Korean Air 2033

There are several sources on the Internet that claim the Korean first officer called for a go around prior to the landing and the Canadian captain refused. The normally reliable www.tailstrike.com has a "transcript of the Cockpit Voice Recorder" that includes the following line: "At this moment, the Co-pilot decides that there is insufficient distance for the Airbus to land safely without crossing the end of the runway. The Co-pilot, when the aircraft is only 30 feet off the ground, decides to "go around" and pulls back on the yoke." Wikipedia shows a "Cockpit voice recorder transcript" that says the copilot said "Go around" at 100 feet. I cannot find any primary source that backs any of this up.

When I first read these reports I was suspicious because the scenario goes against previous Korean Air crashes and the culture at the airline. At the time, the authority gradient between captains and first officers was quite steep. First officers in many cases sat humbly, knowing they were about to die. The reported CVR transcripts ran contrary to the airline's history. Parts of the report are available but they are written in Korean. I've had the report translated and it includes only CVR excerpts. What is printed doesn't show that the first officer called for a go around against the captain's wishes, only that he started to pull against the yoke while the captain was executing the flare, the captain said "get off" and the copilot then said "go around.""

• Date: 10 AUG 1994
• Time: 11:22 LT
• Type: Airbus A300B4-622R
• Operator: Korean Air
• Registration: HL7296
• Crew: Fatalities: 0 / Occupants: 8
• Passengers: Fatalities: 0 / Occupants: 152
• Total: Fatalities: 0 / Occupants: 160
• Airplane fate: Written off (damaged beyond repair)
• Location: Cheju Airport (CJU) (South Korea)
• Phase: Landing (LDG)
• Nature: Domestic Scheduled Passenger
• Departure airport: Seoul-Gimpo (Kimpo) International Airport (SEL/RKSS), South Korea
• Destination airport: Jeju (Cheju) International Airport (CJU/RKPC), South Korea

1 — The crew

2 — What we know happened

3 — My Analysis

4 — The lessons

Korean Air has through the years employed foreign nationals as captains, at first just because they needed the expertise and later because other nations threatened to refuse their airspace until the airlines of Korea improved their safety record. The U.S., for example, did just that in 2001. More about that: The Airlines of Korea.

The captain of Flight 2033 was a 53 year-old Canadian national who didn't speak Korean. It appears he had 13,207 total flight hours, of which 2,758 were in type. The first officer was a 36 year-old Korean who was said to have poor English skills. It appears he had 3,195 total flight hours, of which 316 were in type.

Piecing together what happened from the Accident Report, requires "reading between the lines," which I suppose is where much of the speculation comes from. What follows is what is in the translated report and where I need to speculate, I'll do so explicitly.

Why did the captain say "get off" at that moment, just prior to or during the flare? The captain was an English-speaking Canadian and I believe what might have been happening was the captain had his hands full with the gusty winds and believed the first officer's hands were on the yoke and directed him to "get off" the controls.

My first thought when reading reports of Korean Air 2033 was that it didn't make sense. In the eighteen years prior to this accident, there were 9 crashes of airlines of Korea: Korean Air Lines 642, Korean Air Lines 902, Korean Air Lines 015, Korean Air Lines 007, Korean Air Lines 084, Korean Air 803, Korean Air 221, Korean Air 852, and Asiana Airlines 733. Of the crashes directly attributable to the pilots, one crew failed to turn after takeoff and flew into a mountain, another crew ducked under the ILS glide slope, another captain taxied to the wrong runway and attempt to takeoff, another crew attempted to land using an ILS they knew was out of service, another crew failed to deice before takeoff, and another crew failed to use their before landing checklist and landed with the gear retracted.

In the six years that followed Korean Air 2033 there were three more crashes, each directly attributable to the Korean national captain's incompetence: Korean Air 801, Korean Air 6316, and Korean Air 8509. Two common themes emerge from many of these accidents: pilot incompetence and what looks like a squashing of evidence to suppress what really happened.

Korean Air 2033 follows these themes, at least in part. There is very little information publicly available about the accident other than an untranslated and incomplete accident report and news reports that appear to have little actual evidence behind them. I do have a few other sources but many of these appear to have their own agenda because of subsequent efforts in Korea to send both pilots to prison. (Korea has done this before, sending the pilots of Korean Air 8052, for example, to between 8 and 10 months in prison.) In the case of Korean Air 2033, both pilots were arrested but prosecutors ultimately declined to pursue convictions.

As for the pilots, that is a bit of an exception from all the rest. Of each of the accidents I've cited here, Korean Air 2033 was the only one that involved an expatriate pilot, that is, a pilot working in a country that he or she is not a citizen. Both Asiana and Korean Air were downgraded internationally in 2001 for their abysmal safety records, and hired many expats to help "right the ship." But that was long after Korean Air 2033.

If you watch some of the videos and read some of the translated transcripts, you will be told the first officer said "go around" and then the captain said "get off," leading you to believe the first officer commanded a go around which the captain ignored. But if you examine the untranslated Korean accident report, you see that the captain said "windshield wiper," which the first officer replied with, "windshear?" Then, the captain said "get off," to which the first officer said "go around?" We don't know the inflection of the voices to know if that last bit, "go around," was a question or a command. The English translation I have says that the first officer interpreted "get off" as "go around."

I was sent a copy of a Technical Report by J.M. Bélanger, a Canadian Safety Investigator who attended the reading of the Flight Data Recorder (FDR) and Cockpit Voice Recorder (CVR), with the authorization of the Republic of Korea Civil Aviation Bureau.

I cite only this passage as it clearly shows there was interference with the captain's control inputs during the flare. While Bélanger's qualifications as a professional engineer, air safety investigator, and pilot for Air Canada are not in dispute, the writing of his report shows a clear bias in favor of the Canadian Captain. He ignores or excuses the following:

1. The captain flew an unstable approach.
2. The captain's phraseology throughout the landing was nonstandard, making the first officer's poor English understanding even worse.
3. When it became obvious they would not have touched down in the first half of the runway, the captain should have aborted the landing and gone around.

It is important to examine the captain's actions with the lens of what we knew and how we flew airplanes back in 1994. The idea of a stable approach gate at 500 feet was in its infancy and I doubt many professional airline pilots back then would have thought twice about stable approach parameters in this situation. Korean Air as an airline had poorly written and enforced Standard Operating Procedures leading to the nonstandard callouts by both pilots. The idea of "continuation bias" may have been known by many in aviation at the time, but it wasn't an emphasis item in most aircrew training.

We pilots are result driven; we are there to get the job done. The captain's role is to safely fly the airplane from Point A to Point B and once in that mode, it is easy to ignore warning signs that get in the way. Windshear warnings? Heard them before, we can fly through most windshear alerts. A little high and fast on final? Easily dealt with. Landing going to be a bit long? The brakes can handle it. When you are in the heat of battle, it is easy to have a bias toward continuing the job at hand. This is why we need strict stable approach rules and very strong SOPs. They did not have either at Korean Air in 1994 and for many years to follow.

The probable cause section of the report:

In my view, mixing a highly experienced captain with a first officer with very little experience can only work if the captain is a patient instructor and the first officer is confident and willing to speak up. But a prerequisite to success is that both pilots are proficient in the same language. The report's recommendations provide a good list of lessons, but could go further:

Standard Operating Procedures in the form of proper checklist usage, call outs, and stable approach criteria empower the pilot monitoring to speak up and depersonalize corrections so that captains can realize they aren't being personally challenged by the younger pilot, only called on to adhere to a higher standard.

Landing performance needs to be considered prior to every landing, to include the impact of landing flap selection in view of the winds and runway contamination. Knowing you will need three-quarters of the runway should be adequate motivation to not salvage a landing halfway down the runway.

Rejecting a landing after touchdown and thrust reverser deployment is rarely a good idea.

Pilots must have a common language. A successful strategy employed by some airlines flying culturally mixed crews is to have a definite list of vocabulary and phrases that covers all normal and anticipated non-normal situations.

항공기 사고조사보고, 대한항공 A300-600, HL7296, November 1994, Ministry of Transportation (Republic of Korea)

Technical Report, Korean Air 2033, August 10, 1994, Accident Investigation in Cooperation with Bureau of Enquites Et Accidents Paris, August 24, 1994 by J.M. Bélanger, P.Eng.