Imagine that you're sitting in a cockpit of a Boeing 737 below minimums on final approach to a remote island airport and below you there's nothing but crystal clear turquoise water. Now there are some menacing clouds up ahead, but the runway is in sight. The aircraft is fully configured and everything tells you that this will be just another routine landing. And then a violent rainstorm hits and you lose all sight of the runway ahead. What would you do? Well, on the 28th of September 2018, the crew of Air New Guinea Flight 73 found themselves in exactly this situation. And the decisions that this crew then made would set in motion a chain of events that nobody could have anticipated. Events that were shocking even to me. So, let me now tell you the story of what led up to this and its ultimate conclusion because it's much more complicated than it first looks. Air New Guinea is the national flag carrier of Papa Newu Guinea, a country occupying the eastern half of the island of New Guinea, which is one of the most remote and rugged regions on Earth. The airline was founded in 1973, actually 2 years before Papa New Guinea gained its independence. And by 2018, the airline had built up a very good safety record, having operated for a full 45 years without a single fatal accident. And that record was not an accident. pun intended. The airline leaned heavily on long-standing relationships with Australian aviation experts, and with that came a supply of experienced Australian pilots and technical staff, all of whom also helped establish a healthy airline safety culture quite early on in Air New Guiney's development. Now, to understand why the airline was operating routes like the one that we're going to look at today into comparatively small airports, we also need to talk briefly about the federated states of Micronesia and why they are an unusual environment for any airline to operate in. Micronesia consists of around 600 small islands and atolls whose water cover nearly 3 million square kilmters of the western Pacific Ocean. For reference, that's close to the size of India. But the total land area of all of these islands put together is a bit smaller than New York City. Scheduled international routes into Micronesia are heavily dominated by international carriers from other countries. But the islands themselves are connected to each other and to the outside world almost entirely thanks to air travel with Air Newini providing some of these scheduled services. Flight 73, the route that was being flown on this day, was scheduled from Pompei International Airport, close to Balakir, the capital of Eronia, over to Shook International with an onboard later service to Port Morrisby. It had been only introduced two years earlier in 2016 as part of an agreement between Papa New Guinea and Micronia in order to boost tourism. Now initially at least this was not a commercially profitable route and as such airini received a subsidy from the Micronesian government to continue operating it. So on the 28th of September it was carrying just 35 passengers and 12 crew bringing the total souls on board to only 47 meaning that the aircraft would only be about 1/5 full which wasn't unusual on this route. Juke International Airport, the destination for this particular leg of flight 73, had one single runway, just over 6,000 ft or 1,830 m long, which is well within the performance margins of a Boing 77, especially one that is super light. But it still doesn't leave much room for error, particularly when you consider that the runway has no runway and safety areas. Both of the runway ends went straight up to the coastline. So whichever side the aircraft approached from, if anything would go wrong, it would end up in the water if it couldn't stop within the available paved distance. And that in turn meant that every approach into Chuke demanded absolute precision, which is kind of important to keep in mind. Another complication here is that even though Chuk is an international airport, it has no control tower. Pilots manage their own traffic flow by broadcasting their position on a common traffic advisory frequency or seat communicating directly between each other. Now this is quite normal for smaller airports in this part of the world but it does add some extra workload and need for vigilance by the pilots. Then there's also the weather which can be especially critical in a region like this. Juke sits deep in the tropics in one of the most convectively active regions on Earth. Which means that towering cumulan nimbus clouds capable of producing torrential rain, lightning, and severe turbulence can be quite common and also form quite quickly here. This means that they can quickly turn what was a clear approach path just minutes earlier into an impenetrable wall of cloud and heavy rain. But for crews who regularly operate in this region, that's just part of life. It has always been like that. and it will continue to be. So in summary, this was a destination with a short runway, no overrun area, no tower to look into things if something would happen, and heavily convective weather and pretty unforgiving place which demanded precision, attention, and adherence to rules from its operating crews. Now, the aircraft that was going to be used for flight 73 on this day was a Boeing 737800, which is the very plane that I have spent nearly all of my 21-year pilot career flying. And this particular one was registered as Papa 2, Papa X-ray Echo. It had been manufactured in April of 2005, so it was just over 13 years old on this day. And it had during that time accumulated just over 37,000 flight hours and 14,788 cycles. It was in a good technical condition with no outstanding mechanical defects or system failure. So by every available measure, it was fully airworthy. Now Air Neonini didn't actually own this particular aircraft outright. They had leased it in 2013 from Loft Leader, an Icelandic lesser. And unlike a conventional dry lease, this was an aircraft and maintenance arrangement, which meant that Loft Leir retained responsibility for much of the aircraft's maintenance. So under this agreement, one of Loftle's own maintenance engineers was required to accompany the aircraft when it flew outside the borders of Papa New Guinea. So on this flight, that engineer was sitting on the cockpit jump seat. And during the approach into Chuke, he did something that would turn out to be extraordinarily valuable to the investigation. He took out his phone and started filming. You see, cockpit voice recorders and flight data recorders are obviously standard equipment on modern airliners and provide investigators with a wealth of different information, but they do not record video, which is actually quite a hot topic in the aviation industry and something that I've covered in an episode over on the Mentor Now channel. So in the case of this flight, the engineers recording gave the investigators an invaluable visual record of the cockpit environment and flight instruments, and it would therefore become an example of what all air accident investigations could look like if cockpit cameras were available. Filming from jump seat, the maintenance engineer captured a clear view of the cockpit instruments from approximately 3,000 ft all the way down to what was about to happen. And you will soon see why that was important. Now, there is also one final detail on this aircraft that we need to talk about. A system called an integrated approach navigation system or eons. This isn't something that's essential for instrument approaches. But it should have actually helped change this story's outcome if it had only been used properly. You see, an instrument approach relies on two forms of guidance. lateral guidance to keep the aircraft aligned laterally with reference to the runway center line and vertical guidance to maintain the corrected descent path that we call glide slope on an ILS or glide path if it's generated by our onboard systems. The gold standard for instrument approaches is of course an instrument landing system ILS which using groundbased transmitters provide both localizer guidance and glide slope guidance and can accurately guide the aircraft down towards the runway in some cases even down to an outland. Unfortunately though Chuke airport didn't have an ILS. Instead, it used an ARNAV GPS-based approach where the aircraft would follow a series of GPS way points with published altitude constraints. This procedure provides accurate lateral guidance, but unlike the ILS, it doesn't include a transmitted glide slope, and that's where EANS comes in. EONS uses the aircraft's flight management system, combining GPS with other onboard navigation sensors to determine the aircraft's precise position, and it then generates a path guidance that looks and behaves like an ILS, displaying two magenta deviation diamonds on the pilot's primary flight displays, with one for lateral guidance, like a localizer, and one for vertical guidance, like a glide slope. And because the enhanced ground proximity warning system is integrated with eons, it can also generate glide slope alerts during an arnav approach. But in this case, there was also a catch. You see, Papa 2 papa extra echo was the only boring 737 in air new fleet that was fitted with eons. And because of that, the system was never fully incorporated into the airlines manuals or standard operating procedures. So the pilots were never formally trained on it. Now in aviation we spend a lot of time thinking about redundancies and protections backups for the backups basically. And today's sponsor has given me a challenge to share how I protect myself in other situations outside of aviation. So here we go. 60 seconds on the clock. Go. Movie spoilers. Uh well on this channel I am normally the one who gives them but I will hush anyone who starts to discuss a movie that I am waiting to see. Neighborhood gossip. Uh, this is normally my wife's department, but a tip is never mention that you're a pilot. You will just end up with a lot of turbulence questions during the next neighborhood barbecue. Your partner stealing your leftovers never been a huge problem to be honest, but I do hide them if I really want them. You see, the same rule applies here as in the cockpit. The captain, my wife, and the first officer, me, never eats the same food. Online scams. Now, this one deserves a serious answer. I always use NordVPN's next generation antivirus. You see, unlike traditional antivirus software who wait until a threat is already on your device, Nord works to intercept it at the source instead. It identifies fake shops, blocks scam websites, and neutralizes fishing links before they can ever even reach you. Plus, it's built right into the NordVPN app, so the VPN is still there doing what it's always done. And now you can get an exclusive NordVPN deal plus four months extra at nordvpn.com/pilot or using this QR code. And with a 30-day money back guarantee, you've got nothing to lose by trying them out. Thank you Nord for sponsoring this video. Now, let's continue. In December of 2015, almost 3 years before this flight took off, Air New Guinea had issued a flight standing order, FSO, briefly explaining eons and instructing engineers to remove this system from this aircraft in order to standardize the fleet. Now, that might sound counterintuitive. Why remove a perfectly good superior system? But standardization is actually often more important from a safety perspective. You see, one aircraft with a vastly different system could introduce confusion and lead to important callouts from the system being disregarded even when they shouldn't. So, the fact that the company had decided to remove it was not very strange. But that order then expired 3 months later before its content had been incorporated into any permanent manuals, which meant that engineering never carried out this removal. No placard was added to the aircraft to indicate that EANS was in fact still installed and no written guidance on its use was ever added to the airlines training program. So that was the aircraft. As for the crew in command of this flight was a very experienced captain. He was 52 years old and a Papa Newu Guinea National holding an airline transport pilot license as well as a valid Boing 77 type rating. His total flying time stood at 19,781 hours accumulated over a flying career nearly two decades long. And out of those 2,276 hours had been flown on the Boeing 737. All of them in command. But despite that overall experience, there are a few things worth looking at in more detail when it comes to his background. The first is Air New Guinea's process for qualifying pilots to operate on specific routes and airports. The investigation would later find that the captain's aerodyrome qualification for routes like this had been conducted by observation only. That meant that the captain had likely sat on the jump seat and observed the route being flown but hadn't actually done any hands-on assessed check where he actually flew the procedures himself. This is one of the more subtle concerns that came up during the investigation because obviously watching something and actually doing it, especially in demanding environments like Chuke, are really not the same thing. The other concern with the captain was his rest. The official report recorded that the captain had gone off duty in Pompei at 11:00 p.m. on the previous night and that he had then signed back on at 7:50 a.m. in the following morning which gave him a duty free period of 8 hours and 50 minutes. Now, the Air Neon Guini flight administration manual in line with the Papa Newu Guinea civil aviation rules required a minimum duty-free period of 9 hours, including the hours between 1000 p.m. and 6:00 a.m., which means that in this case, the minimum had not been met. Now, and this is important. I am not suggesting that fatigue it its own was responsible for what was about to happen because during the investigation, a specialist aviation medical practitioner with more than 20 years of accident investigation experience assessed all of the relevant evidence and found no indication that fatigue contributed to this. But it is still worth highlighting because it provides another insight into the airlines attitude towards its own procedures. As for the first officer, he was 35 years old and came from Australia, also holding an airline transport pilot license and a Boeing 737 type rating. He had accumulated a total of 4,618 flying hours, of which 1,820 had been as pilot in command on other aircraft types, but he had only logged 368 hours on the Boing 77 as a first officer, which meant that even though he was reasonably experienced overall, he wasn't that experienced on the Boeing 737 that he was now about to fly. There is also another interesting detail about the first officer that the investigators noted during the approach. He was wearing non-prescription tinted sunglasses and the investigation noted this because the flight involved metological conditions like clouds and heavy rain with little or no external visual references. And in those conditions, tinted lenses can reduce the contrast and legibility of certain flight instruments. Again, on its own, this was a small detail, but in the context of what ultimately happened, it is worth mentioning. In any case, for this flight, the crew had agreed that the captain would be pilot flying, meaning that he would have the controls from departure on through to the landing, and the first officer was therefore going to be pilot monitoring, meaning that he would handle radio calls, but most importantly, it would also be his duty to catch anything that the other pilot might miss. Since the pilot flying tend to have higher workload, this meant calling out deviations, challenging unsafe decisions, and if absolutely necessary, taking control of the aircraft if the pilot flying would fail to respond. It is super important to understand that the pilot monitoring is not just a passive observer. In a cockpit with effective crew resource management or CRM, dividing work and having two independent set of eyes significantly reduces the chance that deviations or unsafe decisions get out of hand and it is actually a cornerstone of modern aviation safety. But in this context, the investigation later noted that Air Neonini's training program had not included practical training for the actions expected from the pilot monitoring. specifically during a sustained unstable approach below 1,000 ft. And that was despite the fact that Air Newogini's own standard operating procedures did require the pilot monitoring to challenge the pilot flying and even if necessary even taking control if an unsafe situation developed and was not corrected. So in other words, the right procedures to follow were written down, but the pilots were never properly taught how to use them. Now, the story I'm soon about to tell you should stand as a great example of why this is so important. But as it turns out, this was not the first time that this particular crew had faced issues in this area. If we back the tape just one day to the 27th of September, these two pilots have been flying the same aircraft on Air Neonini Flight 72. And what the investigation found provided a fascinating insight into how these two dealt with certain issues. During the approach on that flight, the aircraft had descended significantly below the standard 3° glide path. And this wasn't just a brief deviation, it actually remained consistently and substantially below the desired profile throughout much of the approach. >> >> Now, because this aircraft was equipped with EANS, the enhanced ground proximity warning system therefore generated glide slope glide slope alerts in response. And not just one of them. No, incredibly 28 separate glides slope audio alerts had sounded in the cockpit during that approach. And how have the crew responded? Well, according to the cockpit voice recorder, they simply just continued their ongoing non-standard conversation. They didn't discuss the approach or the warnings at all and they took no corrective action to return the aircraft up to the proper descent path and that was even though this warning was very much valid. Now in commercial aviation there's a principle known as the sterile cockpit rule. Now, some details can differ between airlines and different authorities, but in general, this rule states that below 10,000 ft and especially during the critical phases of approach and landing, all communication must be strictly limited to what is operationally necessary. This rule exists for a reason. Distractions during critical phases of flight have contributed to many, many accidents throughout aviation history. But on that previous day, even though this crew managed to land their aircraft without incident, they had done so whilst engaging in completely unrelated conversations as well as ignoring key warnings. And the problem here is that a bad habit that becomes repetitive without anyone challenging it eventually basically stops feeling like a bad habit. It simply becomes the norm in what we refer to as normalization of devian. The investigation report stated that the crew's previous experience of receiving similar audio alerts on approaches where the aircraft landed safely would have contributed to the perception that those alerts were just nuisance warnings and could therefore be disregarded. And the fact that these alerts only were suned on GPS approaches in this particular aircraft likely added to that belief. Now before I continue, I would like to also add a couple of thoughts here beyond what the investigators highlighted in this story. During unexpected heavy rainfall on short final, which was the exact conditions that this crew was about to encounter, the intense noise of rain striking the aircraft could possibly mask some of the warning sounds, at least in part. And at the same time, the tension from the sudden loss of external visual cues can degrade situational awareness in ways that are not always fully within a pilot's conscious control. And that's true even if they're not primed to ignore warnings as these two pilots seem to be. And then another point worth keeping in mind here is that there would be really no urgency for these pilots to land into Chuke on their first try. Back when they were planning the flight, they had uploaded so much fuel that when the aircraft would reach the missed approach point for Shook, it should still have about 10.4 tons of fuel remaining. And to put that into perspective, the fuel required for the next sector from Chuke to Port Moresby was approximately 6.2 tons. So this meant that the crew had a comfortable margin with sufficient fuel for several missed approaches at Chuke before diverting if necessary. Now, if you have watched some of my other videos, which I really hope you have and that you have subscribed to the channel. If you haven't, please do so now. You might remember a concept known as get there itis which is a tendency to press on towards a destination when the safer option would clearly be to go around and try again or to divert and we will soon have to revisit that concept. On the morning of the 28th of September 2018, just after 9:00 a.m. local time, flight 73 pushed back from the gate at Poppe International Airport. The weather there was good for departure and everything was entirely routine. The sector from Pompei to Shook is approximately 380 nautical miles long and for a Boing 77800 that's a relatively short flight. Actually barely enough for the aircraft to level off at cruise altitude before it would be time to start to descent again. So sure enough, the crew eventually climbed to flight level 400 because they were very light. They just stayed there for a couple of minutes and then began preparing for their arrival into Chuke. And this is where the investigators noted something about the way that this airline and this crew in particular were following their own procedures. Shortly before the top of the sand, the first officer used this electronic flight bag. basically a company provided iPad to access the Boeing operational performance tool or OPT which is a software that assists pilots with takeoff and landing performance calculations. Using it, the crews enters information such as the aircraft weight, runway conditions and weather and the software then calculates performance data such as approach speeds, landing distance requirement and the recommended flap setting and auto brake settings. Now this is genuinely a useful piece of software and Air New Guinea had purchased it from Boeing and then made it available to some of its pilots but technically at the time when this flight took place software was still intended for training purposes only. You see, the OPT had not yet been approved by the Papa Newu Guinea Civil Aviation Safety Authority for operational use, and it was also not listed as an authorized performance calculation tool in the Air New Guinea flight crew operational manual or in their standard operating procedures. So, as far as the airline and the regulators were concerned, officially the approved method for calculating landing performance was still paper charge and the performance tables carried on board the aircraft. But in this case, the first officer used it anyway, probably for convenience. And based on the aircraft landing weight and the runway at Chuke, the software calculated that using flaps 40 would reduce the required landing distance by approximately 150 m compared to using flaps 30. So the first officer passed on this information to the captain who agreed flap 40 would be better. Now, the investigation concluded that using the software did not contribute to the accident and that flaps 40 was in fact a sound choice. But the investigator's concern beyond the use of non-standard procedures was that the captain simply accepted the first officer's calculation without cross-checking it either by using the approved performance charge or even by running his own OPT. And if you are a regular viewer on this channel, you know that this is something that has gotten Cruz in a lot of trouble before. But to be perfectly honest here, I think that the investigators just use this OPT discussion to make a point around how crews were using nonapproved software. Anyway, with those landing performance calculations now out of the way, even if they weren't done properly, the crew then should have conducted a thorough approach briefing. This is a formal briefing that we do before starting any type of instrument approach and sometimes it can be relatively short but it should always cover the approach complete with altitudes, visual references, minimums, approach aids and importantly also the missed approach procedure on flight 73. That briefing was started normally but it then wasn't followed through. The missed approach was just mentioned with one word the MI which was the point that the crew should proceed towards and it contained nothing about when to turn what altitudes to follow etc. The whole briefing was just made in a really rushed and nonchalant manner and it would later turn out that the crew only had one plate to share between them which was placed in between them on the center of display. Something that was also against the stipulated procedures. Now, this is a problem because poorly executed briefings is often precursors to problems to come. If everything goes fine, it might not make much of a difference. But if anything happens that the crew then didn't count on, the pilots might not share a common view of what to do. And that's especially true when it comes to the missed approach procedure. Now, it is hard to say whether this crew were just used to not briefing properly or if it had to do with the fact that the captain was at this point becoming a little bit stressed about not having started the descent early enough. You see, at time 0854 and 36 seconds, San Francisco radio sent them a quite convoluted message. ATC clearance New Guinea 73 descent to reach flight of a 340 by time 2305. Sudo requirement to reach flight level 340 by time 2305 Sudo and report level flight level 340 cruise flight 340 to Chuke airport report arrival report leaving flight level 280 flight level 180 and flight level 080 time check 254 and a quarter. This indicated that they had clearance to descent on their own discretion following those requirements, but it contained so many different points that it would likely have been quite confusing. In any case, the captain soon realized that they were becoming a bit high on profile and therefore started descending whilst still discussing what reporting points that they had and how the rest of the descent should be planned. Up until this point, the discussions in the cockpit had been quite relaxed. But once the descent started, it was obvious that the workload in the cockpit had started to rise at time 0908 and 16 seconds. The first officer then contacted Chute Radio and asked for a weather update for the field and the radio operator soon answered with the following. Wind variable at five. Visibility 14. Scattered 012. Charlie Bravo broken 120. Overcast 280. Temperature 26. Dupoint. 25. Altimeter 2973. And in this case, Charlie Bravo stood for CBS or thunderstorm clouds. This report didn't seem to worry these pilots much. Maybe because CBS were so frequent in this area, but it would soon become an issue for them. And here is where something known as threat and error management would have come in handy. In my previous airline, this forms an integral part of the briefing. Basically, it means to discuss any potential threats that we might face and also how to deal with them. And that goes for basically any time that a new threat is identified as well. Again, this is to keep a shared picture of what's going on and to keep the workload as low as possible. In any case, the crew then kept descending down towards their destination. Then for their approach, they had selected the Arnav GPS approach into runway 04, which they had kind of briefed and which was the standard published procedure for any aircraft arriving from that direction. Now, we have already talked about what ARNAP means in general terms. And for this particular runway, the aircraft was supposed to follow a sequence of GPS way points, each with a published crossing altitude leading up to the final approach, which started from the initial approach fix. From that point, the crew were to track inbound on a course of 041° towards the runway, and they should then cross the final approach fix called Faspo at 1,700 ft. From there they needed to continue descending towards Hamax where the missed approach point was located with a minimum descent altitude of 420 ft. Now that was the MDA or the minimum descent altitude. And we can think of HMAX and the MDA as kind of a line in the sand or a decision point. When the crew reached it, one of two things should happen. Either the required visual references would be clearly in sight and the approach was stable enough to continue in which case they could continue to watch the landing or if the crew didn't fulfill any of those requirements well then they had to execute the missed approach. Now because this aircraft was equipped with eons as I mentioned the crew also had an extra ace up their sleeves. An electronic glide path would be displayed on their flight instruments showing whether they were above or below the ideal 3°ree glide path just like it would on a normal ILS. Now we define a stabilized approach as follows. By 1,000 ft above the aerodyome in instrument metrological conditions, meaning inside cloud, the aircraft must be fully configured for landing, established on the correct flight path, descending at an appropriate rate, which normally means no more than 1,000 ft per minute, unless it's specifically briefed, and flying at the correct approach speed. This can then be brought down normally to about 500 ft if the weather is good. But if any of those criteria cannot be met with understand operating procedures are clear, you have to go around from the very first hours of flight training. Pilots have a founding principle drilled into them. A good landing is not created in the last few seconds before touchdown. It is created through good circuit planning and energy management as well as a stabilized final approach. And in order to help with this, the poppy lights beside the runway also provide another independent reference for altitude management. Now, for our viewers who might be unfamiliar with this system, poppies are based out of four lights situated next to the runway by the touchdown zone. And when this system displays two whites and two red lights, well, then you are exactly on the three° path or whatever they are set for. Three whites and one red means that you're slightly high and vice versa. So this also means that with four red lights, you are significantly below the desired approach path and you also really have no way of knowing exactly how low you are. Now on a correctly flown approach, all of these cues work together. The pilots continuously monitor, cross-check, and make corrections as needed in order to maintain a safe and stabilized descent. And in this case, if this crew had actually briefed the published missed approach procedure, they would have seen that it required them to climb straight ahead to the missed approach point and then turn left to a heading of 3 or 6°, climb into 2,500 ft and track towards a waypoint called Dami. This was a very straightforward missed approach and more importantly, the investigation found that the weather along that missed approach track was also completely clear of weather. And had it been briefed properly, it would likely also have been easier to start, which is worth keeping in mind. But what actually happened then? Well, at time 0920 local, flight 73 established themselves on the final approach track. And the captain confirmed this by saying, "Okay, we're on Arnav 041 and I'll go to 1,00 which is a bit curious as 1,00 was not an altitude stipulated in procedures, but maybe this was just him referring to the stabilized criteria. Who knows?" In any case, a moment later, the first officer broadcasted their position on the common traffic frequency. Air New Guinea 73 established inbound ornab GPS runway 04. And remember, there was no tower here to guide them. Descending through 4,100 ft, the captain then called for gear down on flaps 15 by saying, "Uh, we just configure as we can," which was yet another incomplete sentence and another small detail that pass without comment. With Juke still about 15 nautical miles ahead of them, the latest weather report sounded manageable, indicating light and variable winds, 14 mi visibility, scattered cumulan nimbus at 12,200 ft with broken clouds above. But on the weather radar, the pilots could now see that a storm cell was sitting directly between them and the airport. The first officer was first to notice it, and he now pointed it out to the captain. "Well, that must be some storm," the captain replied. but I'll be out of it soon. And so the approach just continued. It is a bit hard to know what he meant by that. Maybe he thought that the storm would dissipate or most likely he thought that they would just be able to pass through it without any big difficulty. But here it is worth mentioning that we should never start an approach if we have to cross a red weather radar return to complete it. It is much better to just hold and let the storm pass by. And as I've mentioned before, this flight had more than enough fuel to do so. Anyway, the captain now called for flaps 30, then flaps 40, and also requested a landing checklist. But whilst the first officer did call out a number of the items on that checklist, including gear flaps and lights, he didn't really completed in full. Passing 1,000 ft, the enhanced ground proximity warning system automatically called out 1,00. And to that the first officer responded, "Okay, stable." Which was absolutely true since the autopilot was still engaged and both the tracking path was being tracked reasonably accurately. 4 seconds later, the captain replied, "Continue." To which the first officer replied, "And visual 900 cloud base." This meant that the runway was now in sight. So at just under 900 ft, the approach appeared entirely normal. The runway was clearly visible and looked to be only moments away. But at this moment, the aircraft started to drift slightly above the recommended path. And the first officer responded to this by calling out what he saw. One red three whites, meaning that the palpus was now showing them getting high. Selecting flaps 40 had initially produced a little bit more lift than flaps 30 had, causing something known as a ballooning effect, which had pushed the aircraft up just a little bit. But the only thing needed to fix that would be a small correction, maybe an increase of descent rate from 750 to 1,000 ft per minute for a short while. So now at 677 ft, the captain disconnected the autopilot and said, "I'm going back on profile as he gently increased the rate of descent by pushing his yoke forward." But then just two seconds later, the aircraft entered in under the storm cloud and was hit by an intense shower. As the rain hit, the captain called for windshield wipers to be activated, but that didn't help. The runway, which had been clearly visible only seconds earlier, now disappeared completely behind a solid wall of tropical rain. So outside the windshield, there were now no visual cues, no horizon, no runway, and no visual confirmation of how the aircraft was actually doing. Now, if this was you, what would your instinct tell you to do? Well, for this crew, the instinct was apparently to just keep going. After all, the runway had been just there, so it has to still be just ahead. If they could just keep going and get back down onto the glide path, they should break out on the other side soon, right? Well, less than 10 seconds after entering the rain, the captain showed that he had made his decision. He simply said, "Okay, landing." But the recorded data now started contrasting hugely from what he thought was happening. Almost immediately after the aircraft had entered into the heavy rain, the approach had started to become unstable. The rate of descent had increased well beyond the stabilized approach criteria and both the flight data recorder and the engineers cockpit video showed the captain now making increasingly aggressive pitch corrections in an attempt to regain what he thought was the desired flight path. As a side note here, there exists a phenomenon called VFR to IMC, meaning a flight that is performed under visual flight rules inadvertently entering into instrument methological conditions. And that is something that has caused a lot of accidents in smaller general aviation aircraft. But of course, this flight was conducted under IFR or instrument flight rules. Even though the portion of the flight we are looking at right now had been visual and what was about to soon happen clearly shows that even experienced commercial pilots are not immune to the effects of VFR into IMC. Just 3 seconds after the captain said landing the enhanced ground proximity warning system announced >> minimum >> minimums. So, the aircraft was now passing 420 ft, the published minimum descent altitude. And this should have been that line in the sand that I mentioned earlier. But instead of being stabilized and visual here, the aircraft was now descending with approximately 1,490 ft per minute, around 50% faster than the normal maximum permitted. And none of the pilots were actually seeing the runway. But still, neither of them called for a goaround. When the captain had disconnected the autopilot earlier, he had done that because the aircraft was slightly high on approach. And that image had now probably became fixed in his mind. But the condition was now changing very fast and he had no longer any reliable visual picture to understand what he was actually doing. So his instinct was just to lower the nose even more and to regain what he believed to be the correct profile. And the problem with that was that by this point the only reliable information left was coming from the flight instruments which are often quite jittery at low altitude and close to the runway which is one of the reasons that we have to be visual in order to descend below the MDA. But it is likely that this captain wasn't looking at those instruments at all. He was now focused outside anticipating that the runway should once again become visible at any second. >> >> In a realworld situation like this, breaking that mental model can be very difficult. The crew's attention had now narrowed almost entirely onto completing this landing despite the lack of any real urgency. And this is a well-known human factor phenomenon called attention tunneling. And together with get there itis, it can be a really deadly combo. Meanwhile, the first officer had also been using the poppy as his primary visual reference until the rain took it away. So the last indication that he had called to the captain, three white, one red, was the last reliable visual picture for him as well. At this point though, his role as pilot monitoring was more important than ever. With the captain's workload increasing dramatically and his decision-making clearly impaired, he should have been the one that should have continuously monitored the aircraft flight path, cross-checking the instruments and verbalizing any unsafe conditions. basically becoming the voice of reason that could break this dangerous situation. But those calls never came. Instead, the aircraft just continued descending below the minimum descent altitude and now also well below the electronic glide path with a vertical speed of nearly 1500 ft per minute. Now, this was obviously taking them straight down towards the waters of Chuke Lagoon. And the enhanced ground proximity warning system was about to start telling them exactly that. Now, what happened from passing the minimum descent altitude to the end of this flight took only about 22 seconds. And in that time, the enhanced ground proximity warning system would generate 13 separate oral alerts, seven glide slope, and six sync rate warnings. And the video from the flight engineers phone also showed that they got at least two visual pullup warnings on their primary flight displays. At time 0924 and 0 seconds as the aircraft passed 307 ft above the water. The EGPWS announced its first warning rate. >> Sync rate sink rate. >> Sink rate. >> And to that the captain responded by not to call for a goound but to justify what he was doing. Uh okay. I just want to get on profile. 4 seconds later, the GPWS added glide slope, glide slope, glide slope as the magenda vertical deviation diamond on the primary flight displays had now reached the very top of the scale. >> 2 seconds later, the aircraft again blared out sync rate synr. And again, all of this was happening without any visual cues. But instead of either pilot calling for a goaround, the captain now seemed to try and reassure the first officer by saying, "That's fine. I'll just go a little bit more." 3 seconds later, at time 0 924 and 9 seconds, another glide slope glide slope warning came and went. And 2 seconds after that, the first officer finally blurted out a question. See the runway? Now, this should have been an indication to the captain that the first officer likely didn't see the runway, but he didn't answer. Instead, 3 seconds later, the aircraft called out >> 100. That meant that the aircraft's radio altimeter now sensed that they were only 100 ft above the waters of Juke Lagoon. And almost at the same time, another glide slope warning came to which the captain just responded slow, >> monitor air speed. Okay, got it. And the glide slope warnings just kept sounding. >> Glide slope. >> Then just two seconds before the end of the flights, when the aircraft was approximately 30 ft above the water, the last sink warning was heard. And this was where the first officer fully recognized what was about to happen and now shouted, "Too low, we're too low. We're too low. We're too low." With the red pull-up warnings on the pilot's screens basically now saying the same thing. Any flight 73 now struck the surface of Chuke Lagoon at time 0924 and 19 seconds approximately 460 m short of the runway 04 threshold. But this wasn't quite the end. The 737 now skipped on the sea surface almost like a flat stone decelerating violently before turning clockwise through 210° and finally coming to a rest roughly 140 m southeast of the runway center line with its nose pointing back out the way that it had come. Inside the cabin, most passengers actually survived the initial impacts and were now able to stand up and see what was happening. At rows 17 and 22, the fuselage had ruptured around its lower circumference as it hit the water. And with the slight nose up attitude that the plane now had, the back of the cabin quickly begun to flood. What followed was chaotic. But it could have been much, much worse. You see, Chuke airport had no dedicated water rescue services despite being surrounded by water on three sides. But the passengers and crew of this flight now had an amazing stroke of luck. A group of US Navy divers, complete with boats and diving gear, happened to be on an assignment very close to where the plane hit and managed to see everything from the shore. They therefore quickly launched an inflatable boat and reached the aircraft within minutes. Local residents and fishermen familiar with these waters also moved immediately towards the wreck. And between the Navy divers and the local boat, a pretty substantial rescue operation now materialized out of nowhere. The locals soon pulled out 28 passengers and two cabin crew from the left overing exits. While on the other side, the Navy divers pulled out six passengers, four cabin crew, and a load master. The two pilots and the engineer then came out through the forward exit door. On top of this, the cabin crew had managed to deploy a life raft with the passenger's help from the left rear overwing exit. So all in all, 46 out of the 47 people on board made it out. Unfortunately though, the 47th person was a passenger seated in 23A, directly beside the point where the fuselage had fractured. Other passengers later told investigators that they had seen him stand up during the approach, so he was likely not wearing his seat belt at the time of impact. Three days later, local divers found him lying on his back on the cabin floor between rows 22 and 23, close to the fracture, only meters away from his seat. And the autopsy found no evidence of drowning. Instead, he had sustained severe traumatic head injuries on impact that according to the pathologists would have caused death within minutes. Six more passengers were seriously injured and none of them had any seat belt bruising, suggesting that they likely hadn't been using their seat belts either. On top of this, the evacuation was also found to have been hampered by several passengers and crew trying to take their hand luggage with them as they exited. As I said at the start, the investigators were soon able to cross-reference the engineers phone footage from the jump seat frame by frame with the flight data recorder data and they copied voice recorder data, allowing them to see exactly what the instruments were showing. And even better, the flight was so short that the CVR and FDR data from the previous flights were also available, revealing a pattern of ignored EGPWS warnings, a consistently below flight path approach, and a nonsterile cockpit. Both pilots were later interviewed, and the captain maintained that he had had visual contact with the runway edge lights when he said, "Continue," whilst the first officer stated that he believed that the approach had been stable throughout. Now clearly the data and the video painted a completely different picture. So how was this allowed to happen then? Well, the investigators found that the captain had experienced fixation, a state in which attention become so narrowly concentrated on a goal that contradictory information is simply filtered out rather than processed. So he likely wasn't ignoring the warnings, he simply tuned them out. Similarly, the first officer's failure to intervene was caused by an equally outdated mental model. The last poppy reading that he called out was one red, three white, and that picture stayed with him even as they crossed the minimum descent altitude with a descent rate of 1500 ft per minute. But there is just no doubt in my mind that a more disciplined cockpit crew that had spent time on their briefing and respected warnings and stabilized approach criterias would never have found themsself in this type of situation. Now, the official investigation produced a number of safety recommendations, including one directed at Honeywell, which manufactures the EGPWS, suggesting that an oral call of pull up might have broken through where the glide slope and sink rate alerts did not. But Honeywell and the United States National Transportation Safety Board did not agree. The NTSB made their reasoning on human factors grounds. The crew had ignored 16 EGPWS alerts in the 19 seconds before impact, and they didn't believe that yet another warning would have likely changed things. Now, Air New Guinea made several changes after this accident. Simulator training was updated to include sudden loss of visual reference scenarios and EGPWS response training was improved. All of which matters. But what I want to leave you with is one final thought, especially if you're a pilot or thinking of becoming one. The fact that the approach of this flight was actually filmed made this a very well understood accident. And yet with all of that understanding, the central lesson remains one of the oldest in aviation. You can always go around. Never convince yourself that an unstabilized approach can be saved or that a lost runway is just around the corner. Go around early and go around often. This and exactly how we fly the Boing 737s professionally is something that Ben and I will show you in detail if you check out our virtual 737 course. If you're a flight simmer out there or just curious about how we work, this course will revolutionize the way you see things. So scan this QR code or if you're interested, go to virtual737course.com and use the code mentor pilot for a 20% discount off the price for a limited time. Now before you go, I just want to give a quick shout out to our collaborators over at Veritasium who have just launched a brand new channel called Ingenious. Their latest video tells the incredible story of how one trucker's idea and a tiny piece of metal called the twist lock ended up turning China into an economic superpower. It is truly brilliant. So, click here and go check it out. My name is Peter Hornfeld and you're watching Mentor Pilot. Have an absolutely fantastic day and I'll see you next time. Bye-bye.