Hmmmm this seem to have dropped out of sight but here goes again;
Been waiting for this to appear on the website to save me typing it out the hard way.
It is IMHO the final word on the fallacy of "see and be seen" in the modern environment at least the one that I work in.
See-And-Avoid A Dangerous Way To Separate High- and Low-Performance Aircraft
Jan 17, 2007
By Patrick Veillette, Ph.D./Business & Commercial Aviation
The smoke plume on the east side of Los Angeles on the afternoon of Aug. 31, 1986, was clearly visible from the balcony of our apartment adjacent to LAX. The local news was reporting that an airliner had crashed into the suburb of Cerritos. My roommates, all Los Angeles-based pilots from a half dozen airlines, stood on the balcony watching that sick black plume rise. No one said the obvious, that the dark column marked the place of death for many.
We subsequently learned that a pilot of a Piper PA-28 had errantly wandered into what is now Class B airspace without a clearance and without an operative altitude encoding transponder. The flight crew of the Aeromexico DC-9 had only the slimmest of chances to spot and avoid the Cherokee, and unfortunately luck wasn't with them on that day. When the NTSB finalized its report, it found limitations with the "see-and-avoid concept to ensure traffic separation" as a contributing factor to the disaster that took 82 lives.
That inflight collision was eerily reminiscent of another eight years earlier when a Pacific Southwest Airlines Boeing 727 lost sight of a Cessna 172 while on approach to San Diego's Lindbergh Field. The flight crew of the fast moving 727 lost sight of the Cessna and thought they had actually passed the slower aircraft. Unfortunately they had not and 138 people died in the collision that followed. A photograph of the last moments of the 727's dive serve as a gut-wrenching reminder of the inadequacy of eyeballs as a primary collision avoidance tool.
On Jan. 15, 1987, a Mooney M20 pilot was practicing holding patterns near the extended approach path into Salt Lake International Airport and inadvertently entered SLC's airport radar service area. Approach Control wasn't receiving an altitude read-out from the Mooney's transponder and unfortunately the practice holding pattern roughly coincided with the traffic pattern for Salt Lake City Municipal No. 2 Airport, a general aviation facility located roughly six miles to the south of SLC, thus adding further confusion as to the exact location of the Mooney. At the same time a SkyWest SA-227 Metroliner was being vectored to the final approach course for SLC, its pilots attempting to locate traffic pointed out by ATC. Despite the good weather, the Metro pilots never spotted the Mooney because the two aircraft collided over Kearns, Utah, raining aircraft parts on the residential neighborhood and killing the 10 people aboard the two aircraft but somehow avoiding any loss of life or injury to people on the ground.
Just five days later, a U.S. Army U-21 King Air collided with a Piper Chieftain near Independence, Mo. Six people died in that crash. The NTSB determined that the probable cause of the accident was the failure of radar controllers to detect the conflict and to issue traffic advisories or a safety alert to the flight crew of the U-21, and the inadequate vigilance of the pilots. However, the Safety Board also cited deficiencies of the see-and-avoid concept as a primary means of collision avoidance, and the lack of automated redundancy in the air traffic control system to provide conflict detection between participating and nonparticipating traffic.
Shortly thereafter, the NTSB recommended the FAA "expedite development, certification and production of various low-cost proximity warning and conflict detection systems for use aboard general aviation aircraft."
Unfortunately, that recommendation was still unrealized when five years later, on Sept. 11, 1992, an MU-2 pilot attempting to pick up an IFR clearance while departing VFR from Greenwood (Ind.) Municipal Airport, collided with a Piper Saratoga inbound for the same airport. The controller pointed out the Greenwood airport to the PA-32's pilot when the aircraft was approximately three miles out, at which time VFR radar service was terminated. The MU-2 had just departed the airport and asked ATC for his IFR clearance to CMH. The controller looked away from the radar screen to locate the proper flight progress strip and did not see the fast-moving turboprop depart from Greenwood. The controller then issued a squawk code and altitude to establish radar identification. It was during those seconds that the incoming PA-32 and outbound MU-2 collided. The Mitsubishi pilot and four passengers and the Saratoga pilot were killed. Two people on board the Piper were seriously injured.
The NTSB determined that the probable cause of the accident was "the inherent limitations of the see-and-avoid concept of the separation of aircraft operating under VFR that precluded the pilots from recognizing a collision hazard and taking actions to avoid the collision." The report concluded that the accident "again underscores the need for low-cost proximity warning and conflict detection systems for use aboard general aviation aircraft."
Now, 20 years have passed since the NTSB issued its original recommendation to expedite development, certification and production of low-cost proximity warning and conflict detection systems for general aviation. And while some systems are in place, "see-and-avoid" remains the primary means of separation between high-performance turbine and low-performance general aviation aircraft sharing the same airspace.
While a "Mode C veil" has been erected 30 nm around Class B airports that seems effective at preventing inflight collisions there, a system-wide solution for preventing collisions in other airspace, especially where high-performance turbine aircraft mix with lower performance general aviation aircraft, has yet to be implemented.
In the meantime, the wreckage continues to pile up. On April 4, 1998, a Cessna CE525 collided with a Cessna 172 over Marietta, Ga., killing five. On June 23, 2000, a Learjet 55 collided with an Extra 300S over the busy skies of Boca Raton, Fla.; four people died as a result. On Oct. 17 of that same year, a Gulfstream GIII collided with a King Air C90 while both aircraft were on approach to Van Nuys (Calif.) Airport; the pilots managed to land their aircraft without injury to anyone aboard. Most recently, a Hawker 800XP descending into Reno collided with a Schleicher ASW-27 sailplane near Minden, Nev., a popular soaring location. Somewhat miraculously the glider pilot managed to parachute to safety and the crew of the damaged Hawker was able to execute a single-engine, gear-up emergency landing at the nearby Carson City Airport. Everybody walked away.
FAR Part 91.113 places the responsibility to see and avoid other aircraft squarely on the pilot, stating, "When weather conditions permit, regardless of whether an operation is conducted under IFR or VFR, vigilance shall be maintained by each person operating an aircraft so as to see and avoid other aircraft."
Using this regulatory precedent, the NTSB has found "failure to see and avoid, inadequate visual lookout, or failure to maintain visual and physical clearance" as the probable cause in 94 percent of the inflight collisions. The Safety Board can keep issuing such causal statements, but that won't change the underlying problem that the see-and-avoid concept isn't reliable as a primary or even secondary means of separating traffic.
Harold Marthinsen, former director of the Air Line Pilots Association (ALPA) safety engineering department, asks, "Do pilots really need a regulation telling them to avoid midair collisions? Pilots have a self-vested interest in preventing that from occurring. It is publications like the FAA's Advisory Circular on collision avoidance that help perpetuate the idea that all you have to do is pay attention, look out the windshield, and you won't have a midair collision. Rather, the FAA should be telling pilots how dangerous the see-and-avoid concept really is as a means of separating aircraft."
See-and-avoid involves a number of steps, all of which are inherently prone to error. First, the pilot must be looking outside the aircraft. Second, pilots must search the visual field and detect objects of interest, most likely with their peripheral vision. Next, the object must be looked at directly so it can be identified as an aircraft. If the aircraft is identified as a collision threat, the pilot must decide what evasive action to take and then follow through correctly and in a timely fashion.
So how well does the concept work? According to Craig Morris of the DOT's Bureau of Transportation Statistics, each year there are an average of 15.6 midair collisions in U.S. civil aviation, and that number may only hint at the scope of the problem. In an average year, NASA's Aviation Safety Reporting System (ASRS) receives approximately 577 pilot reports of near in-flight collisions between various types of aircraft. Specifically with regard to business jets, the FAA's near midair collision (NMAC) database included 226 NMAC reports filed by pilots of business jets for a recent 10-year period. In addition, the ASRS database for the same 10-year period had 806 reports of near midair collisions involving business jets. Just how close did some of the aircraft approach each other in these reports? Half of the reported incidents had a separation of less than 500 feet; some were even closer than that. And those are just the reported events. We really don't know how many close calls went unreported or how many pilots were simply unaware of a near disaster.
Those of us who fly the line don't need a bunch of statistics to tell us that flying into Martha's Vineyard or Nantucket (or Santa Monica, Van Nuys, etc.) on a VFR summer weekend is akin to running a gauntlet.
Writing in Aviation, Space and Environmental Medicine, a peer-reviewed journal of the Aerospace Medical Association, Morris held that "The 'see-and-avoid' concept has considerable physical and behavioral limitations such that pilots cannot reliably see and avoid conflicting traffic." A panel of reviewers would not permit an author to make such a statement unless it was backed by a preponderance of respected scientific research. Morris is among many respected aviation authorities who have expressed concerns regarding the limitations of the see-and-avoid concept.
The Australian Transport Safety Bureau issued a lengthy research report entitled "Limitations of the See-and-Avoid Principle" that stated, "Numerous limitations, including those of the human visual system, the demands of cockpit tasks, and various physical and environmental conditions, combine to make see-and-avoid an uncertain method of traffic separation." In addition, the Australian Bureau of Air Safety Investigation stated that "see-and-avoid is completely unsuitable as a primary traffic separation method for scheduled services."
Marthinsen stated in the International Society of Air Safety Investigators (ISASI) Forum (December 1989) that "See-and-avoid was originally a maritime concept developed for slow moving ships that is now out of place in an era of high-speed aviation." He went on to say that "No one is suggesting that see-and-avoid is not a useful tool for general aviation airplanes operating at uncontrolled airports -- it's the only thing available, and it works most of the time in this environment for which it was adapted. But it should not be relied upon to separate high-performance aircraft from lower performance general aviation aircraft -- a situation for which it was not designed."
There is a big difference between seeing an object under laboratory conditions vs. detecting another aircraft on a possible collision course out in "the real world." The NTSB considers 0.2 degrees as the threshold angle for detection (that is, the viewer could detect an object 20 feet wide at a distance of one mile -- Ed.), although Marthinsen says that figure should be increased by a factor of two or three for targets with low contrast or difficult patterns. The former ALPA director believes no one minimum visual angle can be specified because none can accommodate all variety of conditions such as haze, visible moisture or smoke. He noted that many visual acuity studies were done under favorable light conditions and with the subject staring directly at the object, in which case the image is focused directly on the fovea of the eye. The fovea contains the majority of the eye's cones and is responsible for the sharp vision we use in daylight conditions. If an object is six degrees from the fovea, that is, six degrees off a direct line by the viewer, it would have to be twice the size of an object directly in line in order to be detected. The threshold acuity drops off very rapidly for objects placed at angles away from the fovea. Further, all of these theoretical studies were conducted with fixed targets and viewers, never in motion.
FAA Advisory Circular 90-48C provides military-derived data on the time required for a pilot to recognize an approaching aircraft and then execute an evasive maneuver. The calculations do not include search times but assume that the target has been detected. The total time to recognize an approaching aircraft, recognize a collision course, decide on action, execute the control movement and allow the aircraft to respond is estimated to be around 12.5 seconds. Therefore, to have a good chance of avoiding a collision, a conflicting aircraft must be detected at least 12.5 seconds prior to the time of impact. It should also be noted that this study is over-optimistic for aircraft reaction time because the military study was in reference to agile fighter aircraft. The higher inertia and lesser maneuverability of civilian transports would add considerably to aircraft reaction time. The NTSB has used 15 seconds as the absolute minimum time for detection, evaluation and evasive action if the collision is to be avoided. Other studies suggest somewhat higher values.
Target detection is primarily a function of target size and target contrast, with size being, by far, the more important parameter in the ability to detect other aircraft. Unfortunately business jets and general aviation aircraft are on the "small" side of this spectrum and thus would be very difficult to see at sufficient distances to avoid a collision.
Detecting a target at jet speeds leaves very little time to see and avoid. For example, a jet descending at roughly 400 knots groundspeed covers 1.39 nm in 12.5 seconds. Let's say that the intersecting general aviation aircraft is on a roughly perpendicular flight path, thereby presenting more surface area to be detected and making it easier to see. Both the visual angle and its rate of change remain very small until imminent impact. At seven seconds to impact, a 40-foot-long aircraft would subtend only a 0.5-degree angle, which is still very small.
Walton Graham and Robert H. Orr, in a research paper entitled "Separation of Air Traffic by Visual Means: An Estimate of the Effectiveness of the See-and-Avoid Doctrine" and published in Proceedings of the IEEE (Volume 58, 1970) stated, "As speed increases, the effectiveness of 'see-and-avoid' greatly decreases. It is estimated that see-and-avoid prevents 97 percent of possible collisions at closing speeds of between 101 and 199 knots but only 47 percent when the closing speed is greater than 400 knots."
Morris continues, "Pilots can find it physically impossible to see converging traffic, especially when climbing or descending. Also, because human information processing is biased toward detection of contrast and sudden change, the small, motionless, camouflaged target projected by a rapidly converging aircraft is difficult to detect within the random and narrow window of opportunity to see it."
The human visual system is particularly attuned to detecting movement but is less effective at detecting stationary objects. Unfortunately, an aircraft on a collision course will usually appear to be a stationary object in the pilot's visual field. From each pilot's point of view, the converging aircraft will grow in size while remaining fixed at a particular point in the windscreen.
FAA Advisory Circular 90-48C recommends scanning the entire visual field outside the cockpit with eye movements of 10 degrees or less to ensure detection of conflicting traffic. The FAA estimates that approximately one second is required at each fixation. Thus, to scan an area 180 degrees horizontal and 30 degrees vertical could take 54 such fixations at one second each. A jet descending at an approximate groundspeed of 360 knots would cover roughly six miles per minute, so a pilot who faithfully follows the procedure exactly as spelled out in the Advisory Circular would see an entirely different kind of scene before completing the scan. Marthinsen has a rather negative opinion of the FAA scanning technique, stating, "It is incompatible with the physiological capability of the human eye. The time it would take to scan is substantially longer than the time available to see-and-avoid in many of the midair collision accidents."
Do pilots practice the recommended scanning pattern? According to research cited in the Australian Transport Safety Bureau's report, "Visual scans tend to be unsystematic, with some areas of the visual field receiving close attention while other areas are neglected. Areas of the sky around edges of the windscreens are generally scanned less than the sky in the center, and saccades [motion of eye between fixations -- Ed.] may be too large, leaving large areas of unsearched space between fixation points."
Furthermore, researchers have known for many years that in the absence of visual cues, the eye will focus at a relatively short distance. In an empty field such as a limitless blue sky, the eye will focus at around 56 centimeters (1.8 feet). This effect is known as "empty field myopia" and can reduce the chance of identifying a distant object. Because the natural focus point is around a half meter (1.6 feet) away, it requires an effort to focus at greater distances, particularly in the absence of visual cues.
A U.S. study in 1976 found that private pilots on VFR flights spend about 50 percent of their time in outside traffic scan during cruise flight, although this drops off to 40 percent during departure and approach. Even motivated pilots under ideal conditions frequently fail to sight conflicting traffic. A research project conducted by John W. Andrews of the Massachusetts Institute of Technology's Lincoln Laboratory involved 24 general aviation pilots flying a Beech Bonanza on a VFR cross country. The pilots were not aware that their aircraft would be intercepted several times by a Cessna 421 flying a near-collision course. The pilots spotted 36 out of 64 encounters, a 56-percent detection rate. Keep in mind that this study involved comparatively low-speed traffic operating under ideal detection conditions. The MIT study concluded that the ability of pilots to detect aircraft on near-collision courses is not great.Capt. Harry Orlady, a well-respected human factors researcher and former United Air Lines pilot, published a paper with the National Research Council estimating that airline pilots spend about 20 percent of their time in outside scan. A United-ALPA study of airline crews determined that "no one is looking during climb at least 62 percent of the time and 52 percent of the time during descent."
cont'd