OpenAirDocs.org

Ditching

Open the Markdown

NTSB defining event · CICTT: Other

CICTT occurrence categories (PDF)CICTT phases of flight (PDF)

All 4 cases186 anywhere in the sequence
Defining event in
4 cases
In the sequence of
186 cases
Fatal cases
2
NTSB code
441

NTSB has coded findings and defining events since 2008. Earlier cases used its older occurrence and cause codes, so they aren’t counted here.

When and where

In green, cases where it was the defining event; in gray, cases where it came elsewhere in the sequence. The map shows the cases where it was the defining event. Events are coded with the final report, so the latest years are incomplete.

Cases per year

2008Defining eventElsewhere in the sequence2026
Query with SQL
duckdb -csv -c "SELECT year(c.\"eventDate\") AS year, count(DISTINCT e.case_mkey) AS cases,
    count(DISTINCT e.case_mkey) FILTER (WHERE e.\"isDefiningEvent\") AS defining
  FROM 'https://openairdocs.org/data/latest/ntsb_events.parquet' e JOIN 'https://openairdocs.org/data/latest/ntsb_case_detail.parquet' c ON c.\"mKey\" = e.case_mkey
  WHERE e.\"tier2Number\" = '441' AND e.case_mkey IN (SELECT case_mkey FROM 'https://openairdocs.org/data/latest/ntsb_events.parquet' WHERE \"isDefiningEvent\")
  GROUP BY 1
  ORDER BY 1"

Prints the counts as CSV. It reads the release’s Parquet files over HTTPS, so there’s nothing to download first. Install DuckDB to run it.

In these cases

Phase of flight
  • Unknown1
  • Maneuvering-low-alt flying1
  • Maneuvering1
  • Takeoff1
Query with SQL
duckdb -csv -c "WITH defining AS (SELECT * FROM 'https://openairdocs.org/data/latest/ntsb_events.parquet' WHERE \"tier2Number\" = '441' AND \"isDefiningEvent\")
  SELECT * FROM (
  SELECT 'Phase of flight' AS kind, e.\"tier1Number\" AS code, max(e.\"eventTier1Name\") AS name, count(DISTINCT e.case_mkey) AS cases
  FROM defining e
  GROUP BY ALL
  UNION ALL
  SELECT 'Finding in the probable cause', left(f.\"findingCode\", 8), max(f.\"tier4Name\"), count(DISTINCT f.case_mkey)
  FROM defining e JOIN 'https://openairdocs.org/data/latest/ntsb_findings.parquet' f ON f.case_mkey = e.case_mkey AND f.aircrafts_position = e.aircrafts_position AND f.\"inProbableCause\"
  WHERE f.\"findingCode\" GLOB '[0-9][0-9]*'
  GROUP BY ALL
  UNION ALL
  SELECT 'Aircraft model', a.aircraft_model_code, any_value(concat_ws(' ', m.manufacturer, m.model)), count(DISTINCT e.case_mkey)
  FROM defining e JOIN 'https://openairdocs.org/data/latest/ntsb_aircrafts.parquet' a ON a.case_mkey = e.case_mkey AND a.position = e.aircrafts_position
  LEFT JOIN 'https://openairdocs.org/data/latest/faa_aircraft_model.parquet' m ON m.code = a.aircraft_model_code
  WHERE a.aircraft_model_code IS NOT NULL
  GROUP BY ALL
  )
  QUALIFY row_number() OVER (PARTITION BY kind ORDER BY cases DESC) <= 10
  ORDER BY kind, cases DESC"

Prints the counts as CSV. It reads the release’s Parquet files over HTTPS, so there’s nothing to download first. Install DuckDB to run it.

Newest cases · 4

NTSB GAA25WA232AccidentClass 5 · ForeignRobinson R22 Beta · Qld, AustraliaMar 27, 2025 · VH-KNG · DestroyedNTSB GAA23WA220Accident · No injuriesClass 5 · ForeignPiper PA18 · Biscarosse, France, FranceJul 3, 2023 · F-HYHB · Substantial damageNTSB GAA23WA045Accident · 2 fatalClass 5 · ForeignBell 412EP · Ocean, MexicoDec 22, 2022 · XA-BNANTSB GAA22WA307Accident · 2 fatalClass 5 · ForeignRobinson R44 · Off shore, EcuadorJul 28, 2022 · HC-CLZ · Substantial damage
Query with SQL
duckdb -csv -c "SELECT c.\"ntsbNumber\", c.\"eventDate\", c.\"eventType\", c.\"eventCity\", c.\"eventStateOrRegion\", a.n_number, a.\"aircraftMake\", a.\"aircraftModel\", c.\"highestInjuryLevel\", c.\"totalFatal\", c.\"completionStatus\", c.report_url
  FROM 'https://openairdocs.org/data/latest/ntsb_case_detail.parquet' c LEFT JOIN 'https://openairdocs.org/data/latest/ntsb_aircrafts.parquet' a ON a.case_mkey = c.\"mKey\" AND a.position = 1
  WHERE c.\"mKey\" IN (SELECT e.case_mkey FROM 'https://openairdocs.org/data/latest/ntsb_events.parquet' e WHERE e.\"isDefiningEvent\" AND e.\"tier2Number\" = '441')
  ORDER BY c.\"eventDate\" DESC, c.\"ntsbNumber\" DESC"

Prints the whole list as CSV. It reads the release’s Parquet files over HTTPS, so there’s nothing to download first. Install DuckDB to run it.