Air Tractor AT-502B accident near Belle Glade, Florida, September 15, 2004
On September 15, 2004 at about 4:05 pm local time, a Air Tractor AT-502B, registered N90375, was substantially damaged in an accident near Belle Glade, Florida (Belle Glade, Fl airport). It was an aerial application (crop spraying) flight under agricultural flying rules (Part 137). No one was hurt; 1 person was on board or involved. The weather was visual conditions (good weather).
The NTSB's probable cause their words, unchanged
The pilot's inadequate preflight inspection and a loss of engine power due to water contamination of the fuel.
Source: NTSB aviation accident database, copy made October 5, 2026. Docket and reports at the NTSB.
What the record shows
- Date
- September 15, 2004 · about 4:05 pm local time
- Place
- Belle Glade, Florida · Belle Glade, Fl · map
- Type
- Accident
- Injuries
- No one was hurt; 1 person was on board or involved.
- Weather
- visual conditions (good weather)
- Aircraft
- Air Tractor AT-502B
- Registration
- N90375 · no longer on the register · serial 502B-0580
- Damage
- Substantial damage
- Flight
- Aerial application (crop spraying) flight · agricultural flying rules (Part 137)
The NTSB's narrative final · quoted from the NTSB record
The airplane was topped off with 31 gallons JP-5 from the operator's above-ground 8,000-gallon supply tank. The pilot stated he did not sump the airplane's fuel tanks prior to departure. The flight departed the airstrip and was at 500 feet above ground level when the airplane lost engine power. The pilot's attempts to restore engine power were unsuccessful, and the pilot executed an emergency landing on a dirt road. The airplane collided with a dirt berm and spun into a ditch. Examination of the airplane revealed approximately two quarts of water was retrieved from each wing tank of the airplane, approximately one pint of water was retrieved from the fuel filter bowl, and the engine fuel filter was found saturated with water. Examination of the fuel supply tank revealed water in the tank and tank filter. The pilot stated the supply tank had not been used to fuel any airplanes since before Hurricane Frances impacted the Belle Glade, Florida, area with heavy rain and high winds on September 5, 2004. The pilot examined the tank and reported the vent lids on the tank could be lifted open, the pick-up tube for the nozzle extended to the bottom of the tank, and the tank was not equipped with a means by which it could be sumped or drained from its lowest point.
The complete narrative as the NTSB published it. The NTSB's docket holds the report as a PDF and any photographs, statements and other documents from the investigation.
The factual record from the NTSB's investigation tables, in plain English
Pilot
- Certificate: commercial pilot
- Ratings: multi-engine land; single-engine land; instrument: airplane
- Flight time: 8,000 hours in all; 2,000 in this make and model; 100 in the last 90 days
- Medical certificate: Class 2
- Injury: no injuries
The aircraft
- Landing gear: fixed
- Engine: Pratt & Whitney PT6A (turboprop); 0 hours total
The flight
- Departed from: PVT Belle Glade FL
- Flight plan: none
- Runway 18
Weather at the time
- Light: daylight
- Temperature: 0°F (-18°C), dew point 0°F (-18°C)
Injuries
| Fatal | Serious | Minor | None | |
|---|---|---|---|---|
| Crew | 1 |
Documents from the investigation the NTSB's docket: the evidence folder behind the report
The docket, the folder of records gathered during an investigation (maintenance records, photographs, witness statements, examinations), has not been read from the NTSB for this case yet; the list appears here once it has. Open the docket at the NTSB.
Everything on this page comes from the NTSB's public records. The narrative, probable cause and findings are the NTSB's own words; the coded tables behind the report are written out in plain English, with pilots' ages, home towns and medical details left out. The documents and photographs are the NTSB's docket, shown as the NTSB released them. This site's own text never names anyone involved. A preliminary report can change; the final report usually follows one to two years later, and the page is refreshed when it does.