Data sources and limitations

Where does the data in TaxiwayClosed AI come from?

TaxiwayClosed AI builds its view from three layers: NOTAM-style closure information, OpenStreetMap-based airport surface geometry, and an AI agent that reads the NOTAM package and matches each closure to a taxiway, runway, or apron on the map. Each layer has known failure modes. NOTAM-style feeds can lag or miss items; OpenStreetMap geometry can be incomplete or mislabeled, especially at smaller airports; and AI interpretation can misread dense NOTAM text. The app exposes these limits through confidence and mapped-status cues instead of hiding them. This page documents what each layer contributes, what it cannot guarantee, and what a low-confidence result means in practice. Because any layer can be wrong, TaxiwayClosed AI supports flight preparation as an awareness aid only — pilots must always verify closures against official NOTAM briefings, current charts, approved EFBs, ATC, and company procedures.

A NOTAM (Notice to Air Missions, formerly Notice to Airmen) is a time-critical notice alerting pilots to hazards or changes — such as a closed taxiway or runway — at an airport or along a route. TaxiwayClosed AI works with NOTAM-style information: closure text in the familiar NOTAM format whose coverage, availability, and refresh timing vary by airport and by data-provider status. It is not a direct feed from the official NOTAM system, and there can be a lag between the moment a NOTAM is issued and the moment a matching item appears in the app.

The airport surface itself — runways, taxiways, aprons, terminals, gates — comes from OpenStreetMap, a community-maintained map database. Large international airports are usually mapped in good detail. Smaller airports can have missing taxiways, unnamed aprons, outdated layouts, or labels that differ from the official airport diagram.

On top of those two layers sits the AI agent. It reads the whole NOTAM package, extracts the subject, location, and schedule from each item, and tries to match the closure to a specific surface feature so it can place a cross (X) exactly where the closure applies. How that parsing works is explained on the AI NOTAM parser page. Like any AI system, it can misread dense abbreviations, pick the wrong taxiway segment, or miss a condition in the schedule.

What do confidence and mapped status actually mean?

Every closure item in TaxiwayClosed AI carries two cues. Confidence expresses how sure the AI agent is that it identified the right surface feature for a NOTAM. Mapped status says whether the item is drawn on the map at all. High confidence means the NOTAM text and the map geometry agreed cleanly — a named taxiway in the text matched a named taxiway in the geometry. It does not mean the closure is confirmed, current, or complete. Low confidence, or an unmapped item listed below the map, means the agent is telling you openly that it could not pin the location, and that you should read the raw text yourself.

These cues are honesty indicators, not safety determinations. They never confirm that a taxiway, runway, or apron is open, closed, authorized, or suitable for use, and they never approve a taxi route.

What does a NOTAM that will not map cleanly look like?

Here is a realistic case from Miami International Airport (MIA) where the app deliberately shows low confidence:

!MIA 07/038 MIA APRON WEST CARGO APRON EAST SIDE CLSD 2607140800-2607262200

• !MIA / MIA — the notice belongs to Miami International Airport.
• 07/038 — the 38th NOTAM issued for MIA in July.
• APRON WEST CARGO APRON EAST SIDE CLSD — the subject is an apron: the east side of the west cargo apron is closed.
• 2607140800-2607262200 — effective 14 July 2026 at 08:00 UTC through 26 July 2026 at 22:00 UTC.

The subject and the dates decode cleanly; the location is where mapping breaks down. "East side of the west cargo apron" describes a portion of an area, and in OpenStreetMap that whole area may exist as one polygon with no name at all. There is no feature called "west cargo apron" for the agent to anchor a cross to. In this case TaxiwayClosed AI shows the item with low confidence, or lists it as not mapped below the map, rather than guessing at a position. Contrast that with a crisp NOTAM like "TWY S BTN TWY P AND TWY R CLSD" — named line features that produce a precise cross, as illustrated on the taxiway closure map page.

How do I report a data problem?

If you spot an incorrect map label, a wrong cross placement, a bad summary, a stale timestamp, or any safety concern, email contact@etwintech.com with the airport, the date, the visible NOTAM text, and what looked wrong. Reports like these directly improve matching quality. Do not use email for operational decisions or urgent aviation safety matters.

Related pages