2015 A320 fog undershoot Hiroshima Japan
The A320 was severely damaged after touching down short of the runway threshold at night.
The crew conducted an RNAV approach to R/W 28 after changing from an initial plan to land on r/w 10. They noted that a "hump" in the runway resulted in the far end being invisible after touchdown, but apparently did not discuss the potential for visual illusions due to the initial 0.5% up runway slope, and the "black hole" surroundings of the airport with a deep valley just before the runway threshold.
After seeing the runway from about 6 miles, the Captain (PF) disconnected both autopilot and flight director at about 1000ft. Both pilots were evaluating the visual cues and the descent rate increased.
The crew were advised of RVR 1700m about a minute before reaching DH, but in fact the visibility was significantly decreasing. By the time the DH was reached visual cues were at least partially lost, but the Captain elected to continue. He eventually called for a go-around, but too late to avoid striking the approach lights, the runway 10 localiser antenna and other facilities.
The aircraft subsequently left the side of the runway, fortunately stopping just short of a cliff, otherwise there probably have been significantly more casualties.
Although the crew had discussed "bad weather" it appears that the Captain expected reasonable visibility conditions, which seemed to be confirmed by the fact that the runway was visible from about 6 miles out, although with a "strange appearance" and possible cloud about.
He later said that he "could have made a little more careful approach if he had received the information about the weather worsening. He remembers that he was not informed of RVR value but only informed of wind information when the Aircraft was cleared for landing. He could not anticipate the meteorological conditions might deteriorate."
It is evident that the Captain elected to transition to visual references far too early for the actual conditions, having disconnected both autopilot and flight director at about 1000ft aal, and was expecting the "good visibility" to be maintained. Subsequently these cues began to become "a bit ambiguous due to cloud". From about 900 feet aal. the aircraft started drifting below the correct approach path, being the equivalent of 3-4 reds PAPI at DH.
Both pilots were head-up for a significant portion of the final approach, and when the visual cues started to disappear the Captain attempted to revert to some form of instrument approach, including asking the F/O (PM) for radio altimeter readings. These were extremely misleading due to the steep valley immediately under the final approach - for example, the "500" RA reading came only 5 seconds before "40", just before impact.
There were discrepancies in what the 2 crew members said they could see: the Captain maintained that from DH "he could continuously see the runway and never lost the sight of it". The F/O is recorded on the CVR just above DH as saying "getting invisible" followed by "minimum" and then "runway not in sight", to which the Captain's response is "wait a second". His eventual go-around call did not come until some 24 seconds later.
If the crew had been using a PicMA procedure, it is likely that
1) The Captain might have been significantly better prepared for the actual conditions, as he would have been responsible for the ATC communications including the ATIS and final approach RVR information.
2) It is likely that the final approach path to DH would have been more accurately flown by the F/O using the autopilot.
3) The Captain would have been far better placed to study the varying visual cues approaching DH. In particular it is likely that the PAPI and close runway area became invisible in the thick fog patch around touchdown, while the far runway remained visible, which might well have triggered the Captain to ask the tower for an update.
4) The F/O would have called "Decide" to signal arrival at DH, and the end of the Captain's assessment of visual cues to confirm the aircraft's position and trajectory. The actual automated and F/O call of "minimums" did not trigger a decision, but allowed the Captain to continue in the hope that the cues would improve.
5) The F/O was required by the actual procedure to call for a go-around and if necessary take control of the aircraft because he could not see the cues, but he failed to do so because of the cockpit authority gradient. If a PicMA procedure had been in use, the Captain would have had to positively take control in these deteriorating conditions.
6) At DH the aircraft would likely have been closer to the runway, with better slant visibility to the approach and runway lights.
7) In these circumstances the F/O would have concentrating on continued instrument monitoring to draw attention to any build-up in descent rate or flight path angle. As it was his attention was apparently split between normal instrument monitoring, external cues and making the misleading and irrelevant radio altimeter callouts.
8) Even if the Captain had taken control at DH and then subsequently gone around due to a deterioration, the aircraft trajectory would have been safer.