The Crash of Southern Airways flight 932
On the 14th of November in 1970, a McDonnell Douglas DC-9 crashed on final approach to the Huntington Tri-State airport in West Virginia, killing all 75 on board. The aircraft had been chartered as Southern Airways flight 932 to carry the Marshall University Football team, along with their coaches and fans (known as boosters) back home to Huntington after a football game in Greenville, North Carolina, which they lost.
Tri-State airport is surrounded by hills. The runway is 828 feet above mean sea level (amsl). There was a tree to the east of the outer marker at 990 feet amsl. Two rivers passed by the airport at elevations of about 500 feet amsl north, south and west of the airport.
The key point to take away here is that the geography around the runway was uneven, with the elevation ranging between 500 feet and 1,000 feet amsl. At the time, this wasn’t reflected on the charts; the pilots flying that evening probably didn’t know much about the surrounding area.
The weather was 300 feet scattered cloud, 500 feet variable broken ceiling, 1,100 feet overcast with a visibility 5 miles in light rain, fog and smoke. This means that at 600 feet, the flight crew would have struggled to see the ground. With the rain and fog, they might not be able to land at Huntington Tri-State at all, but everyone was hopeful that they would make it home.
In order to land in such conditions, a pilot has to use external navigational aids. In 1970, it was still very expensive for an airport to install Instrument Landing Systems (ILS) for their runways and non-precision approaches were much more common. A non-precision approach is simply an instrument approach to a runway which doesn’t have an instrument landing system. This means that the flight crew aren’t receiving a glide slope that will lead them directly to the runway like they would with an ILS approach. Instead, they use navigation beacons to align the aircraft with the runway and then they follow ‘step downs’ which allow them to descend safely towards the runway. The flight crew remains at a specific step-down altitude until they’ve passed the appropriate point, at which point they descend to the next step-down altitude.
The allows the flight crew to safely approach the airfield even though they are flying above or through clouds and cannot see the surrounding terrain, let alone the airfield.
The crew must remain aware of the minimum descent altitude, which they should not fly below unless they have the landing environment clearly in sight (that is, the runway and surrounding terrain). The approach can continue at that altitude until the ‘missed approach point’. If the flight crew still don’t have the runway in sight when they reach the missed approach point, they have to go around.
At the Tri-State Airport, the minimum descent altitude was 1,240 feet amsl, where the runway was 828 feet amsl. This means that flight crew can continue their approach at 400 feet above the runway threshold until they have the airport in sight, which will keep them clear of terrain. If the flight crew still can’t see the runway when they reach the missed approach point, they break off the approach and turn north while climbing to 3,100 feet. From there, they can either try again, if they think the visibility is improving, or they can divert to an alternative airport: one which has better weather or a runway with an instrument landing system, which will guide them to almost over the threshold even if visibility is bad.
In 1970, the Tri-State airport had a localiser which offered horizontal guidance to the flight crew. Once they are ‘on the localiser’, they know they are flying on the approach path towards the runway. If the airport had an instrument landing system (ILS), then it would also offer vertical guidance to the flight crew. The glide slope indicator in the aircraft would tell them exactly how to follow the path down to the runway to remain clear of terrain and reach the runway at the right height. Because Tri-State airport did not have an ILS, there was no vertical guidance.
There were other external navigational aids available to airports at the time, including Precision Approach Radar and VASI systems. Huntington didn’t have Precision Approach Radar but normally an airport of that size with no ILS would at least have a VASI system.
VASI is a visual approach slope indicator which uses red and white lights to offer visual descent guidance information. The light units are arranged in bars: near, middle and far; most VASI installations are two bars, near and far. The lights should be visible from up to 5 miles (8 km) during the day and up to 20 miles (32 km) at night.
These days, it’s much more common to see a Precision Approach Path Indicator (PAPI) but in the 1970s, the VASI system was much more common. Similar to PAPI, the lights appear as white or red depending on the angle from which it is viewed. If the aircraft is on the glide slope, then the first set of lights (near) should appear white and the second set (far) as red. This visual glide path allows the aircraft to remain safely away from obstructions within 10° of the extended runway centreline.
At the time, FAA policy was to provide VASI systems primarily where no other electronic guidance was planned or available. As the Tri-State airport was actively negotiating for a glide slope for an instrument landing system, no VASI system was installed. There was just one problem with this: they’d been negotiating for thirteen years, since 1957. In the mean time, they had no ILS and no VASI.
The airport lighting system, which consisted of high intensity approach lights, sequence flashers and high intensity runway lights, was operational and it’s been confirmed that on the 14th of November in 1970, everything was properly set.
However, another issue for the flight crew that night was that Tri-State was an off-route airport, which means that it was not an airport that the charter airline trained the crew on or tested them on. There was no standard way that the flight crew could demonstrate that they had sufficiently familiarised themselves with the airport approach. And there was nothing on the charts to warn them about the high terrain as they continued inbound to Tri-State in the rain in the dark.
The captain and the first officer chatted about the chart en route; they weren’t worried.
Captain: You might try it again
Sound of tuning of ADF
First Officer: Southern nine thirty two out of eleven thousand five hundred.
Captain: Approach plate’s two years old.
First Officer Yeah [unintelligible]. On these charter kits they don’t keep those things up like they’re supposed to.
Sound of laughter
Captain: How many miles you got to Pulaski?
First Officer: About to run out [unintelligible]. It’s pointing that way Frank. Can’t get a code on it though.
Captain: Let’s run the rest of the in-range check… How many miles you got on it? I [intelligible] it’s gone off.
First Officer: Yeah, it’s gone off. (Bugs) one two three
Captain: Put Charleston on yours [unintelligible]
Captain: Center, Southern nine thirty two
Indianapolis Center: Southern nine thirty two, descend and maintain five thousand, say again.
First Officer: Okay, Southern nine thirty-two, we’re out of eight now, we’re going to five, and approximately how far do you show us from the Huntington Airport?
Indianapolis Center: Nine thirty-two approximately twenty miles south-east of Huntington Airport.
First Officer: Roger.
Indianapolis Center: Southern nine thirty two squawk zero four zero zero, contact Huntington Approach Control one two zero point niner, radar service terminated.
First Officer: One two zero point nine, good day sir.
The fact that the charts didn’t show the terrain shouldn’t really matter. The procedure on the charts was still clear.
The flight crew that night were experienced: the captain had worked for Southern since 1949 and had 18,557 flying hours, the first officer had been hired by Southern five years ago in 1965 and had 5,872 flying hours. Both had rest periods (20 and 18 hours respectively) before reporting for duty.
So it isn’t surprising that they weren’t all that worried about the chart. The big question from the pilots’ point of view was whether they’d be able to get into the airfield at all.
The weather was 300 feet scattered, 500 feet broken, 1,100 feet overcast, with a visibility of five miles in the rain and fog. This means that at 600 feet, the flight crew would have struggled to see the ground. Once the crew descended to four hundred feet over the runway height, they could not descend further unless they were in clear sight of the runway.
As they approached the airfield. a previous aircraft reported that they broke out of the clouds right at four hundred feet and were able to land. It seemed hopeful that they would, as well, allowing Southern 932 to continue the approach and land.
Today, flight crew are told to keep a sterile cockpit, where they avoid any discussion and distractions that aren’t directly related the approach. But in 1970, this wasn’t a concept that anyone had heard of and it wasn’t really seen as a big deal that the flight crew were chatting during critical phases of flight.
Even so, on that cloudy night, it was particularly bad, because the charter coordinator was also in the cockpit, a break in procedure which wasn’t acceptable even then. Worse, he seemed not to care that the approach needed the pilots’ full concentration as he chatted to them, talking about picking up fuel in Huntington.
First Officer: Forty-two DME. How many you got?
Captain: Thirty-seven. Coming over middle marker.
First officer: Middle marker there.
Charter Coordinator: Frank, you want full fuel load out of here?
Captain: Might as well.
First Officer: Minimum is nineteen —- wonder how much they’ll charge us?
Charter Coordinator: Well, we get contract price, whatever that is, whatever we pay for it.
First Officer: We got a mile or two to go, Frank, ‘s all
Captain: We’re showing on the localizer
Charter Coordinator: Hope we don’t have this all the way in. It’s rough.
First Officer: There she is.
First Officer: Southern nine thirty two, we’re over the marker now, proceeding outbound.
Huntington Approach Control: Southern nine thirty-two, roger, report the marker inbound.
The lack of focus becomes obvious when the captain said, with surprise in his voice, that he had captured the glideslope. He was right to be surprised. Tri-State did not have an Instrument Landing System and there was no glideslope.
Captain: That thing captured! How did it capture?
First Officer: Yeah, it ought to.
Captain: You getting a glide slope capture and you ain’t got a glide slope.
First Officer: I might capture on the, ah, on ILS, ah, Frank, regardless of glide slope. I don’t have no capture, though.
Captain: Okay, give me twenty five
First Officer: Yeah, it’s good, it’s got the capture.
Captain: I got it cut off there now.
First Officer: Got twenty-five flaps, all is squared.
Captain: We ought to be over the outer marker at twenty — two hundred feet
First Officer: Yeah.
There is no ILS at the airport so they do not have a capture of anything.
Charter Coordinator: I’m sorry, Frank.
Captain: You going to call out minimums?
First Officer: Yeah, I sure will. I’ll sing ’em out to you.
There was no glide slope. The captain must have set up the instruments wrong; it seems likely that he was seeing the ‘automatic descent’ which is a standard descent of 700-800 feet per minute. This had nothing to do with the actual sector that he was in and it was in no way safe to follow the slope down.
Now, as long as the flight crew were still monitoring their altitude separately and following the non-precision step-downs, that wouldn’t matter. But instead, the captain became fixated on the (fake) glideslope and the autopilot. Both of the flight crew stopped paying attention to the step-downs.
The first officer continued to monitor the instruments as they descended into the dark, unaware of the rising terrain surrounding their destination.
First Officer: A thousand feet above the ground, rate and speed good.
First Officer: Speed a little fast, looks good. Got bug and twelve.
Captain: See something?
First Officer: No, not yet. It’s beginning to lighten up a little bit on the ground here at ah… seven hundred feet.
First Officer: Bug and five.
First Officer: We’re two hundred above.
Charter Coordinator: Bet it’ll be a missed approach.
By two hundred above, the first officer probably meant that they were 200 feet above the minimum descent altitude. The minimum descent altitude for Huntington was 1,240 feet above mean sea level or 400 feet above the runway height. The Charter Coordinator still hasn’t learned when to shut up.
Following procedure, they can descend another two hundred feet and then they continue to fly at 400 feet above the ground until they make visual contact with the runway.
A few seconds later, the first officer called out “Four hundred”, which presumably was meant to that they had reached the minimum descent altitude of four hundred feet above the ground.
Captain: That the approach?
First Officer: Yeah.
This sounds to me like the captain was asking if they reached the minimum descent altitude, which they had. The first officer confirmed and the aircraft began to level off.
Normally, the captain would have anticipated reaching the minimum descent altitude when the first officer called two hundred above, so that he would be levelled off by the time they reached it. Instead, he didn’t start leveling off until the first officer called four hundred feet. The aircraft sank an additional ninety feet.
Here’s the big question: what, exactly, did the call-outs mean? My first impression was that they make a lot more sense if the first officer was using the radio altimeter rather than the barometric altimeter. The barometric altimeter would have been set to show them their altitude above sea level. The radio altimeter times how long it takes radio waves to reflect off the ground, so it gives the “true” height because it is measuring the distance between the aircraft (or specifically, the antenna) and the ground directly below it.
The radio altimeter is only useful over a short range but it is extremely useful for maintaining an exact height above the terrain below, an issue which we looked at in the case of Air France flight 296 at Habsheim.
But on this foggy night, they were flying over rapidly changing terrain. In this scenario, it is much more useful to know your height over mean sea level, rather than know that you are 400 feet over whatever it was you just flew over, without knowing if it was a hill or a valley. For this approach, the barometric altimeter makes a lot more sense.
From the report:
Southern’s training program distinguished between the use of radio altimeters in instrument approaches over level and irregular terrain. However, the Southern Airways DC-9 Operating Manual did not make such a distinction, but rather accentuated its use for all instrument approaches by stating that, “Two separate radio altimeter systems…are provided to obtain precise altitude information above the ground at minimum decision [sic] altitude (MDA). This information is essential to the pilot in his decision to land or initiate a go-around maneuver.”
(Long term readers will know that I had a bad habit of talking about minimum decision altitude when I meant minimum descent altitude, so I’m kinda pleased to see that I’m not the only one…)
Notwithstanding the fact that the crew may have been formally trained to use the radio altimeter as a secondary reference, the tabulation comparing the available altitude references indicated that the first officer may have relied on the written material and was using the radio altimeter for altitude information.
It’s not possible to be sure what they were thinking. There is no clear evidence of any malfunction. The captain could have seen the dim glow of the approach lights through the weather, although the investigators think it was unlikely. Certainly the captain never said that he had the runway in sight. There was never a vocalised positive decision to continue the descent past the minimum descent altitude.
Nevertheless, by the time the aircraft was two miles out, they were 300 feet below the minimum descent altitude. The flight crew didn’t seem to be aware that they had descended past it. As the DC-9 came clear of the cloud, they discussed the lights just beginning to become visible in the distance, probably the glow from the refinery lights.
First Officer: Hundred and twenty six.
In a modern airline, the first officer is trained to be forceful, especially when he notices something wrong. Something like the fact that they are dangerously below the minimum descent altitude and no one has called out that the runway is in sight.
These days, we’d expect the first officer to make a direct instruction: WATCH HEIGHT! FULL POWER! GO AROUND! If the captain still didn’t respond, which can happen when someone is fixated, and the situation did not improve, then the first officer would be expected announce that he was taking control of the aircraft and to put full power on and climb away.
But that’s now, with a much greater understanding of Cockpit Resource Management and the benefit of decades of safety studies. In 1970, the first officer was clearly in a position of being subservient to the captain. He was meant to support him, not to question him. He didn’t have the power to insist on an action or take control of the aircraft. Again, this is similar to the A320 demo flight over Habsheim, in that the Pilot Monitoring, in that case a pilot who was a captain in his own right, was calling out the height but not empowered to take control when the aircraft went dangerously low.
The first officer shouted a single word: hundred. According to the flight data recorder, at this point the DC-9 stopped descending and someone applied power, almost certainly the captain, as not a word was spoken, leading the aircraft to enter a slow climb. Was the captain initiating a go-around? And if so, was the first officer’s shouting out “hundred” referring to their altitude or their airspeed?
Just over a mile west of the runway, there was a hill with a tree. The top of the tree was at 990 feet above mean sea level; 150 feet higher than the runway, which was at 828 feet above mean sea level.
With a minimum descent altitude of 1,240 feet amsl, they would have cleared the tree by 250 feet. If they’d had the runway in sight, they would have seen the silhouette of the tree, a clear obstacle blocking out the runway lights.
Instead, they flew right into it. The aircraft dipped to the right, almost inverted, and then crashed nose-first into a hollow. As it hit the ground, the aircraft burst into flames, an intense fire that reduced the fuselage to a powder-like substance. There was never an opportunity to evacuate and no chance of survival.
The increase in power and halting the descent imply that the captain had decided to go around, which fits in with the swath cut through the trees. But it was too late.
The Aircraft Accident Report AAR73-11 concluded only that the aircraft had descended into the tree.
The National Transportation Safety Board determines that the probable cause of this accident was the descent below Minimum Descent Altitude during a non-precision approach under adverse operating conditions, without visual contact with the runway environment. The Board has been unable to determine the reason for this descent although the two most likely explanations are (a) improper use of cockpit instrumentation data, or (b) an altimetry system error.
The two explanations are both interesting and both hard to believe. I look at them in more detail in Part 2: An Analysis.