The exceedance and the reportare one event.
Import the exceedances your flight data programme already produces, keep them on their own severity axis, and let the platform propose which crew report describes the same event. FDM events are their own record type, so they never enter an occurrence count, and the correlation is a suggestion a person accepts, never an automatic link.
FDM events cannot corrupt your occurrence numbers.
Importing exceedances as occurrences is the mistake that ruins a reporting-culture metric: four thousand machine events a month drown the reports people chose to file. Here the segregation is structural rather than a filter somebody remembers to apply. An FDM event is a different record type, so it is out of every occurrence list, closure rate and reporting-culture figure by construction.
The capabilities, on one record.
Event code, level, parameter, recorded and threshold values, exceedance magnitude, phase, route, date and tail, mapped through the shared import pipeline.
An event level is an event level. It is never auto-written into the ICAO severity pair, which a person sets at the assessed stage.
The field pack carries no crew identity, and the importer refuses a crew name, email or staff id mapping rather than accepting and hiding it.
The platform proposes the crew report that describes the same event. A person accepts the link; nothing is joined automatically.
A significant exceedance with no matching report inside your window is raised as a signal, which is the reporting-culture gap the data can actually see.
Real records, one reference each.
Every record carries a stable reference and a source on every field. Nothing is silently overwritten, and the trail is the record.
An FDM event is not an occurrence, and the schema enforces it
The event level never becomes an ICAO severity on its own
Crew identity is refused at import, not hidden after it
A proposed correlation waits for a person to accept it
It links to the rest of the graph.
Machine events never become crew reports.
An FDM event is a different record type, so four thousand exceedances a month cannot enter an occurrence count or a reporting-culture figure. The segregation is the schema, not a filter somebody remembers to apply.
The operation has been read from the recorder. The next chapter reads it from the people: safety promotion, and what the frontline is told.