Home Page || Flight 587 Page || Your Comments @ U.S.Read's Web Log || Send E-Mail to the Editor || Search U.S.Read or the Web ||

U.S.Read's Flight 587 Preliminary Report

Part 4:  The Tollbooth Video And The FBI's "Ongoing Investigation"

August 18, 2004

See end of article for a February 14, 2005 update in re the "original" videos

by Brett Hoffstadt and Victor Trombettas

(We include some "Late Breaking Info" at the end of this article which came to us after the completion of this part of our report)

Two cameras mounted at the Marine Parkway Bridge tollbooth plaza in Brooklyn captured, in a series of still shots, the most critical portion of Flight 587's departure from JFK.  These pictures include the entire sequence of the pilot's control inputs, the loud bang, the loss of control, the "flash" and large smoke or mist trail, and the first 1,000 feet of descent.  Unfortunately, the quality of the video is very poor; it is monochrome and the image of the aircraft is little more than a small horizontal line.  Figure 1 shows how the aircraft appears on the video:

Figure 1

A low-quality, compressed Windows Media file of the relevant portion of the tollbooth video is freely available from the NTSB's web site at:

The NTSB's Video Study (a pdf file) is available at:, but includes no information concerning the analysis the NTSB performed to determine if any debris was seen falling from Flight 587 (FL587). 

The first half of the video shows the aircraft from the Lane 1 camera until it passes behind the building.  The video is then superimposed with the Lane 5 view of FL587.  The switchover from Lane 1 to 5 provides a seamless view of FL587 for approximately 19 seconds.  The video time (counter) stamped on the screen was approximately 49 seconds ahead of Air Traffic Control (ATC) time, the latter being the official (and accurate) time used in reference to FL587.

According to the NTSB, the tail separates during the Lane 1 view, when it is one second beyond the pole just to the left of the building. But they also acknowledge that nothing is seen falling in the Lane 1 view (see Figure 2).   If the tail departed here, it is very unusual that a flash and misting of the hydraulic fluid is not visible (as occurs later in the Lane 5 view).  There are three hydraulic lines that run to into the tail/rudder of the aircraft, each pressurized at up to 3,000 psi. 

Figure 2 - with the tail allegedly already departed

FL587's Altitude And The NTSB's Assumptions

The NTSB used a copy of the video to calculate the altitude of the aircraft after the Digital Data Recorder (DFDR) lost its data stream and the aircraft's transponder ceased sending altitude data to the radar facilities. 

Surprisingly, the NTSB's altitude calculations derived from the tollbooth video contradicted a plot of FL587's altitude in their Aircraft Performance Report.  The Video Study showed that at 9:16:06 a.m. (8 seconds after the time the tail allegedly separated) FL587 was at an altutude of 2,428 feet.  The NTSB's Aircraft Performance Report showed an altitude of 1,843 feet.  U.S.Read has informed the NTSB of this contradiction, and has been told they will be corrected to reflect the more accurate altitude from the Video Study. Figure 3 shows the altitude calculations the NTSB derived from the tollbooth video. Figure 4 shows the NTSB's altitude calculations that had not taken the video into consideration.

Figure 3 - from the NTSB's Video Study (the Time column shows UTC time - the 14th hour is 9 a.m. local time)

Figure 4 - from the NTSB's Aircraft Performance Report

The incorrect altitude plot (Figure 4) was likely based on an NTSB assumption that the aircraft started to lose altitude immediately after the time the tail and engines separated, 9:15:58.5 a.m. and 9:16:01.2 a.m., respectively.  But the altitude data derived from the video disproves this original assumption.

This error cannot be quickly brushed aside by the NTSB as a data entry error or as an insignificant miscalculation on their part.  Their incorrect altitude data was a logical assumption based on their conclusion of how and when the aircraft broke apart and descended, with the separation of the tail identified by the NTSB as the initiating event. 

Now that those initial altitude assumptions have been acknowledged as incorrect, the theory that gave rise to those numbers (that the tail had separated at the time of the loud bang) must be re-examined. 

Although of very poor quality, the tollbooth video shows that the aircraft lost only 70 feet of altitude up to 9:16:06.2, when the flash and smoke/mist trail appear on the video.  However, the NTSB still maintains that the vertical tail separated almost 8 seconds earlier at 9:15:58.5 (the time of the loud bang).  They have stated to U.S.Read that both engines departed the airplane at or around 9:16:01.20, five seconds before the flash and smoke/mist trail, and five seconds before the aircraft started descending. 

NTSB assumptions about when the engines separated

In regard to the "flash" and smoke/mist trail, Bob Benzon (the NTSB's Lead Investigator for FL587) said the following in his opening statement at the Hearings: "The staff believes that this could be misting fuel, smoke, or even flame that spread from the airplane after the engines broke away from the wing."

If this is so, why does the flash and smoke/mist trail not begin until five seconds after the engines separated?  Absent other clues, wouldn't it be more reasonable to assume that as soon as the engines were breaking off the flash and smoke/mist trail would appear on the video? 

If the engines departed later than 9:16:01, perhaps around the time of the flash on the video at 9:16:06, then the NTSB must answer the question: Why did the DFDR stop recording all flight data at 9:16:01.2?  There is no DFDR evidence that the engines separated at 9:16:01 –– as the DFDR was still powered (by the engines) for several seconds beyond the point where it had lost the entire data stream.  Neither the NTSB or any investigator that we interviewed, has provided anything remotely approaching an explanation for the DFDR data ending 13 seconds before impact.  We'll deal with this issue, the premature cessation of all DFDR data, later in our multi-part report. 

It appears the NTSB's conclusions of how and when FL587 broke apart have not been reconciled with the tollbooth video.

"That Can't Be"

Based on their conclusions about the video, the NTSB's position is that FL587 lost its tail and two engines and flew level (did not lose altitude) and did not produce a flash or smoke/mist trail for up to 8 seconds.  One FL587 Investigator who was not aware of these issues until we presented them to him responded, "that can't be"

But according to the NTSB, that is what happened.

One theory that can explain why FL587 did not lose altitude prior to 9:16:06 or produce a smoke/mist trail is that the tail and engines were still attached, contrary to what the NTSB has concluded.

Additionally, our conclusion that the aircraft's in-flight breakup (including the tail) occurred later than the NTSB claims is consistent with the video evidence and all the physical evidence –– including the eyewitnesses.

Items of Interest

There are a several items of interest from the copy of the tollbooth video we received from the NTSB.  The last version of the video we received from them was not compressed (as earlier ones were –– although this last tape was analog).  This provided a noticeable quality gain compared with the previous "mpeg" version we had received. This is important as it means the copy we examined does not have "compression artifacts", although there is still some overall quality loss due to the analog copy process. 

1.  NTSB has misidentified the start time of the video

The NTSB stated that the airplane was visible in the video from 9:15:54 a.m. and onward.  This is incorrect by at least 4 seconds –– a large error for a video clip that lasts only 19 seconds.  The aircraft appears in the video at ATC (Air Traffic Control) time 9:15:50  (based on the "pole 2" timing marker the NTSB identified).  This is a notable error because 9:15:51 is the moment when the pilot began his aggressive control inputs.  Unfortunately, many of these early frames of the video are heavily distorted (see Figure 5).

Figure 5 - Distorted portion of the tape and a portion the NTSB does not have - this is one of the few less-distorted frames from the first second or two of the tape

There should not be any distortion at that point in the video because the FBI confiscated over 6 hours of original video (on one tape) from the tollbooth facility both before and after FL587 flew by.  The distortion seems to be a result of the FBI creating a short version of the original tape for the NTSB.  The FBI should have been more meticulous and thorough in their preparation of the video for the NTSB.  Unfortunately, the NTSB didn't complain about receiving this shortened, distorted segment of the video.  They also didn't complain about receiving a copy from the FBI, instead of the original video. 

2.  NTSB missing slice of video

It is evident in viewing the video that several seconds of the flight before 9:15:50 should be available on the original video, yet the NTSB does not have this portion of the video (see Figure 5 above –– the white highlighted box shows a portion of FL587's flight –– at least a few seconds –– that is missing from the tape).

The NTSB either never noted these issues or was indifferent.  They never indicated (in two years of email exchanges with us on issues related to the video) that they were pursuing acquisition of the original video from the FBI.

3.  Something trailing behind the aircraft?

There is at least one portion of the video (in the Lane 1 view) that appears to show a trail behind the aircraft while it is ascending, during the time of the pilot's aggressive control inputs.  An accident video analyst who examined the tollbooth video for U.S.Read had this to say about this apparent trail:

"In this frame it does appear that there is something trailing behind the aircraft ... at approximately a negative 19-degree angle with respect to the average horizon line of the earth and the aircraft.  The aircraft appears to be traveling in a horizontal path parallel to the average horizon line of the earth."

Figure 6 - streak behind the aircraft?
(in Figure 6 we adjusted the color levels to attempt to better isolate the aircraft and the distortion trailing it –– we did not "enhance" the distortion whatsoever –– our adjustment affected the entire picture)

A split second later, this portion of video shows another interesting event.  The analyst stated:

"This frame appears to show the aircraft coming out from behind either smoke or debris" (see Figures 7 and 8).

Figure 7 - the aircraft obscured

Figure 8 - aircraft emerging from distortion

A video forensics specialist contacted by U.S.Read also noted that something is happening at this point on the tape (Figures 6 - 8).  But given that our examination was based on an analog copy of the NTSB's analog copy, he could not speculate if the event was an artifact due to the copy process or an actual event associated with the aircraft.  He also added that any conclusions (as to what is or isn't there) made from these copies of the videos are not reliable.  The original video must be examined before any conclusions can be reached. 

With this qualification in mind, the trail behind FL587 occurs approximately three seconds before the time the NTSB states the tail separated.  Thus, even if the NTSB's assumed time of tail separation is correct, the copy of the video suggests something may have occurred before that point.

U.S.Read is not declaring that we have found video evidence of a fire or explosion prior to tail separation.  However, we can say that the copy of the video we received from the NTSB suggests something may have been trailing the aircraft.  Therefore, the original video must be examined using the best tools in the world, before any conclusions can be reached.  Yet the NTSB has concluded that there is nothing notable on the tape –– and they have not examined the original.

Why would the NTSB be satisfied doing an intense examination on a second or third generation analog copy of the original video?  If they were intent on a serious investigation of the video, logic would dictate that they would settle for nothing less than the original.  Since they did not feel compelled to acquire the original video from the FBI, we are left wondering as to the seriousness of their video analysis. 

Didn't they notice the "trail" at this point in the video?  Were they even looking for these types of anomalies?  U.S.Read made the NTSB aware of this and other issues related to the video –– and the NTSB did not respond.

We are interested in this portion of the video, and in this possible "trail" emerging from FL587, because FL587 was already experiencing electrical anomalies with their radio communications and the Cockpit Voice Recorder (CVR).  In addition, witness John Power (among others), who observed FL587 during the time of the pilot's aggressive control inputs (before the tail separated), stated:

"I see smoke coming from his right wing ... and I see fire coming from his right wing."

4. When did the tail separate?

The NTSB stated at the 2002 Hearings that although they believe the tail separation occurred in the Lane 1 view of the video (at the time of the loud bang on the CVR at 9:15:58.5 a.m.), no such event was noted on the video.  Our examination of the video leads us to suspect that tail separation occurred during the Lane 5 view –– after the "flash" and smoke/mist trail at 9:16:06 a.m.  Supporting evidence for this conclusion comes from:

a.  the fact the aircraft begins losing altitude only after 9:16:06.
b.  the cockpit warning system (ECAM) which indicates major structural and systems failures occurred after 9:16:06.
c.  the radar data which shows more airborne debris after 9:16:06.
d.  many eyewitnesses –– who stated the tail departed at this point in the Bay.
e.  the fact there is no significant smoke/mist trail in the Lane 1 view of the video. 

5.  Debris visible in the lane 5 view?

Our analysis of the video indicates the Lane 5 view of the tollbooth video needs to be carefully scrutinized as it appears debris may be visible descending into the Bay from the large smoke/mist cloud released by FL587.  One item is of particular interest (see Figures 9 and 9.1):

Figure 9 - possible debris in the lane 5 view?  Seen at video time 9:16:56 frame 24.

Figure 9.1 - enlargement of Figure 9 with brightness/contrast adjustments

U.S.Read is continuing to analyze this section of the video.   If this is debris from FL587, it can only be the tail or rudder because if any debris would be visible on the video (given its poor quality) the tail/rudder would have to be it, given they were the largest piece(s) that fell into the Bay.   This would contradict the NTSB's timeline, and would support the conclusion that tail separation was not the cause of the crash.  But again we come back to the same issues –– didn't the NTSB notice these specks on the video?  Didn't they at least want to be certain about their conclusions and ask the FBI for the original video? 

Why the half-hearted effort?

The original video is still in the possession of the FBI.  The NTSB has never received it, much less asked for it.  U.S.Read filed a Freedom of Information Act (FOIA) request with the FBI in May 2003 for a digital-uncompressed copy of the original video, but our first request was denied with the reason given that its release might interfere with "enforcement proceedings".   We filed an appeal in November 2003 and are awaiting a reply. 

It is difficult to understand why the NTSB would accept anything less than the original of the only video documenting the most critical moments of the flight.  An expert analysis of the original video might yield important clues.  It is unprecedented to have any kind of images, much less a video, of an aircraft crash as it is unfolding –– this is why we think there should be an unprecedented effort to analyze it. 

If the NTSB deferred the most painstaking video analysis to the FBI, then why didn't the FBI produce a report?  Or why didn't the NTSB make available the details of the FBI's analysis?  The NTSB told U.S.Read that the FBI was "consulted" in the analysis of the video.  What does that mean?

In August 2003, almost two years after the crash, U.S.Read asked the NTSB:

"Do you know why the FBI is holding on to the original tollbooth video and why the NTSB never worked from the original?"

The NTSB reply was:

"[NTSB] can't speak for the FBI.  We felt the version we had was sufficient for our needs."

One NTSB Board Member, speaking to a FL587 family member, stated that the NTSB had their best people looking at the tollbooth video "pixel by pixel" for several months. 

We cannot understand why the NTSB would put such immense effort into examining a second or third generation VHS copy. 

A source within the investigation told U.S.Read that the NTSB did not form a video study group with the parties to the investigation, as they do with other areas of an investigation (the "Structures Group", for example).

In June of 2002, several months after the crash, the NTSB told the parties to the investigation that the video showed nothing of value, and they were shown a few still pictures from the video.  That was the extent of what the NTSB disclosed to the parties.  The NTSB did not even provide enlarged pictures of critical frames.   

Why won't the FBI release a digital uncompressed copy of the original tape to the public or to the families of the victims?  One family member told U.S.Read her lawyers failed in their effort to acquire the tape from the FBI.  Why? 

Why did the FBI not provide the NTSB with footage of flights preceding FL587 on the same flight path?  This would allow the NTSB to compare those normal flights to FL587 and be able to gauge what looks normal, and what does not.  Why didn't the NTSB see the usefulness of this?

Most importantly, why didn't the NTSB demand the original video tape in November 2001?

In addition to the tollbooth video, the radar data also does not reconcile with the NTSB's conclusions about when and where the tail separated, and where over the Bay the major breakup of the aircraft occurred.  We cover this in Part 5 of our report

Late Breaking Info

1.  U.S.Read was recently informed (after we had completed this part of our report) that diligent efforts by certain individuals close to the NTSB investigation had succeeded in getting the FBI to release the original tollbooth video to the NTSB (however, the NTSB has not confirmed this report).  U.S.Read raised this issue in 2002 and 2003 through many communications with the NTSB and the parties to the investigation. 

No party outside the FBI and NTSB has yet to see the original video.  U.S.Read has filed a new FOIA request with the NTSB for a digital-uncompressed copy of the original video.

2.  On August 16th we received a letter from the FBI denying our FOIA appeal for a digital-uncompressed copy of the video.  Their reason was that the video was:

"compiled for law enforcement purposes, the release of which could reasonably be expected to interfere with enforcement proceedings."

This was the same reason the FBI used in their original denial.

3.  On August 17th we spoke with a Department of Justice (DOJ) attorney who worked on our FOIA appeal.  We were informed:

"there is an ongoing FBI investigation that would be complicated by the release of this video –– we can't comment any further on the investigation or proceeding."

We informed the attorney (as we had in our appeal letter) that the NTSB was the lead agency, and that this was an "accident" investigation, not a criminal investigation.  He acknowledged that he was aware of the NTSB's position.  He repeated that the FBI has an "ongoing investigation".  We cannot reconcile these FBI statements with NTSB's conclusions about the crash. 

We reminded the attorney (as we had mentioned in our appeal letter) that "the NTSB had already released a lower quality version of the tape.  How could a higher quality release by the FBI do any harm?"  The reply was:

"the FBI acquired the original video from the tollbooth facility via subpoena –– the NTSB was not authorized to release the video and should not have done so".

We left a message with FBI Public Affairs in New York today, informing them that the DOJ's statements give the impression there is an open and active FBI criminal investigation in regards to FL587.  We asked for clarification and will update this space if we receive it. 

February 14, 2005 Update
After months of inquiry, U.S.Read finally received today from the NTSB a report offering some new details on the status of the original tollbooth videos (this report was on a docket CD dated 11/22/04).  This NTSB report (available here as a 2.6MB pdf file), dated October 7, 2004, reveals for the first time that the NTSB acquired the original videotapes in late May 2004 and that the originals do indeed provide a noticeably higher quality image, and more footage, of AA 587 in flight. 

The NTSB does not identify what new tests they conducted on the original tapes in an effort to identify when debris began departing the aircraft.  It is evident they did not convene a Study Group nor did they solicit the help of outside specialists to enhance and study the video images further. 

U.S.Read is still waiting on the NTSB to deliver to us (under a Freedom of Information Act request) a digital uncompressed copy of the original tapes. 

We welcome your comments and feedback on this article at U.S.Read's new web log at: or you can send us an email by clicking here.

Home Page || Flight 587 Page || Your Comments @ U.S.Read's Web Log || Send E-Mail to the Editor || Search U.S.Read or the Web ||