January 2001 Volume 44 / Number 1
Landmark Accidents: Collision Over Cerritos
The accident that changed everything
The AOPA Air Safety Foundation is offering, as its premier free seminar series this year, a review of collision avoidance procedures and some discussion about the new technologies that are becoming available for light aircraft. Seminar locations are shown in "ASF Action" each month in the "AOPA Action" section of Pilot, each week in AOPA ePilot, and on AOPA Online. The seminar was produced by grants from the Department of Defense, the Air Transport Association, Ryan International (a builder of collision avoidance avionics), and pilot donors to ASF. A companion safety advisor will be available in January.
Some points to remember on avoiding collisions:
- Traffic patterns at nontowered airports are the most likely place for a midair collision in day VFR conditions.
- Proper patterns, looking outside, and concise, accurate radio use are essential.
- Use all available exterior lights when in the terminal area — strobes, anticollision beacons, and especially landing/recognition lights.
- Military operating areas (MOAs) are established to separate civilian traffic from military aircraft that are engaged in tactical exercises, and may be operating at very high speed and be distracted from avoidance tasks. Avoid them when hot.
- Use a sterile cockpit (no unnecessary conversation or activities when arriving or departing an airport).
- Use your transponder at all times. It's a good idea and it's the law to operate in controlled airspace (if the aircraft is equipped with an operable transponder—FAR 91.215).
- Program navigation equipment (GPS, VOR) and avionics before takeoff or before entering the terminal area (at least 10 miles out).
- Brief passengers on how to look for traffic, how to use the "clock system" to point it out, and what constitutes a threat (i.e., no movement relative to your aircraft).—BL
A standard admonishment is to watch carefully for other aircraft, lest the day be ruined by just one midair collision. Cerritos, a suburb of Los Angeles, is remembered for an accident where the ramifications went far beyond those directly involved. The midair encounter 14 years ago between a Piper Archer and an Aeromexico DC-9 changed the airspace, the equipment requirements for most aircraft, and our operating environment as few accidents have—before or since.
The AOPA Air Safety Foundation has been tracking midairs for more than a decade and recently completed a new collision avoidance safety seminar that will be offered in 200 locations during the coming year. Why are these mishaps important and why should we take the time to learn about them, especially when they are relatively rare? Collisions are one of the few emergencies in which the ability to control your aircraft may be lost in an instant, and a high percentage result in fatalities. Proper procedures, good scan technique, and some new technology will increase the odds of avoiding a run-in. Another collision between a GA aircraft and an airliner could result in serious restrictions on general aviation, regardless of who is at fault.
At approximately 11:40 a.m. Pacific Daylight Time (PDT) on August 31, 1986, a Piper Archer with a pilot and two passengers departed Torrance, California, for a VFR flight to Big Bear, California. A flight plan filed, but not activated, with the Hawthorne Flight Service Station showed the proposed route as direct Long Beach to Paradise VOR and then direct to Big Bear at a cruising altitude of 9,500 feet. Radar showed that the Piper turned east toward Paradise after takeoff with the Mode A transponder (nonaltitude reporting) set to the VFR code of 1200. The pilot did not request ATC assistance or clearance into the TCA. (In 1986, what is now Class B airspace was designated as a terminal control area or TCA.)
Aeromexico Flight 498, a DC-9 with 58 passengers and a crew of six, was inbound from Tijuana to Los Angeles (LAX) on an IFR flight plan. Flight 498 reported out of 10,000 feet at 11:46 a.m. and was instructed to contact LAX Approach Control. About one minute later, the flight reported level at 7,000 feet, and at 11:50 approach control asked for a speed reduction to 210 knots, which the crew acknowledged.
At 11:50:46, the controller advised Flight 498 of "traffic, 10 o'clock, one mile, northbound, altitude unknown." Aeromexico acknowledged the call but did not report the traffic in sight. This was not the Archer target. At 11:51:04, the flight was cleared down to 6,000 feet.
At 11:51:18, a Grumman Tiger called approach control to ask for VFR traffic advisories through the TCA at 4,500 feet. The controller assigned a squawk at 11:52:04 and asked the pilot to verify his altitude shortly after that. The Tiger pilot reported climbing through 3,400 feet, and at 11:52:36 the controller advised the Tiger that he was in the middle of the TCA and suggested, "In the future you should look at your TCA chart. You just had an aircraft pass right off your left above you at 5,000 feet and we run a lot of jets through there at 3,500."
The controller then noticed that the radar was no longer tracking Flight 498 and after several unsuccessful attempts at radio contact, notified the arrival coordinator that radar and radio contact was lost. At 11:52:09, Flight 498 and the Piper Archer collided over Cerritos at about 6,650 feet. The Archer had inadvertently penetrated the 6,000-foot floor of the TCA without a clearance. The sky was clear with reported visibility of 14 miles. There were no survivors on either aircraft, and 15 people were killed on the ground. Five houses were destroyed and seven others damaged by wreckage or post-impact fire. The DC-9's cockpit voice recorder had only one comment by the captain relative to the collision: "Oh, this can't be!"
The Archer pilot, age 53, held a private certificate and had logged a total of 231 hours. He had logged only two hours in the 90 days preceding the accident. Interviews with other pilots and flight instructors described him as "conscientious and careful." According to his primary flight instructor, the pilot was taught to "scan left, look at the instruments, scan right, look at the instruments, and repeat the procedure." The pilot was said to be familiar with the wing leveler in the Archer and used it "as it was intended to be used when looking at maps, reviewing charts, or doing other in-cockpit activities."
Another CFI who had provided the pilot with some IFR training said they had discussed sectional charts and the significance of the numbers used to show the floor of TCAs. The pilot liked to look outside and was more inclined to navigate by visual reference than by navigation aids, according to the instructor. They discussed TCAs, along with equipment requirements, and arrival and departure procedures.
The pilot moved to the Los Angeles area from Spokane, Washington, in October 1985. In December, he took a familiarization flight with a local CFI and then flew the Archer down to L.A. in March 1986. Since December, he had logged 5.5 hours on seven local flights. The pilot was just beginning to get acquainted with the complex and crowded airspace that surrounded Southern California. The Archer pilot had purchased an L.A. Sectional and an L.A. VFR Terminal Area Chart that morning and apparently was using it since—when it was later found in the cockpit of the Archer—it was folded to show the proposed route.
There was some speculation that the Archer pilot might have suffered a heart attack that led to the crash when, during the autopsy, a major blockage in the right main coronary artery was discovered. Further analysis revealed "severe atheriosclerosis but no evidence of a myocardial infarction" before the accident.
The Aeromexico captain had 10,641 hours total time, with more than 4,600 in the DC-9, and had been employed by the airline since 1972. The first officer had flown 1,453 hours and had joined the airline two years before the collision.
From paint smears and damage signatures, the NTSB determined that the Archer collided with the left horizontal stabilizer of the DC-9's tail. The stabilizer separated from the aircraft and fell intact about 1,700 feet from the main crash site. The Piper was mostly intact except for the upper portion of the fuselage, engine, and vertical stabilizer. It suffered no in-flight or post-crash fire damage. There was no evidence that either aircraft performed evasive maneuvers. The transponder was set to 1200, and the navigation radios were tuned to the Seal Beach and Paradise VORs. Based on the com radio frequencies and tapes, it was determined that the Archer was not in contact with ATC. The DC-9 caught fire after ground impact, which contributed to significant ground damage at the crash site.
After the accident, the NTSB conducted a visibility study to determine if either aircraft was visible to the other before the collision. The viewing angles for each airplane were plotted at five-second intervals. The study showed that from 11:50:56 until 11:52:01 the captain of the DC-9 could see the Piper and was in the best position to observe the traffic. It was located about 15 to 30 degrees left of the design eye reference point or center of the windshield.
The first officer could see it about 50 percent of the time. The Piper pilot could only see the DC-9 by looking to the far right through the copilot's windshield. The Piper's front and rear seat passengers would have had a better view but presumably were not trained to spot other aircraft.
Massachusetts Institute of Technology's Lincoln Laboratories studied pilots' abilities to spot other aircraft in flight. This was part of an ongoing program incidental to the accident, but the results were sobering. The test counted the number of times pilots acquired or failed to identify an intruder aircraft and at what distance. Subject pilots were not informed that there would be intruder aircraft or that scanning behavior was the focus of the study. Visual acquisition occurred only 56 percent of the time and the median range was just under one mile. An experimental group using a prototype traffic collision avoidance system acquired targets 86 percent of the time at a median range of 1.4 miles.
Aeromexico procedures and training
The NTSB reviewed the flight operations manual with specific emphasis on collision avoidance. It generally repeated the right-of-way rules of FAR Part 91. Recurrent training was essentially identical to that in use by U.S. airlines at the time. This consisted of two simulator and ground training sessions a year, followed by an en route check flight, normally to a U.S. destination.
Company policy required the cockpit door to be closed during flight, and jump-seat privileges were restricted to other company pilots, company technical personnel, and flight standards inspectors. There was no requirement to limit conversation to operational topics, as FAR Part 121 rules mandate today (sterile cockpit), or to restrict flight attendant access during takeoff or landing. Completion of company paperwork was prohibited on short legs (such as the flight from Tijuana) and was required to be completed at cruise altitude on longer flights. Aeromexico also used abbreviated before-landing checklists on short flights to minimize crew distraction.
The autopilot was recommended for all flights and was normally activated at 2,000 agl during climb and remained on until final approach. An Aeromexico check pilot testified that the autopilot would likely have been engaged at the time of the accident, although this could not be verified from the wreckage.
An unintelligible female voice was heard on the cockpit voice recorder at 11:50:05, minutes prior to the collision. The check pilot testified that it was probably a flight attendant advising the captain that the cabin was prepared for landing or part of a cabin announcement that could be heard through the closed cockpit door. The NTSB found no evidence of improper procedure by the Aeromexico crew.
Terminal control radar and ATC procedures
At the time of the collision, Los Angeles Approach Control used two airport surveillance radar systems (ASR-4 and ASR-7, commissioned in 1964 and 1973, respectively). They were augmented by the ARTS III system, which included a conflict-alert program to warn controllers of a potential collision situation. A primary target, or "skin paint," of all aircraft visible to the radar was displayed as a dot of light that faded until a subsequent sweep on the antenna refreshed the target brightness. Transponder-equipped aircraft had a beacon slash that appeared next to the primary target. A 1200 code from a Mode A transponder was depicted as a nonfading triangle if the controller did not suppress the VFR code to declutter the scope. The beacon slash could be suppressed depending on the configuration of the equipment. Mode C targets appeared as a square with the altitude displayed in a three-digit block next to the target. Tracked targets, aircraft with assigned codes, had a data block attached showing aircraft ID, groundspeed, and altitude. Flight 498 was depicted in this manner.
Radar can track only a finite number of targets before becoming overloaded, but the system was not saturated at the time of the collision. There had been some limited radar maintenance outages, but the backups functioned properly and all equipment was determined to be operating properly at the time of the accident.
Primary radar returns can be affected by temperature, pressure, and humidity, known as the refractive index, which can bend or even split the radar beam. Under certain conditions a hole in coverage can occur where a primary target could be missed. On the day of the accident, the refractive index was high at the altitudes where the collision occurred. The NTSB could not determine if a primary target was present on the controller's scope. The secondary beacon target was visible, although not as prominent as a typical primary return. Local procedure required a setting on the radar unit to suppress untracked VFR traffic beacon slashes to reduce target clutter. The NTSB considered this procedure a factor in the accident because the controller failed to spot the small secondary target triangle.
The NTSB concluded that the controller, while not overly busy, had been distracted by coordinating a change for Aeromexico 498's arrival into LAX and by the sudden appearance of the Grumman Tiger, which posed a potential threat to other LAX inbound traffic. Because of this, the controller apparently failed to see the Piper secondary target and advise Flight 498 of the traffic. The controller had no reason to presume an altitude conflict since most VFR traffic operated below the floor of the TCA and there was no other authorized traffic in the area. However, his failure to provide a traffic advisory to the DC-9 was considered a factor.
The probable cause was determined to be limitations of the air traffic control system to provide collision protection including procedures and automated redundancy. Contributing factors were the inadvertent and unauthorized entry of the Piper into the L.A. TCA and limitations of the see-and-avoid concept.
All aircraft in visual conditions, including those operating under IFR, must use the see-and-avoid concept. Nevertheless, the NTSB felt there were some inherent limitations as shown by the MIT study. Pilots miss many targets because of aircraft structure, physiological limitations, distraction, procedural errors, fatigue, and lighting conditions.
Because of this accident and data from a variety of other sources, the NTSB recommended, and the FAA ultimately adopted, the requirement for a traffic alert and collision avoidance system (TCAS) on all air carrier and regional airline aircraft. There can be no doubt that this has reduced the potential for collisions, but TCAS can only see transponder-equipped aircraft. This is why it is essential and required that all aircraft with an operable transponder use it religiously.
There were calls to ban light aircraft from high-density areas, and AOPA worked hard to balance the media and regulatory demands for elimination to a more reasonable solution. The Air Safety Foundation mounted a major education campaign in Southern California in the two months following the accident to educate GA pilots about their responsibilities and the procedural requirements to operate in and around TCAs.
The airspace surrounding high-density airports was consolidated from two classes of TCAs into one. Mode C transponders became a requirement not only to operate within a TCA, but also to fly within a 30-mile radius of the primary airports—the 30-mile veil. Flights within sanitized airspace could be warned of intruder aircraft that, before the requirement, were assumed to be below the floor.
The Mode C rule also enhanced ATC's ability to track intruders and enforce the entry requirements. The number of airspace enforcement actions by the FAA's L.A. Flight Standards District Office jumped tenfold after the accident. That vigilance continues today and most GA pilots have learned that Class B and C transgressions should be avoided, lest we have another Cerritos.
Links to additional information about airspace and collision avoidance may be found on AOPA Online. See also the index of "Safety Pilot" articles, organized by subject. Bruce Landsberg is the executive director of the AOPA Air Safety Foundation.