In June we reported that the registry gap has a paper trail: 284 hospital helipads with federal airspace paperwork on file and no registration anywhere in the FAA's facility record. That piece asked what the record does not know. This one asks a harder question about what it thinks it knows. When the FAA's own case files and its own registry describe the same landing area but disagree about where it is, which one is right?
The raw material is the same archive: 483,006 airspace case files, every aeronautical study the agency has conducted on a proposed structure or landing area, matched against the 5,655 operational heliports in the facility registry. The overlap we care about here is the near miss class: 746 facilities where a determined landing area case sits between half a mile and two miles from a registered heliport. Close enough that the case and the registry almost certainly describe the same facility. Far enough apart that one of them is wrong.
A conflict needs a referee, and for hospital helipads there is a natural one: the hospital. A hospital helipad exists to serve a specific building whose location is public record. Whichever coordinate sits at that building wins. Applying that test, plus name corroboration between the case text and the facility record, split the 746 into three piles: six where the case file wins outright, 274 where the registry wins, and 466 that stayed ambiguous on paper.
The six
Every one of the six is a hospital helipad. In each, the registry's coordinate sits 3,700 to 7,800 feet from the hospital the pad serves, while the FAA's own case file places the pad within a few hundred feet of it. These are not rounding errors. They are the coordinates of a previous location, or of nothing at all, carried forward for a decade or more after the agency's own paperwork recorded the truth.
| Facility | Case file | Registry position sat | Case position sits |
|---|---|---|---|
Jefferson County Health Center, Fairfield IA7IA6 · aix:hp:N6F5Z5ZV | 2008: establish heliport at new hospital location, old helipad to be removed | 7,386 ft from the hospital | 381 ft |
Bay Area Medical Center, Marinette WI4WN6 · aix:hp:39R46NPB | 2016: relocation with the existing helipad decommissioned | 7,777 ft | 336 ft |
Guthrie Troy Community Hospital, Troy PAPN74 · aix:hp:E6A5JYTA | 2012: establish ground level private use heliport | 5,349 ft | 155 ft |
Carris Health Redwood Falls, MN37MN · aix:hp:V4KT4203 | 2009: establish (relocate) hospital helipad | 4,907 ft | 393 ft |
UP Health System Marquette, MI76MI · aix:hp:5QFKVPXP | 2014: helipad adjacent to the emergency wing | 3,969 ft | 520 ft |
Portneuf Medical Center, Pocatello IDID64 · aix:hp:KYBZ5H4P | 2009: heliport to be moved to new location | 3,780 ft | 482 ft |
Fairfield, Iowa is the cleanest example of the failure mode. In 2008 the hospital built a new campus and filed the paperwork: establish a heliport at the new hospital location, remove the old pad when the new facility opens. The FAA studied it, determined it, and filed it. Eighteen years later the registry still pointed at a coordinate 1.4 statute miles from the hospital, on the wrong side of town, while the agency's own case file carried the right answer the whole time. The record did not lack the information. It lacked the join.
We corrected all six on July 31. Each correction moved a pad between 0.6 and 1.3 nautical miles, each is recorded as a permanent event in our public change log, and each preserves the FAA's original coordinate alongside the corrected one. That last part matters more than it sounds, and we will come back to it.
The part that did not go our way
A finding that stopped there would be misleading, because the more common outcome points the other direction. After the six, we took the strongest of the 466 paper-ambiguous conflicts, the fourteen with the highest corroboration scores, into manual adjudication: the case text, the registry, and current satellite imagery of both candidate coordinates, side by side.
The registry won 13 of 14. In most of them the imagery was unambiguous: a marked pad, sometimes with a painted cross or letter H plainly visible, sitting exactly where the registry said, while the case file coordinate landed on a subdivision, a factory, a forest, or an empty field.
What the losing case coordinates had in common is worth recording, because it is a taxonomy of how position error enters federal paperwork. Some were truncated at filing time to whole minutes of latitude and longitude, a precision floor of roughly a nautical mile. One was a single axis transcription error: the case latitude matches the registry to five decimal places while the longitude is off by exactly one digit's worth of distance. Two belonged to different facilities entirely, deactivation paperwork for predecessor pads, one of them beside a hospital that has since been demolished, that our matcher had paired with the successor facility. And at the Palo Verde nuclear plant, the case file itself resolved the conflict: it announces a 2,800 foot relocation and states the new coordinates, and the registry matches them exactly. The agency tracked that move correctly. The conflict was our matcher reading the case's header coordinate, not the FAA being wrong.
Six confirmed errors out of 746 candidates is a 0.8 percent hit rate, and we consider that the credibility of the method, not its weakness. A correction layer that fired on all 746 would be noise. One that fires six times, each with the agency's own paperwork and a hospital's street address as witnesses, and that declines to fire when satellite imagery says the registry is right, is a layer you can build on.
A correction you can audit, and a delta we are watching
Correcting a federal record in a private database is easy to do badly. Overwrite the coordinate and you have destroyed the evidence that the error ever existed, and with it any way for a reader to check your work.
So the corrections are structured the other way. Each corrected facility carries both coordinates: ours, traceable to the specific FAA case file that justifies it, and the FAA's original, preserved as the standing delta between the record and the truth. Our monthly refresh from the federal source now compares the FAA's published position against that preserved mirror rather than against our corrected value. Which means two things going forward. Our corrections do not silently wash away when the federal file updates. And if the FAA ever fixes one of these six records, the next refresh will see the federal position move and flag it for review. The delta is not just documented. It is instrumented.
Why it matters is the same reason it has always mattered, sharpened. These are hospital pads. Air ambulance operators plan to them, obstruction studies measure from them, and the coming generation of advanced air mobility siting decisions will inherit whatever coordinate the record carries. A pad that is 1.4 statute miles from where the record says is not a data quality footnote. It is a different approach path, a different obstruction environment, and a different answer to whether a corridor works.
The registry gap now has a paper trail, a correction record, and a watch list. The next move belongs to the record.