First post here, after years of lurking. Mainly because I can't take it anymore. I'll start off with a rant on data analysis:
<rant>
Why is everyone plotting the Asiana flight data using time on the X axis? For example, there's
post 1210 - the NYT sure set a poor example. Then there's
this guy on Twitter, whose plots are also ragged and misleading because the FlightAware data he used had time truncated to 1 minute resolution.
Not pilots, presumably, because when was the last time any of us said "hmm, I'm a little high for being 2 minutes from touchdown"?? If you haven't seen it yet,
this article by @FlyingProfs is one of the best data analysis pieces I've seen so far.
</rant>
Ok, so the real reason for my post is to share the result of some data wrangling of my own. @FlyingProfs showed some interesting comparisons of the speed/altitude/energy of the accident flight vs. another arrival shortly before. Their article only compares against UAL 852 which landed on the same runway about 10 minutes prior. I got curious and plotted out AAR 214 and UAL 852 arrivals from the previous several days, as well as an ANA 8 arrival just a few minutes before the incident airplane. AAR and UAL were 777-200's; ANA 8 was a 777-300. I think the results reinforce the idea that the accident flight was "out of family" compared with the other approaches.
In the charts below, the accident flight is the redline. Other AAR flights in blue, except Jul 4 in orange; UAL and ANA in green. Vertical bars are "gates" representing the San Mateo Bridge and runway threshold. The grey horizontal in the speed chart is the target Vref = 137 kts for the accident flight.
On
the Quiet Bridge Visual procedure, recommended altitude at the bridge (5 mile final) is 1900', which just about everybody hits - except AAR 214 on Jul 4 and 6. The difference between the July 4 and 6 flights is in speed - with the July 6 flight being significantly over-speed at 3-6 miles out relative to all the other flights. (Since the plots are groundspeed, some variation can be expected given differing wind conditions encountered by the different flights) It seems but for some better speed control, the Jul 4 flight could potentially have met a similar fate.
Bottom line, the arrival profile of the accident flight inside 5 miles stands out (even aside from the fact it's highlighted in red).
This got me wondering just how unusual the accident flight profile was. The following altitude and speed scatter plots show data from the accident flight vs. 103 other B-777 flights landing at the SFO runway 28 complex over the last two weeks. I removed go-arounds as they tended to gum up the works in my quick and dirty plot setup. Note again that the speed plot shows groundspeed, so some variability would be expected for differing wind conditions over the two week period. Along similar lines, there's no guarantee that all plotted arrivals were on a visual approach vs. RNAV, LOC, etc. The glideslope shown is 3 degrees to a touchdown point 1000' beyond the threshold.
Again, it's pretty clear that Asiana 214's flight profile on July 6 was pretty unusual relative to the bulk of other flights.
Many others have commented on how this flight profile might have arisen. I have no experience or qualifications to address that aspect, so I'll stop here and hope everyone enjoys contemplating the charts
(Note: All plots were sourced from FlightAware data)
Edited to fix glideslope in 3rd chart.