By common consent, above, SA includes awareness of other traffic to a greater or lesser extent. I think many of us delude ourselves about just how much, or little, we know of it.
The Air Accident Investigation Branch (AAIB) referred to scan problems in its report on the Cirrus/Grob Tutor collision (2 fatalities in the Grob) in 2009.
‘1.18.11 See-and-avoid
Various studies have highlighted the limitations of the see-and-avoid method of preventing mid-air collisions. .
‘1.18.12 Alerted see-and-avoid
‘ . . . tests determined that one second of alerted visual search is as effective as eight seconds of unalerted search . .’
The report is available to download from:
Air Accidents Investigation: 5/2010 G-BYXR and G-CKHT :
An article by a CAA man included: “. . . the visual scan is the ONLY way to spot likely collision hazards”. It’s official, so it must be right. But the same article made clear that visual scan has many limitations.
(A visual search in the absence of traffic information is less likely to be successful than a search where traffic information is provided because, obviously, knowing where to look greatly increases the chances of sighting the other aircraft. Various field trials have shown that in the absence of a traffic alert, the probability of a pilot sighting a threat is generally low until a short time before impact. Traffic alerts were found to increase search effectiveness by a factor of eight (in relation to the ‘alerted’ threat only).
Just how good is the standard unalerted lookout?
To quote from the AAIB report again:
‘Lookout to the front and scan above and below the horizon, then attitude and instruments…. Move the eyes around the horizon in a series of steps (normally to the right initially), scanning up then down at each point…..continue the scan back to the tailplane and then look above and behind over the top and back to the front.’
The AAIB report continued to observe, and quote references to the effect, that a traffic scan takes time – at least one second at each fixation, so from 54 seconds upwards. A study they referred to included these cumulative time periods, in seconds, to react to an observed collision threat:
See object 0.1
Recognise aircraft 1.1
Recognise collision 6.1
Decide on action 10.1
Muscle reaction 10.5
Aircraft lag time 12.1 secs
Is even 54 seconds per complete scan good enough to detect in time a closing threat from, say, your 8 o’clock (only one such glance per 54 seconds, at best)?
Chris N