Federal Express 1478

About 0516:38, the captain questioned the flight engineer about the weather information, stating, "one thousand scattered, ten miles, uhh is zat what it said . . . there?" About 0516:43 (while the captain was finishing his statement), the first officer began the approach briefing for runway 27 at TLH, stating, in part, "we'll plan on a visual to runway 27 . . . We'll back it up with this . . . ILS runway 27 full procedure . . . 272 is the final approach course inbound." The first officer stated that the minimum safe altitude was 3,300 feet mean sea level (msl) "all the way around . . . missed approach will be as published and we'll talk to 'em and see if we can get something better . . . runway's 8,000 [feet long], plan on rollin' out to the end . . . got a PAPI on the left-hand side . . . pilot-controlled lighting, so if you can . . . click it seven times I'd appreciate it." About 0518:30, the first officer stated, "all right, start on down," and the captain responded, "all right." The captain then radioed Jacksonville ARTCC, stating, "uh, Atlanta FedEx uh fourteen seventy eight, leaving two nine oh for uh, nine thousand."

According to the CVR transcript, about 0519:38, the first officer asked, "you wanna land on nine if we see it?" He added, "we got a PAPI on nine, too." The captain responded, "yeah, maybe . . . be a longer taxi for us, but . . . way we're comin' in probly two seven be about as easy as any of 'em." The first officer said, "okay."" The pilots initiated the in-range checklist about 0521:57 and completed it about 0522:20. During this time, the CVR recorded a sound similar to the microphone being keyed six times in about 1.3 seconds.

According to the CVR, about 0524:23, the captain stated, "runway nine . . . PAPI on the left side . . . I don't know, you wanna try for nine?" The first officer responded, "we're pointed in the right direction, I don't know, like you said . . . kinda a long . . . taxiback." The captain said, "yeah, that'd be all right." The first officer further stated, "I always thought you were supposed to land with the prevailing wind . . . at an uncontrolled . . ." and the captain responded, "well, at 5 knots, it really . . . the only advantage you have, landing to the west you have the . . . glideslope . . . which you don't have to the east." The captain asked the first officer if he was familiar with TLH, and the first officer replied that he was not.

The pilots continued the approach checklist and completed it about 0528:57. About 0529:53, the captain asked the first officer if he wanted to tell the ARTCC controller that they had TLH in sight. The first officer responded, "yeah. I don't see the runway yet, but I got the beacon." About 0530, the captain told the ARTCC controller that they had the airport in sight. Jacksonville ARTCC then cleared the pilots of flight 1478 for the visual approach into TLH and asked if they were aware that runway 18/36 was closed. The captain responded, "no sir, but . . . we're gonna use runway nine."

About 0532:34, the first officer stated, "I hope I'm lookin' in the right spot here." The captain responded, "see that group of bright lights kinda to the south down there and you see the beacon in the middle of it? . . . right over there . . . you're kinda on about . . . ten mile left base or so." The first officer then indicated that he had been "looking at the wrong . . . flashin' light." About 0533:05, the first officer repeated, "I was lookin' at the wrong light," and the captain responded, "yeah okay, yeah."

About 0536:37, the airplane was slightly more than 2.5 nautical miles (nm) from the airport and was transitioning from an angled base-to-final leg to line up with the runway. The Safety Board's airplane performance study indicated that, about this time, the PAPI would have been displaying one white light and three red lights when viewed from the cockpit. About 0536:40, the PAPI display would have shown four red lights. (Figure 1 shows the airplane's descent profile in relation to the PAPI light indications, flap extension, the runway, and treetops. Figure 2 shows the same information with select CVR comments overlaid.) About this time, the power on the three engines increased from about 1.05 to 1.24 EPR (engine pressure ratio),15 then, about 0536:41, the power to the engines decreased from 1.24 to 1.17 EPR. About 0536:43, as the airplane approached 500 feet, the captain asked the first officer if he wanted to go to flaps 30, and the first officer responded, "please." At 0536:47.8, the CVR recorded the GPWS announcement indicating that the airplane was passing through 500 feet agl. About 0536:49, the CVR recorded the captain stating, "stable." The Safety Board's airplane performance study indicated that, at this time, the airplane was 1.8 nm west of runway 9, descending through 500 feet agl at a vertical speed of 1,248 feet per minute (fpm), with engine power settings of about 1.17 EPR and an airspeed of 152 knots.

Also about 0536:49, the CVR recorded the first officer stating, "[I'm] gonna have to stay just a little bit higher . . . I'm gonna lose the end of the runway." About 0536:51, the captain responded, "yeah . . . yeah, okay."

About 0536:58, the FDR data indicated that the engine power began to increase from 1.17 EPR, reaching 1.20 EPR about 4 seconds later. At 0536:59.7, the captain advised TLH traffic that flight 1478 was on short final for runway 9. About 0537:09, the captain said, "it's startin' to disappear in there a little bit, [isn't] it? Think we'll be alright, yeah." The performance study indicated that, about this time, the airplane was 0.9 nm west of runway 9, descending through about 200 feet agl at a vertical speed of 528 fpm and an airspeed of 146 knots; the airplane performance study indicated that the PAPI indication observed from the cockpit would have been four red lights.

About 0537:13, the flight engineer announced that the before landing checklist was complete. This announcement was the last flight crewmember statement recorded by the CVR. At 0537:14, the GPWS announced that the airplane passed 100 feet agl; FDR and airplane performance information indicated that, at this time, the airplane was 0.7 nm west of runway 9, descending at a vertical speed of 432 fpm and an airspeed of 144 knots and that the engine power had increased to about 1.34 EPR. At 0537:19.9, as the GPWS announced 50 feet agl, the engine power increased to about 1.46 EPR. At 0537:20.3, as the GPWS announced 40 feet agl, the No. 2 and No. 3 engine EPRs began to increase rapidly. At 0537:20.7, the CVR recorded the sound of a crunch, and, about 0537:21, the GPWS announced 30 feet agl. About 0537:22, the CVR recorded another crunch sound, and the No. 1 engine EPR began to increase rapidly. At 0537:22.6, the GPWS announced "bank angle, bank angle." The CVR transcript indicates that, about 0537:23, the sound of crunching began again and, about 0537:25, a loud squeal began; both sounds continued to the end of the recording at 0537:26.2.

The airplane collided with trees in a right-wing-low, slightly nose-up attitude during the approach to runway 9 then impacted the ground, coming to rest on a heading of 260º degrees about 1,556 feet west-southwest of the runway. A postimpact fire ensued; however, the three flight crewmembers exited the airplane through the captain's side sliding cockpit window before the fire reached the cockpit. The accident occurred about74 minutes before sunrise.

Source: AAR-04/02, §1.1

During postaccident interviews, the captain told investigators that at 0430 on July 23, 2002, FedEx scheduling personnel contacted him by telephone to advise him that he was assigned to fly FedEx flight 1380, from Shreveport Regional Airport (SHV), in Shreveport, Louisiana, to MEM later that day. He arrived at SHV about 1010 on July 23 and checked into a hotel, where he rested before he returned to the airport about 2030 for his assigned flight. When the captain arrived at MEM (on FedEx flight 1380) at 2353, he was released from duty and returned to his home, arriving after midnight on July 24, 2002.

The captain stated that he did not sleep well that night. He reported that he stayed awake for a "couple of hours" after he got home to take care of the family dog, which was in deteriorating health. He indicated that he slept on the couch the rest of that night so he could more easily care for the dog during the night and stated that his sleep was accordingly interrupted three times during the night. He stated that his rest period ended about 0730 on July 24 and that he engaged in routine activities throughout the day. He went to bed about 2130 that evening, again sleeping on the downstairs couch and getting up several times during the night to care for the sick dog. He stated that he awoke about 0730 on July 25, 2002, and described his sleep quality as "marginal, not really good."

Between 1800 and 1830 that day, the captain checked company scheduling using his home computer and received notification of the TLH flight assignment. He stated that he slept from about 2100 on July 25 to about 0030 on July 26; he described his sleep during that 3 1/2 hours as "pretty good" and said that he did not feel fatigued when he subsequently arrived at MEM for the accident flight.

Source: AAR-04/02, §1.5.1

The first officer reported that he awoke again on the morning of July 24 and engaged in routine activities around the apartment during the day. According to a roommate, during a conversation that morning, the first officer complained about the reserve schedule he was flying because it "reversed day and night sleeping on consecutive days." The first officer stated that he had dinner with his landlord that evening, went to bed about 2100, and slept until early morning on July 25, when he had to get up to report for duty.

The first officer arrived at MEM about 0300 on July 25 and departed on FedEx flight 134 about 0356, arriving at Winnepeg International Airport (YWG), in Manitoba, Canada, about 0645. He went to a hotel and slept for about 5 to 6 hours and had dinner before reporting for duty at YWG again about 1818. He described his quality of sleep as "no better or worse than most day sleeps." The first officer departed YWG on FedEx flight 137 about 1902 and arrived at Grand Forks International Airport (GFK), in Grand Forks, North Dakota, about 1935. Flight 137 departed GFK about 2057 and arrived at MEM about 2303.

After flight 137 landed at MEM, the first officer was notified that he was scheduled to work flight 1478 to TLH, which was scheduled to depart MEM about 0312 on July 26 (about 4 hours after flight 137 arrived). The first officer stated that he accepted the flight 1478 trip assignment after he ascertained that it did not violate existing FedEx/pilot union agreements and would not result in his exceeding flight and duty limits. He indicated that he slept for about 1 1/2 hours in a private sleep room in FedEx's crew rest facilities at MEM before he met the captain to prepare for the accident flight. He stated that, although he described that rest as "good" sleep, he did not recall "feeling alert." A friend and roommate of the first officer's told Safety Board investigators that before the accident trip, the first officer "looked tired, like everyone else at 0330." During postaccident interviews, the accident captain said that the first officer "seemed tired, but maybe it was just his personality; he seemed not as communicative, not as alert. He may have been preoccupied."

Source: AAR-04/02, §1.5.2

The first officer told investigators that he never had a color vision problem in the Navy. However, during a July 24, 1995, evaluation for an FAA medical certificate, the first officer did not pass a color vision screen that was conducted using pseudoisochromatic plates (PIP). The FAA-designated medical examiner who conducted the evaluation contacted the FAA's Regional Flight Surgeon for advice and was told to issue the first officer's medical certificate for use with a SODA for the color vision deficiency. In a July 25, 1995, letter to the Regional Flight Surgeon documenting their discussion, the medical examiner stated that the results of the first officer's color vision screen indicated "a color vision loss on pseudoisochromatic plates, missing numbers 3, 4, 5, and 6. This suggests mild red-green defect. Per your instructions I gave him his certificate . . . . Please notify me about any SODA number that he is issued." The FAA issued a SODA on August 1, 1995, and, in a same-day letter to the first officer, the Regional Flight Surgeon stated, "based on your operational experience, I have determined that you are eligible for a first-class medical along with [a SODA] for defective color vision."

During postaccident interviews, the first officer's wife told investigators that she vaguely remembered hearing about her husband's color vision deficiency. She stated "I recall this was way back when he was in training for the Navy when this came up. I think it was like a blue/green problem . . . he was given a waiver for it." When asked if the first officer's color vision deficiency ever affected anything in his daily life (for example, did she or her daughters ever have to help him match articles of clothing), the first officer's wife responded, "no." Further, the first officer told investigators that he never experienced any difficulty distinguishing red and white on PAPI or VASI visual approach slope indicator) lights.

Source: AAR-04/02, §1.13

At the Safety Board's request, the first officer completed an extensive postaccident ophthalmic evaluation at the U.S. Air Force School of Aerospace Medicine (USAFSAM) at Brooks City-Base in Texas.

The USAFSAM letter concluded that the first officer's "documented proficiency" suggested that he relied on "learned strategies" other than normal color vision to determine the airplane's position when using the PAPI/VASI during an approach.

Source: AAR-04/02, §1.13.2

According to Chapter 6 of FedEx's FOM that was effective at the time of the accident, the stabilized approach corridor begins at 500 feet agl for airplanes that are cleared for a visual approach and at 1,000 feet agl for airplanes that are cleared for an instrument approach. The stabilized approach is defined as follows:

• The aircraft must have landing gear down and locked; the flaps/slats must be in the final landing configuration.
• The engines must be spooled-up[85] and steady at the proper approach setting.
• The proper descent angle and rate of descent must be established and maintained. All available landing aids (ILS, VASI, PAPI, etc.) must be used. Non-precision approaches may require a slightly steeper angle until reaching the MDA (minimum descent altitude).
• Airspeed must be stable and within the range of target speed (+/- 5 knots of target). Momentary and minor deviations are only tolerated if immediate corrections are made.

The FOM emphasized that "the procedures and parameters listed above are not merely targets, THEY ARE MANDATORY CONDITIONS AND LIMITS. ANY DEVIATION OCCURRING AT OR BEYOND THE BEGINNING OF THE STABILIZED APPROACH CORRIDOR REQUIRES A MANDATORY GO-AROUND."

Source: AAR-04/02, §1.1

According to the Safety Board's airplane performance study, when the airplane became established on the final approach course about 2 1/2 miles from the approach end of runway 9, the PAPI lights would have shown a low indication (one white and three red lights). Almost immediately thereafter, the PAPI lights would have shown a very low indication (four red lights), which would have been viewable from the cockpit for the remainder of the flight.

As the airplane descended through 500 feet agl at 1,248 fpm, 152 knots, and with engines operating at about 1.17 EPR, the captain announced that the approach was "stable." The Safety Board notes that, although the airplane's airspeed was within the target range, the airplane did not meet FedEx's criteria for a stabilized approach because its rate of descent was greater than FedEx's recommended 1,000 fpm, the engines power settings were less than the expected 1.3 to 1.45 EPR, and its glidepath was low as indicated by the PAPI light guidance. According to FedEx procedures at the time of the accident, if a visual approach was not stabilized when the airplane descended through 500 feet agl, the pilots were to perform a go-around. The Safety Board concludes that the accident approach was not stabilized as the airplane descended through 500 feet agl and that the pilots should have detected this and performed a go-around.

Source: AAR-04/02, §2.2

The Safety Board's examination of the captain's sleep history revealed evidence of a sleep deficit. The captain was off duty from about midnight on July 24 until the accident flight. However, he described his sleep during the two nights preceding the accident trip as "not really good" or "marginal" because his sleep was interrupted to take care of the family dog. The captain stated that after he learned that he was assigned to the accident trip, he got about 3 1/2 hours of "pretty good" sleep.

The CVR recording revealed that the captain made several small errors during the accident flight that suggest he may not have been fully alert.

Further, the captain's decision-making and monitoring during the approach were not characteristic of his past performance.

The Safety Board's review of crew interviews and the first officer's sleep history revealed that he reported having difficulty adjusting his sleep cycle to the reserve-duty schedule. The first officer told investigators that this was the first reserve schedule he had flown in several years and that it was difficult on his body because the sleep-wake cycle was frequently changing between day and night sleeping schedules. He stated that he normally preferred to bid schedules that allowed either all-day or all-night sleep periods over 1-week blocks.

Although the first officer's errors and occasionally deficient performance are consistent with the effects of fatigue, in the case of the failure to maintain an appropriate glidepath, the investigation revealed a potential alternate explanation. Specifically, there was evidence that the first officer's ability to use PAPI information was limited by a congenital color vision deficiency (see section 2.4.1).114 Therefore, the Safety Board concludes that the first officer's schedule and his reported difficulty adapting to frequently changing sleep cycles were conducive to the development of fatigue impairment that contributed to his degraded performance during the approach to TLH; however, there were also other factors affecting the first officer's performance (for example, his color vision deficiency).

Source: AAR-04/02, §2.3

The National Transportation Safety Board determines that the probable cause of the accident was the captain’s and first officer’s failure to establish and maintain a proper glide path during the night visual approach to landing. Contributing to the accident was a combination of the captain’s and first officer’s fatigue, the captain’s and first officer’s failure to adhere to company flight procedures, the captain’s and flight engineer’s failure to monitor the approach, and the first officer’s color vision deficiency.

Source: AAR-04/02, §3.2