Landmark Accidents: Vineyard Spiral

Low visibility contributes to JFK Jr.'s accident

September 1, 2000

The official NTSB report on the John F. Kennedy Jr. accident was released almost one year after the crash, an event that caused intense media and public scrutiny of general aviation — particularly regarding VFR flight at night. On July 16, 1999, at about 9:41 p.m. Eastern Daylight Time, a Piper Saratoga, N9253N, carrying JFK Jr., his wife, and his sister-in-law plunged into the Atlantic Ocean approximately seven and a half miles southwest of Martha's Vineyard, Massachusetts.

The weather played an important part in this accident and appeared to be benign. The NTSB and National Weather Service determined that the weather at the crash site was VFR. It will become obvious with hindsight that this was not a good flight for a new VFR pilot to undertake. Even with considerable instrument training, the pilot was unable to maintain control, and this accident underscores the need for caution, especially when flying over water and sparsely populated areas after dark.

Flight history

The flight originated from Caldwell, New Jersey (CDW), with a planned stop at Martha's Vineyard Airport (MVY), where Kennedy's sister-in-law was to be dropped off. The final destination was Hyannis, Massachusetts (HYA).

Kennedy informed an FBO employee in the early afternoon that he planned to depart Caldwell at around 6 p.m. Another pilot who was also heading out to the islands that night said that the auto traffic was the "second heaviest he had ever seen" and, as a result, he was delayed almost an hour beyond his own planned departure time. Traffic likely contributed to Kennedy's delayed departure.

Witnesses at Caldwell saw Kennedy using crutches as he loaded luggage aboard the Saratoga. Air traffic control transcripts showed that the flight departed at 8:38 p.m. It was nearly dark. After a quick discussion with the tower regarding the departure route, there was no further communication between the aircraft and ATC or flight service for the rest of the flight.

Radar data was used to reconstruct the flight path and, at 8:40 p.m., a VFR transponder target thought to represent N9253N was observed about one mile southwest of Caldwell at an altitude of 1,300 feet. The target proceeded to the northeast, on a course of 055 degrees, and climbed to 1,400 feet when it reached the Hudson River. Eight miles northwest of the Westchester County Airport in White Plains, New York, it turned north over the river. Shortly after that it turned eastward on a course of 100 degrees, climbing to 5,500 feet. The flight passed just north of Bridgeport, Connecticut, and crossed the shoreline between Bridgeport and New Haven, Connecticut. The ground track paralleled the Connecticut and Rhode Island coastlines; passed Point Judith, Rhode Island; and continued over the Rhode Island Sound.

About 34 miles west of Martha's Vineyard the radar data indicated that the flight began a descent from 5,500 feet. The speed during descent was calculated to be about 160 knots indicated airspeed, and the rate varied between 400 and 800 feet per minute. At approximately 9:38, the flight began a right turn in a southerly direction. About 30 seconds later, the descent stopped at 2,200 feet and the target began a climb that lasted another 30 seconds.

During this time, the target stopped the turn, and the airspeed decreased to about 153 KIAS. About 9:39, the target leveled off at 2,500 feet and flew southeasterly. About 50 seconds later, the target then entered a left turn and climbed to 2,600 feet. As the left turn continued, it began to descend at about 900 fpm. When the target reached an easterly direction, it stopped turning; its rate of descent remained about 900 fpm. At 9:40:15, while still in the descent, the target entered a right turn. By 9:40:25, the bank angle exceeded 45 degrees, the vertical acceleration was 1.2 Gs, the airspeed increased through 180 kt, and the airplane's nose was pitched down about 5 degrees.

After 9:40:25, the airplane's airspeed, vertical acceleration, bank, and dive angle continued to increase, and the right turn tightened until water impact, at about 9:41. The last radar hit at 9:40:34 showed an estimated descent rate of more than 4,700 fpm. There were no survivors.

Wreckage information

The wreckage was located in 120 feet of water about one-quarter of a mile north of the target's last recorded radar position. The Saratoga, based on radar and information obtained from the wreckage, struck the water with its right wing low, in a steep nose-down pitch attitude. The recovered attitude indicator showed a 125-degree right bank and 30 degrees nose low.

Data from the flight and engine instruments showed that the engine was developing power at the time of impact. The tachometer indicated 2,750 rpm, above the redline of 2,700 rpm, and the airspeed needle was slightly above the maximum 210 KIAS value shown on the instrument. The fuel-flow gauge needle was found slightly loose and indicated 22 gallons per hour.

There was no evidence of any preimpact failures of the airframe, engine, flight instruments, avionics, or autopilot. The landing gear was up. Throttle and propeller controls were found full forward. Annunciator lights from the autopilot showed no evidence of filament stretching, indicating that the autopilot was not in use.

Aircraft information

The accident airplane was a Piper PA-32R-301 Saratoga II, a single-engine, low-wing airplane with retractable landing gear. The airplane was originally certificated by Piper Aircraft Corporation on June 9, 1995, and had been owned by several pilots before being sold to Kennedy on April 28, 1999. During the prepurchase inspection, "The aircraft was found to be in very good condition, with only a few minor discrepancies." An annual inspection was completed on June 18, 1999, at a total airframe time of 622.8 hours.

Meteorological information

There is a wealth of weather data and the warning, while there, is subtle. Would you have been concerned? Some information, italicized in the text and table, shows that the weather had the potential to snare a VFR pilot. Had the flight departed as originally planned, Kennedy and his passengers would have arrived well before sunset. Sunset was at about 8:14 p.m., and civil twilight ended at about 8:47. When the accident occurred about 9:40, the moon was 11.5 degrees above the horizon at a bearing of 270.5 degrees and provided about 19 percent illumination. Despite the relatively good weather report from Martha's Vineyard, several pilots reported considerable haze, which would have obscured what little moonlight there was.

According to Weather Service International (WSI), a private weather service, Kennedy — or someone using his password — made two weather requests from WSI's Web site on July 16, 1999. The first request, at 6:32 p.m., was for a radar image and the second, made at 6:34, was for a route briefing from Teterboro to Hyannis, with Martha's Vineyard as an alternate.

Pilot preflight weather requests

The 6 p.m. weather observations from several airports along the route indicated that visibilities ranged from 10 miles along the route to four miles in haze at Caldwell. The lowest cloud ceiling was reported as 20,000 feet overcast at Providence, Rhode Island. Observations for Nantucket (ACK), Hyannis, and Martha's Vineyard were also included in the briefing. At the departure point of Caldwell at 5:53, the sky was clear; visibility four miles in haze; and the winds were 230 degrees at 7 kt.

However, according to WSI, Kennedy did not access the updated National Weather Service (NWS) area forecast (FA). Excerpts from the Boston area forecast, issued on July 16 at about 8:45 p.m. and valid until July 17 at 2 a.m., included scattered clouds at 2,000 feet, occasional visibility 3 to 5 miles in haze, with haze tops at 7,000 feet for the area including Martha's Vineyard.

The National Weather Service does not prepare aviation terminal forecasts (TAFs) for Martha's Vineyard. Excerpts from the closest TAF pertinent to the accident, issued for Nantucket on July 16 about 1:30 p.m. and valid from 2 p.m. July 16 to 2 p.m. July 17, was for clear skies, visibility greater than 6 miles, and winds from 240 degrees at 15 kt. A later forecast was not quite so optimistic. The 7:30 p.m. TAF, valid from 8 p.m. July 16 to 2 a.m. July 17, was for winds from 240 degrees at 15 kt; visibility 4 miles, mist; and scattered clouds at 25,000 feet. Temporary changes from July 16 at 9 p.m. to July 17 at 1 a.m.: clouds 500 feet scattered; visibility 2 miles, mist.

The terminal forecast for Hyannis also deteriorated; the 1:30 p.m. TAF called for clear skies and visibility greater than 6 miles, but the TAF issued at 7:30 called for winds from 230 degrees at 10 kt; visibility 6 miles, haze; and scattered clouds at 9,000 feet — with temporary changes from 8 p.m. July 16 to midnight July 17 of visibility 4 miles, haze.

There were no airmets, sigmets, convective sigmets, or in-flight weather advisories in effect along the route between Caldwell and Martha's Vineyard from 8 p.m. to 10 p.m.

Surface weather observations

Martha's Vineyard used an automated surface observing system (ASOS) that could be edited and augmented by ATC tower personnel if necessary. Despite some assertions by the tabloid press, the NTSB found no anomalies regarding the ASOS at the Vineyard. During an interview, the tower manager stated that no actions were taken regarding the ASOS during his shift, which ended just after the accident occurred. He also stated, "The visibility, present weather, and sky condition at the approximate time of the accident was probably a little better than what was being reported. I say this because I remember aircraft on visual approaches saying they had the airport in sight between 10 and 12 miles out. I do recall being able to see those aircraft and I do remember seeing the stars out that night.... To the best of my knowledge, the ASOS was working as advertised that day with no reported problems or systems log errors."

The Nantucket weather was clearly marginal for VFR operations, especially at night. The one-degree temperature/dew point spreads at Nantucket and at Hyannis show the highly variable nature of weather in the islands. Even without clouds, especially at night, haze can be a significant obscuring factor. In this microclimate, fog forms rapidly and can be localized. The Martha's Vineyard weather, taken out of context, would lead one to believe that VFR was a reasonable option.

Further evidence of that comes from some Coast Guard weather observations that would not have normally been available to pilots. At Point Judith, Rhode Island, the 5 p.m. and 8 p.m. reports were cloudy, with 3 miles' visibility in haze. By 11 p.m., however, it was cloudy with 2 miles' visibility.

Pilot observations

One pilot who flew from Teterboro to Nantucket requested current weather observations and forecasts for Nantucket and other points in Massachusetts, Connecticut, New York, and New Jersey. Visibilities were well above VFR minimums. He asked flight service "...if there were any adverse conditions for the route TEB to ACK. I was told emphatically: 'No adverse conditions. Have a great weekend.' I queried the briefer about any expected fog and was told none was expected and the conditions would remain VFR with good visibility. Again, I was reassured that tonight was not a problem."

The pilot departed Teterboro "...in daylight and good flight conditions and reasonable visibility. The horizon was not obscured by haze. I could easily pick out landmarks at least five [miles] away." Above 14,000 feet, the visibility was unrestricted. During descent to Nantucket, when GPS indicated that he was over Martha's Vineyard, he looked down and "...there was nothing to see. There was no horizon and no light.... I turned left toward Martha's Vineyard to see if it was visible but could see no lights of any kind nor any evidence of the island.... I thought the island might [have] suffered a power failure. I had no visual reference of any kind, yet was free of any clouds or fog." Upon contacting Nantucket Tower for landing, he was instructed to fly south about five miles; however, he maintained a distance of three to four miles because he could not see the island at five miles. Approaching the airport, he made a turn for spacing and "found that I could not hold altitude by outside reference...."

Another pilot flying from Bar Harbor, Maine, to Long Island, New York, crossed Long Island Sound at about 7:30 p.m. The preflight weather briefing indicated visual conditions, but the pilot filed IFR at 6,000 feet. Visibility ran two to three miles in haze throughout the flight. The lowest visibility was over water, but no clouds were encountered.

A third pilot departed Teterboro at about 8:30 p.m. destined for Martha's Vineyard. Climbing to 7,500 feet, the route took him over the north shore of Long Island. The entire flight was conducted under VFR, with a visibility of three to five miles in haze. Over land, he could see lights on the ground when looking directly down or slightly forward. Over water, there was no horizon; he encountered no cloud layers or ground fog during climb or descent. Near Gay Head, on the southwest corner of the island, he began to observe lights on Martha's Vineyard. About four miles from MVY he first observed the airport's rotating beacon and landed at about 9:45.

Another pilot at Caldwell canceled his planned flight from there to Martha's Vineyard because of the "poor" weather. "From my own judgment, visibility appeared to be approximately four miles — extremely hazy. Winds were fairly light. Based only on the current weather conditions at CDW, the fact that I could not get my friends to come with me, and the fact that I would have to spend money on a hotel room in Martha's Vineyard, I made the decision to fly my airplane to Martha's Vineyard on Saturday."

The interviewed pilots who flew that night, despite operating under VFR, were apparently experienced and qualified to fly IFR if needed. We will never know how many noninstrument-rated pilots successfully made the trip that night and how many canceled.

Pilot information

Kennedy obtained his private pilot certificate in April 1998 and received a high-performance airplane signoff in his Cessna 182 in June 1998. His complex sign-off in the accident airplane was completed in May 1999. Estimated total flight time, excluding simulator training, was about 310 hours, of which 55 hours were at night. Seventy-two hours were without a CFI on board. His estimated flight time in the accident airplane was about 36 hours, of which 9.4 hours were at night. Only about three hours of that flight time was without a CFI on board, and Kennedy had flown the aircraft by himself at night for less than an hour. In the 15 months before the accident, Kennedy had flown 35 flight legs either to or from the Essex County/Teterboro, New Jersey, and the Martha's Vineyard/Hyannis, Massachusetts, areas. At least half of these trips were without a CFI on board, and five occurred at night.

The CFI who prepared Kennedy for his private pilot checkride observed that he had "very good" flying skills for his experience level. The pilot examiner who administered the checkride said he successfully recovered from two unusual attitudes while wearing a hood. During 1998, Kennedy flew approximately 179 hours, including about 65 hours without a CFI on board. In March 1999, he passed the instrument pilot knowledge (written) examination.

In April 1999, Kennedy went to a highly respected flight academy for concentrated instrument training, where he completed about half of the course. His instrument instructor noted progression was normal and that he "grasped all of the basic skills needed to complete the course." His instrument flying skills and simulator work were observed to be excellent. However, there was trouble managing multiple tasks while flying, which the CFI felt was normal for the pilot's level of experience.

Kennedy continued to receive flight instruction in his new Saratoga from CFIs in New Jersey. On one flight from Caldwell to Martha's Vineyard with an instructor — less than a month before the accident — an instrument approach was made into Martha's Vineyard through a 300-foot overcast. The CFI requested an IFR clearance and demonstrated a coupled ILS approach to Runway 24. Kennedy performed the landing, but the CFI assisted with the rudder because of the pilot's injured ankle.

Another CFI who flew with Kennedy for 39 hours between May 1998 and July 1999 accumulated 21 hours of night flight and 0.9 hour in instrument conditions. On July 1, 1999, the CFI flew with the pilot in the Saratoga to Martha's Vineyard. The flight was conducted at night with IMC at the airport. During the flight, Kennedy used and seemed competent with the autopilot. The CFI had to taxi the airplane and assist the pilot with the landing because of Kennedy's leg injury.

The instructor stated that Kennedy had the ability to fly the airplane without a visible horizon but was not ready for an instrument evaluation as of July 1, 1999, and needed additional training. The CFI observed that he would not have felt comfortable with Kennedy conducting night flight operations on that route and in those weather conditions. On the day of the accident, the CFI offered to accompany them that night but Kennedy replied that "he wanted to do it alone."

A third CFI flew with Kennedy for nearly 60 hours between May 1998 and July 1999, including 17 hours of night flight and eight hours flown in IMC. This CFI had conducted the complex airplane sign-off in May 1999. On one or two occasions he noted a disparity in the airplane's autopilot where it turned to a heading other than the one selected, but did not feel that the problem was significant. As noted earlier, no discrepancies could be found in what was left of the autopilot.

The CFI made six or seven flights to Martha's Vineyard with Kennedy in the accident airplane. Most were night flights, and Kennedy did not have any trouble flying the airplane. He was methodical about flight planning and very cautious about his aviation decision making. The CFI felt that he had the capability to conduct a night flight to Martha's Vineyard as long as a visible horizon existed.

In early June, Kennedy fractured his left ankle in a hang-gliding accident; it was placed in a cast, and later, a walking cast. The walking cast was removed, and on July 16, 1999, the day of the accident, he was given a "straight cane and instructed in cane usage." The orthopedic surgeon felt that, at the time of the accident, the pilot would have been able to apply the type of pressure with the left foot that would normally be required by emergency brake application in an automobile.

The probable cause of the accident, according to the NTSB, was the pilot's failure to maintain control of the airplane during a descent over water at night, which was a result of spatial disorientation. Factors in the accident were haze and the dark night. The depth of the NTSB's human factors investigation in this accident is commendable, recognizing the high-profile nature of the victims. This emphasizes the need for thorough human factors data collection, especially in weather-related accidents like this. The more we can learn about a pilot's background, training, and the decision processes used, the better we are able to educate ourselves to avoid similar circumstances.

Kennedy had more than the average amount of instruction and experience for his low total number of flight hours and flew far more frequently than most noncareer pilots. He had already shown that he could control an aircraft solely by reference to instruments and had done so before under similar conditions with a CFI on board. He was conducting a flight in an area with which he was very familiar and had been successful both at night and in IMC.

Two other factors worth mentioning are Kennedy's relatively low time in the aircraft type — a common predictor in accident scenarios — and Kennedy's low pilot-in-command hours. Most of his flight time was dual. The dynamic changes when a pilot must make all the decisions on his own when he is used to having an experienced CFI call the tough ones.

The weather at the reporting points was reasonable for a day VFR flight. However, the reports must be taken in context. At night, over the water, it becomes marginal. As reported by the interviewed pilots, en route conditions were very hazy. A point that cannot be emphasized too strongly is that a pilot's decision making must be based on what is seen (or not seen) through the windshield. It must not be based on forecasts or on observation points some 10 or 15 miles distant. What you see is what you get! Weather forecasting is both art and science, and the weather will change — something that many pilots fail to take into account.

The flight proceeded normally until the descent, when it appears that Kennedy disengaged the autopilot. That was precisely the time, in retrospect, to leave it on. The stable platform of level flight becomes more treacherous when the aircraft is climbing or descending. The fluids in the inner ear move in multiple directions, and the final radar plot shows a classic disorientation spiral where the aircraft descended and climbed, made several turns, and then the pilot — when level — was tricked by his senses into rolling the aircraft into a gentle bank. As the nose pitched down and the airspeed increased, the pilot, sensing he was level, pulled back rather than rolling wings level. This further increased the bank angle, which caused the nose to drop farther and G forces to build. The only salvation is to believe the flight instruments and ignore your senses. It is counterintuitive and essential.

With hindsight, what should have been done differently? Obviously, complete the flight before dark, get updated forecasts, ask for (and give) pilot reports, and have an alternate plan. Kennedy had no contact with anyone after departing Caldwell. VFR traffic advisorys, frequently called flight following, from ATC would have been an excellent idea in the high-density traffic of the Northeast. Filing a flight plan is also recommended, but flight following could facilitate a faster search-and-rescue response if needed. It also helps to keep VFR flight separated from IFR traffic, but that is dependent on controller workload. The Golden Rule is to always have an out, an alternative for escaping a bad situation. By the time the flight was descending, the options were rapidly diminishing. The focus had shifted to landing at the destination, not diverting to an alternate, which could have easily been done as the flight started out over the water and the poor visibility became evident.

This accident is ironic in one sense that Kennedy had invested in quality training, was current, familiar with the area, and had the best equipment. The softer part of the analysis is the pilot's mindset. He was a successful entrepreneur, possibly under some business stress, pressured to make a schedule for a family activity, and must have experienced the self-imposed stress that most pilots feel to complete a trip. Couple that with his earlier successful exposure to the night/haze/IMC environment of the islands, and it seems he believed that the reward of completing the trip outweighed what seemed like a relatively low risk of spatial disorientation.


Spatial disorientation: When left is right

According to the Aeronautical Information Manual, illusions that lead to spatial disorientation are one of the leading factors in fatal accidents. The graveyard spiral is a likely scenario in this accident. "An observed loss of altitude during a coordinated constant-rate turn that has ceased stimulating the motion-sensing system can create the illusion of being in a descent with the wings level. The disoriented pilot will pull back on the controls, tightening the spiral and increasing the loss of altitude." Once the fluid in the semicircular canals of the inner ear has stabilized, the sensation is generally of being in level unaccelerated flight. When another control input is made, unless the pilot correctly interprets and follows the flight instruments, it is likely that the aircraft will be placed in an unusual attitude.

A very strong disorienting sensation can be induced by putting the aircraft into a prolonged constant-rate turn, allowing the sensation-sensing organs to stabilize, and then abruptly turning the head. This might occur where the pilot bends down to look at a chart, reach for something on the floor, or to change fuel tanks if the switch is poorly located. The FAA's Vertigon simulator, frequently seen at airshows, is a great place to experience this. This is not recommended for those with weak stomachs or after ingesting large quantities of greasy food. The results can be spectacular.

Obviously, if pilots maintain reference to the horizon, disorientation is not an issue. However, a benign situation can quickly deteriorate. Turn on the autopilot, if so equipped, before attempting to resolve the attitude question yourself (see " On Autopilot: Autopilot Directing"). Autopilots, even basic wing levelers, will do a far better job of keeping the aircraft upright than will a VFR pilot. Once the aircraft is stable, a 180-degree turn should return you to safe conditions. Autopilots are not intended as a substitute for the instrument rating, but they can save lives in critical situations. VFR pilots who check out in autopilot-equipped aircraft should be proficient in the autopilot's use. — BL


See also the index of "Safety Pilot" articles, organized by subject. Bruce Landsberg is executive director of the AOPA Air Safety Foundation.