmed is not uploaded, and anything the reviewer flags is refitted and shown again.
8. The reviewed geometry is written to a JOSM .osm file, validated, and uploaded manually by the account holder in staged changesets, additive changes first.

At KRIC the review changed the algorithm's own answer on 18 of 21 lines. At KJAC (building the files) it changed the shape of the markings themselves. The gate is the substance of the method, not a formality.

Validation

Every uploaded object is read back from the OSM API afterwards and checked for length, straightness, duplicate node references and expected tags.

When

• KRIC (Richmond International): 3 September 2026, complete.
• KJAC (Jackson Hole) and KSAV (Savannah/Hilton Head): planned, after the notice period below.
• Further US towered airports with surface awareness systems (SAI) deployed thereafter, at a rate of a few per month. Each airport is a separate set of changesets.

Consultation

Announced on the OpenStreetMap Community Forum, United States category, on 3 September 2026: https://community.openstreetmap.org/t/adding-missing-airport-holding-position-markings-from-faa-sources/147259

Opting out

Email cyriel.kronenburg@uavionix.com, or comment on any changeset. If a local mapper or airport operator objects to the edits at their airport, that airport is removed from the list and any edits already made there are reverted on request, no argument.