Originally posted by Alfie1
reply to post by tezzajw
I don't think you have a clue how far out of your league you are with 911 files.
I should stop digging your hole if I was you mate.
Indeed. When I said that a very few here may already have
run the new numbers that Warren has uncovered, 911files was
the person I had in mind.
Perhaps I can give some small idea of what is involved. If
you simplify the problem enormously by considering only the
position, velocity, and acceleration data taken from the
flight data recorder, you have what is called an initial
value problem. In this case, the "initial value" is really
the set of final values for the three position coordinates
and the three components of the velocity. Given those six
final values, you can use the recorded accelerations to run
the flight backward, following a generalization of the
process I attempted to explain earlier within this thread
with respect to the altitude alone.
You can estimate the final values from the FDR data, while
checking your estimates by comparing your computed data
against data recorded by the FDR and other sources. Your
computed values will include some error due to errors in
your final values, the use of single values to represent
intervals ranging from 1/8 to 4 seconds (depending on the
measurement), the unknown/uncertain placement of those
values within each interval, the error in each recorded
instrument, and the amplification of those errors as the
computation goes further back in time.
It is quite unreasonable to ask for an "exact flight path".
On the other hand, it is reasonable to ask for flight paths
that are consistent with the FDR and other data to within
the known margins of error.
Using the extremely crude techniques I illustrated earlier
in this thread, I was already getting altitudes within 20%
of the radio heights except for one point, where the plane
most likely flew over a building or other obstacle. That
corresponds to a point made by 911files: radio heights are
not altitudes. One of the many remaining steps I should
take is to look up the elevation and topography at each
position, use that data to calculate altitudes from the
radio heights, and compare the altitudes calculated that
way with altitudes calculated from the estimated final
values and the known accelerations. That process would
lead to refinements of my estimates and (in all likelihood)
to better agreement with the known data.
That is just one very small example of what 911files means
by "time correlation of the data sets". There are many
other data sets besides the FDR and geographical data, and
911files knows those data sets way better than most of us.
I'm looking forward to his book.
If I have anything to contribute here at all, it is to
perform a crude calculation independently of 911files,
and perhaps to explain a little of why the full story
is too complicated for 911files to squeeze into a single
post.
Will
_), 
