A Few Words of Caution For dMole

[BCR]

I was checking the referral links to AAL77.COM this morning and came across some interesting work by dMole at the P4T forum. He seems to be a sharp individual who is attempting to use some of the FAA TRACON data I have posted at the site to come up with a viable flight path. Unfortunately he is asserting inconsistencies in the data and of course people over there tend to scream "alteration". So I thought I would explain those inconsistencies and help the boys out.

The first is a time offset. NEADS time is running around 25.1 seconds slower than SEADS time. There is also a difference in one-two radar sweeps (4.7 seconds) between the TRACON's. I am awaiting radar data from the ZNY area so that I can do a full time correlation between the sites and broadcast time, but for the time being, it is enough to realize that there is a time offset between the site data sets.

Another aspect of the data that is critical to understand is that the positional data for the FAA TRACON's is calculated using the azimuth. The azimuth is given in whole degrees in the files provided by the FAA. This means that there is an inherent error in the positional values due to rounding which increases over time. Using a point at Columbia Pike and South Glebe as a reference the following errors must be considered.

DCA positional error = 0.05 nautical miles (translates to 40 knots in speed calculations)
ADW positional error = 0.19 nautical miles (147 knots)
IAD positional error = 0.33 nautical miles (254 knots)

So of course, there are going to be significant variations in speed estimates based on this data. As such, central trend lines should be examined, not one-to-one speed estimates.

When examined in this context, the speed trends correlate well with some predicted wild swings in speed estimates in the IAD data. The time offsets are visible in the graph, but what should be noted is the speed trends. For the purposes of dMole's assertions, all of the data is consistent with a sharp increase in speed which is increasing during the final seconds of flight and exceeding 450 knots at the terminus. The inconsistencies noted by P4T are due to measurement and rounding errors.

 

[ElMondoHummus]

Pardon me for going pedantic here, but for the benefit of younger students and folks who've not concentrated in science in school, I want to make a point about data analysis here: In addition to clarifying a point about the radar data and what it indicates about Flight 77's speed, 911files' post can serve as a real-life object lesson about significant figures. John brings up an important point about rounding errors, something often missing in much conspiracy peddling analysis. If a person does not take care with identifying what are and are not significant figures in their analyses, they can easily screw up their conclusions, especially if they attribute significance to certain measurements without regard to noise, rounding error (as John here has explicitly mentioned above), or other sorts of error.

The most recent example of this I can think of was the brew-ha-ha over Szamboti's and MacQueen's video analysis. Although for accuracy's sake, I have to say that that was much more about averaging obsfucation than anything else... but my point here is that there are quantifications that actually mean something, and quantifications that don't. Instead of worrying about point data, the important thing to remember is to observe the overall trend and see what that tells you. That's what John is excellently highlighting here.

Apologies for the derail, but I sometimes feel that meta points are often ignored when we all dive into the messy details of a topic. Ideally, threads contain something for all readers, including ones who're just starting to internalize concepts, and that's what I'm choosing to do. Anyway, I'll leave everyone else to discuss the TRACON data and John's analysis itself from here on out. Thanks for allowing me to indulge myself here.

 

[BCR]

Apologies for the derail, but I sometimes feel that meta points are often ignored when we all dive into the messy details of a topic. Ideally, threads contain something for all readers, including ones who're just starting to internalize concepts, and that's what I'm choosing to do. Anyway, I'll leave everyone else to discuss the TRACON data and John's analysis itself from here on out. Thanks for allowing me to indulge myself here.

You did not derail at all, because the topic is exactly what you addressed, rounding errors (significant figures) and how some folks can go entirely off-track if they don't allow for such in comparative analysis.

 

[beachnut]

Your work matches the FDR. If p4t or CIT were real investigators they could line up RADAR fixes with FDR information using groundspeed.

They are the dumbest investigators in the world. That is expected when selling DVD of implied lies.

dMole and the other p4t expert speed guys are not good at this. I get sick watching their failed work. Even when they get something right, they tie it to a flawed conclusion.

It is great; your speeds are matched to the Throttle push up at 37:16 on the FDR!

It is confirmation of the FDR, or RADAR data if CIT/p4t could understand evidence.