reducing "Crew-caused"
approach and landing
accidents 

Pilot-in-charge Monitored Approach

2015 A320 snowstorm CFIT Halifax Canada

Brief account : 

Inadequacies and omissions in the official investigation report make it difficult to be precise as to exactly what happened but it is possible to deduce many aspects from other public domain material. For that reason a detailed critique of the report is available at a link at the bottom of this page.

The night non-precision (localiser only) approach was being conducted in marginal weather conditions of low cloud, low visibility, crosswinds and heavy snow, using the autopilot "flight path angle" descent angle mode. The actual flight path was steeper than the crew had intended but this was not detected. At MDA some lighting was called out by the F/O: the Captain committed to landing and started looking for the visual cues, leaving the autopilot engaged.  Just above the threshold elevation the autopilot was disconnected and a go-around was commenced shortly after.  The aircraft struck the ground in the approach lights and was destroyed after bouncing onto the runway, fortunately with no fatalities.

Crew-related factors : 

This accident is a classic example of the factors that make transition to visual cues in marginal conditions so hazardous when using "traditional" procedures.    

Because of the low (ISA-20) temperature it was necessary for the crew to calculate the appropriate altitudes and selected descent angle.  For reasons that are not fully explained, the aircraft flew a steeper than planned descent, but as the crew were not expected to (and did not) make further crosschecks after leaving the final approach fix, this was not detected. 

The weather was extremely marginal and had been above and below minima at various times prior to the accident. The runway had only the most basic approach lighting, which had not been set in accordance with the crew's expectations, and the runway itself was likely to generate visual illusions.     

At the MDA the First Officer saw sufficient cues to meet Canadian regulations for visual reference, leading the Captain to commit to landing before he had actually confirmed the aircraft's position and trajectory. Both pilots then remained head up during a "conversation" about the cues, while the aircraft continued under autopilot control below its minimum permitted usage height.  The autopilot was disconnected just above runway level and a go-around was subsequently attempted, but the aircraft hit approach lights prior bouncing on the ground and then onto the runway. 

The aircraft went below runway level prior to impact but sloping terrain before the runway and the runway's own slope were the only reasons the initial impact was not catastrophic. As a result the accident was not fatal, although the aircraft was destroyed. 

A significant factor in this accident appears to be that the Canadian required visual reference for descent below MDA is far less demanding than that assumed in ICAO standards and frequently referred to on this website. It does not require the pilot to HAVE MADE an assessment of the aircraft's actual position and rate of change of position by the DH/MDA before descending further, only to decide that cues have been identified that make an assessment possible, even if it results in a conclusion that the position and trajectory are unsafe. This is not mentioned in the report.

A detailed critique of this and many other aspects of the report is available here.

The "traditional" cockpit procedure and call-outs used also meant that the initial decision as to visual reference adequacy was delegated to the First Officer. These two factors combined meant that the intended protection which is normally provided by altitude and visibility minima against instrument flight path errors (such as that experienced during the descent from the Final Approach FIx), in the conditions encountered, was simply not provided to the passengers and crew of this aircraft.

Crew procedural aspects: 

If the crew had been using a PiCMA procedure with the First Officer flying the approach, and with the ICAO required visual reference: 

  • It is possible that the better workload distribution would have provided some opportunity to detect the initial deviation from planned descent path.
  • There would have been a call from the F/O at 100 ft. above the MDA; 
  • The Captain would have started assessing the visual cues well above the MDA rather than some distance below;
  • There would have been an imperative call of "DECIDE" from the F/O at the MDA, rather than an advisory one of "minimums"
  • It is almost certain that the Captain would have called for a go-around at the MDA. A go around was subsequently initiated after the autopilot was disconnected some 23 seconds later, and it is highly unlikely that the cues would have been significantly more favourable to a "land" decision at the MDA and subsequently deteriorated to such an extent.
  • The go-around would have been the one briefed for, executed by the F/O using the autopilot, rather than an attempted manual one by the Captain.  
  • In the unlikely event that the Captain had decided to continue based on the cues available, the trained requirement to positively take control would have provided the salient trigger to also disconnect the autopilot at its minimum use altitude; 
  • In the unlikely event of continuation without that disconnection, the F/O would remained "head down" and likely have observed the "disconnect autopilot for landing" annunciation;
  • The F/O would also then possibly have been alert to the very rapid decrease in radio altitude prior to impact due to the sloping terrain, and made his call for ago-around earlier.    
Type: 
A320
Where: 
Halifax Canada
Expected weather: 
Instrument
Pilot in charge: 
Capt
Early transition: 
Yes
Go-around : 
Below DH/A
Damage: 
Serious
PicMA potential: 
Major
Year: 
2015
Time: 
Night
Deterioration: 
No
Vert Guidance: 
None
Both Head Up: 
Yes
LoC: 
No
Operator: 
Air Canada
Fully prepared: 
Yes
Actual Weather: 
Snow
Autopilot : 
Y
CCAG: 
Normal
Source material: