CSV/ANIMATION ALTITUDE DISCREPANCY

CSV/ANIMATION INITIAL ALTITUDE DISCREPANCY
Adam Larson / Caustic Logic
The Frustrating Fraud
August 16 2007
last updated 8/21

Note: Pilots for Truth senior member “Undertow,” who I admit has far more knowledge in this area than I, commented on the following piece “You can not possibly be this moronic about this. Please for the love of god stop confusing everyone.” (see comments below)

I read this as his seeing too many mistakes to be accidental, and so it’s probably an attempt to confuse, a feeling I'm familiar enough with. No, in fact, I can be, and was, this moronic. I am off on a lotta spots here, numbers and otherwise. Sorry, I was tired and inspired at the same time. The general gist is still right, but numerous minor alterations and further notes are to come… (in red) Apologies for any remaining confusion and please don’t take this as a final word on anything without consulting a non-obfuscationist expert first. In fact I'm not even saying much at all here.

But as “mod” says below “CL is going in the right direction IMHO. What does the Video say vs. the CSV file, in terms of pure numbers? 99% should match.” I also feel I’m onto something here or I wouldn’t’ve posted it… I just don’t know what yet. Maybe it’s nothing. But UT’s vehemence has inspired me: “Jesus, please stop and really think about this before posting anymore absolute nonsense.” Have done, and I got lots of red here now. Jesus. I'm done.

For more in-depth, piloty, and udeful analysis of pressure altitude readings, please see the lively and enlightened Pilots for Truth discussion thread at this link. Hopefully you're a member, or perhaps they'll provide me with a better link to help explain what I'm missing here. Otherwise do your own research or just hang tight to see what I come up with.
---
Where I was Wrong
Aha! I feel stupid but not really. I had previously concluded that the original, takeoff altitude for Flight 77 in the csv fille matched the same in the NTSB animation. It turns out I was incorrect. It was a mistake on my part but kind of odd how easily it lent itself to misreading.

1) The actual altitude at Dulles is about 300 ft msl as I’d heard. I verified this with Earthtools; runway 30 is on an slight uphill slope from about 285-295 feet MSL. I’ll call it 290.

2) The NTSB animation shows precisely 300 feet in the altitude dial on the ground. It’s not quite right, but close enough to be accurate. It also happens to match almost exactly the magnetic heading on that runway of 300 degrees from magnetic north, also displayed as 302 with the plane at rest.

The csv agrees on mag heading and lists it as 302 for the long runway pause, which is where I got mixed up. When I first dug in for altitudes, I saw it was luckily only the fifth of several hundred settings so would be easy enough to find. I had heard takeoff altitudes matched, so looked first for an altitude of about 300, I counted over five, six, whatever values and saw 302. Not realizing it was actually the mag heading, I took it as a verification of the animation’s rough accuracy. The reason I missed the actual pressure altitude is that it’s unusually different from the display alt at that point, while the mag heading is still much closer, if it were altitude in feet. And they’re listed right next to each other so I got mixed up.

In fact the beginning altitude listed in the csv is 40 feet, 250 feet too low to be accurate. This is a point of disagreement I haven’t heard anyone mention yet, though Undertow tells me I’m “covering ground which we plowed last summer.”

Tho it didn't come across as I noted a discrepancy, I do also understand the csv and animation attitudes should not exactly match – as UT usefully points out “The FDR produces a Pressure Alt and Baro column in the CSV. Together they create what the pilot would see on the Actual Alitmeter. Which is what is supposed to be shown in the Animation.” That is, the pressure alt reading “+” barometer correction = real altitude. The exact relation and expected differences I’m not totally clear on but getting there [another post]. What I'm really noting here is a massive difference I hadn't noticed before, apparently the mirror image of the 300-foot part of John Doe X's correction.

Tracking Back
I stumbled on this error while looking into another, related anomaly, a search that involved lots of backtracking. I saw that the final altitudes at 9:37:44 are near identical – 180 for the animation’s dial and 173 for the csv (of course they aren’t supposed to be, because of the omitted barometer reset). This final altitude is however misleading, and seems to be from a frame of data that didn’t play out and would have left the animation considerably lower. In the four seconds before that, the two reading carried a rapidly decreasing difference of 50 feet to 28 feet, with the animation lower. I didn’t bother tracking this discrepancy, but noted that they match up closer back around 9:24 (animation about 15-20 feet lower than the csv), both before and after the mysterious FL180 reset recorded in the csv that had no effect on the animation readout.

So clearly the altitude difference between the two is all across the board, but nothing in the difference range of hundreds of feet. Tracing back further, they also match roughly after the initial FL180 reset on ascent (at 8:28) with difference of 20-30 feet but with the animation higher this time. Since it reads lower near the end, there is a roughly 75-foot fudge room, at least, between the two, that comes in somewhere between 8:28 and 9:24. I don’t feel like tracking this down, especially since I had been looking for something else.

An Inexplicable 240-foot Gap
At the initial FL180 reset, on ascent, at 8:28, Pandora’s Black Box shows an altitude drop from 18273 to 18058, a difference of 215 feet. Despite having matched the csv before that (I thought) and after being only about 27 feet off. There is 242 feet of total discrepancy here, most of which is lost at 8:28 leaving a near-match after, which means the animation must’ve been set much higher than the csv before the change. Another error of the cartoon, no doubt.

I’d never noted it before, but lo and behold, the altitudes do not match before that onscreen drop. The difference is in fact about 280 feet, animation high. Since I “knew” they matched on the runway at 8:20, I traced it back over the intervening eight minutes to see where the discrepancy crept in. Actually I started at the beginning and made the same mistake of identifying 300 and then watched for the up-tick to match the animation after it started rising at 8:20:16. I suspected the csv would be slower, but it remained at about 300 all-down the 300-oriented runway second after second as the animated plane rose. Then I realized I was looking at the wrong column, that the discrepancy runs back to the first frames and the csv file’s original altitude is wrong – to the tune of 250 feet underground, the same discrepancy nearly erased with the onscreen altitude change (leaving as the remainder the smaller discrepancy).

Here is a table of data at the relevant points: "baro cor" is the setting re-set at 18,000 feet both ways, once apparently by Capt. Burlingame, once by capt. Hanjour. + differences mean animation high, - means csv higher than animation.
time |baro cor | anim alt | csv alt | diff.
8:20:15 | -- | 300 | 59 |+241
8:20:16 | 30.20 | 304 | 52 |+252
8:20:17 | -- | 311 | 49 | +262
8:20:18 | 30.21 | 323 | 53 | +270
8:20:19 | -- | 339 | 61 | +278
8:20:20 | 30.20 | 363 | 83 | +280
8:20:21 | -- | 392 | 111 | +281
8:20:22 | 30.21 | 420 | 144 | +276
8:20:23 | -- | 456 180 +276
8:20:24 30.20 494 210 +284
8:20:25 -- 530 248 +231
8:20:26 30.21 568 290 +278
8:20:27 -- 614 330 +284
8:20:28 30.20 662 372 +290
8:20:29 -- 704 417 +287
8:20:30 | 30.21 | 755 | 469 |+ 286
…
8:27:54 | 30.21 | 18081 | 17785 |+296
8:27:55 | -- | 18129 | 17823 |+306
8:27:56 29.91 18140 17861 +279
8:27:57 -- 18179 17899 +280
8:27:58 30.21 18219 17938 +280
8:27:59 -- 18252 17976 +276
8:28:00 29.91 18166 18015 +151
2:28:01 -- 18083 18056 +27
8:28:02 30.21 18118 18093 +25
8:28:04 29.91 18198 18170 +29
8:28:06 | 29.94 | 18278 | 18247 | +30
… …
9:24:10 | 29.92 | 18264 | 18285 | -21
9:24:11 --- 18222 18245 -23
9:24:12 29.91 18185 18205 -20
9:24:13 -- 18151 18168 -17
9:24:14 29.92 18106 18126 -20
9:24:15 -- 18071 18088 -17
9:24:16 30.23 18030 18049 -19
9:24:17 -- 17993 18011 -18
9:24:18 | 30.01 | 17956 | 17972 | -16
… ... ...
9:37:40 | 30.23 | 445 | 496 | -51
9:37:41 --- 351 399 -48
9:37:42 30.24 279 307 -28
9:37:43 --- 211 239 -28
9:37:44 | 30.23 | 180 | 173 | +7
---
Here is the graph of five seconds at each of the four key spots. One can see how a roughly 300-foot jump in the animation altitude (red) would be expected at the second re-set. I'm not sure what exactly to make of this, this is just to illustrate. (view in new window to see it full-size)

Clearly, there is a massive discrepancy only for the first eight minutes of flight and back on the runway. So this piece stands as two points:

1) I correct myself on a silly mistake – the initial csv altitude I had previously worked with was grossly wrong
2) Whether or not they should, the animation dispaly and csv pressure altitudes roughly match all through the flight, as I have previously noted, except for these first eight minutes.

And since I just now figured this out I don’t want to say what it means – if the JDX correction is boosted by this or if this reveals something else. That's in the math I'll do a bit later. But it cleary raises questions about the data itself, if not about what physically happened.