Jet Crash Forum

Although this forum will be weather oriented, and rightly so, there are some new facts relevant to Tim's analysis.

The Brazilian Air Force has reported finding two bodies and other items "470 miles - about 900 kilometers from Fernando de Norohna" and "69 kilometers from the final communicated location." The miles must be nautical, and this would place the recovery location WNW of the last (0214Z) ACARS message location at about 3° 49' N and 30° 57' W. This location is consistent with the 0214Z location being close to the crash location, since one of the nearby drifter buoys (31919) has moved about 60km in a westerly direction since the time of the accident.

See:
http://www.estadao.com.br/noticias/cida ... 3405,0.htm

-rer47

. . .  and another buoy to the east (31526) has drifted about 80 km to the NE, making 3° 30' N and 29° 45' N a second possible recovery location. Either way, these points are only 69 km from the 0214Z aircraft location.

If variables such as position, speed, direction, etc along with weather conditions for this flight are known, can they be inserted into a simulator to estimate in which direction the fragments of the plane would have impacted the water? The same concept can be applied to the water current strength and direction to aid in locating the plane fragments. If the rate of descend of these fragments into the sea can be calculated and sea currents and direction can be known, the path of these fragments can be plotted using a simulator. I've heard of an array of listening devices for military or seismic measurement purposes. if an object traveling at 500 or 600 miles per hour hit the waters, it should have created a "sound event" that can be plotted to determine the impact location. This may sound radical, but can anyone tell Airbus to ground all flights until the causes of this catastrophe are known? Many people are telling me they don't want to flight on these fly-by-wire Airbuses....

The last ACARS message was at location: 3°34'40"N, 30°22'28"W (3.5777°N, 30.3744° W or 3.5777, -30.3744)
Search region is indeed northwest of this location - map here:

http://avherald.com/h?article=41a81ef1/0004&opt=0

More details on the search and recovery here:

https://www.defesa.gov.br/mostra_materi ... ERIA=33101

I have done some work in computer modelling.

I suppose one could create such a model, but it isn't clear to me that this would be very useful for a couple of reasons.  The most serious of these is that a computer model is always a conveniently simplified version of reality.  Especially when looking at weather systems, falls from 34000 feet, etc, it isn't clear to me that the accuracy from the model would be good enough to beat old-fashioned searches.

On a smaller level, though, I am sure that rough calcultions (either computerized or by hand) are being used to determine the search area.  I would not expect these to be the result of sophisticated computer models so much as required search areas based on past experience.

billb - the revised 6-6-09 PowerPoint presentation is a good find.

There are several discrepancies to note here, though. First, there are two kinds of data plotted in the PowerPoint. One is actual lat/lon data plotted in Google Earth (GE) and the other is simply graphics and text drawn in PowerPoint on the slides. The first kind has some geographic credibility, the second perhaps little. Some of the plotting of points in GE seems to have been done correctly, because reporting point TASIL is in the right location. To see the underlying GE plot, download the ppt presentation and delete the graphics that lie on top of the GE layer. You can then reproduce this in GE, positioning points using details of the sea floor bathymetry to eyeball locations.

First, the distance from Fernando de Noronha to the "AFR 447 ultimo reporte" location (the little red airplane symbol) is given on the ppt as 849 km, but when you measure it in GE it is only 814 km. However, if you plot the original ACARS 0214Z location given in AV Herald (N3.5777 W30.3744) and then measure the distance from Fernando de Noronha to that point, you get 851 km - very close to the 849 km given in the ppt. So, the AF447 Ultimo reporte "red airplane" is plotted in the wrong location in the GE layer - about 60 km too far south.

Second, the location given for the Recovery Site is actually plotted in GE (3° 34.08" N, 30° 27.30' W) so it is probably a real GPS coordinate obtained by FAB. If you plot this location it is only 9.8 km WSW of the original ACARS 0214Z last report - extremely close (see attached).

The graphics (and news reports) report a 69 km distance between the final ACARS 0214Z location and the Recovery Site, but plotting the data shows it is only 9.8 km. To do much more with this kind of data we will need to wait until the original source data is released. For example, there is no real confirmation yet of the 0214Z ACARS location except in the press. Perhaps there are coordinates for each of the 24 ACARS events - only the investigators know at this point. The short distance between the 0214Z ACARS location and the Recovery Site suggests the aircraft crashed shortly after the last message. Perhaps the "AF447 Ultimo reporte" and "ACARS 0214Z" points are both correct - the first at 0210Z and the second at 0214Z, with the 60 km distance between them being traversed at about 500 kts before impact.

It is useful to remember that the wing section of Pan Am 103 impacted the ground less than one minute after the explosion destroyed the aircraft at FL310.

http://www.aaib.gov.uk/cms_resources/df ... 503158.pdf

-rer47

-evidently you must login to see the attachment

FAB has published photographs of the SAR effort for 6 July. Among these is a photo of one of the SAR field maps - complete with water droplets that have smeared the ink, and evidently the signatures of SAR team members. It is interesting to see one of the maps used in the search:

http://www.fab.mil.br/portal/voo447/FOT ... sinada.jpg

I have warped the northern portion so it is flat and referenced to latitude/longitude. On this I have plotted in white the aviation waypoints and the location of St. Peter and Paul Rocks. As with my earlier post about the FAB PowerPoint maps, the actual location of TASIL matches that point on the SAR map fairly closely, but other points do not. In yellow I have again plotted the 0214Z ACARS location (N3.5777 W30.3744) and the 6 June Recovery location where two bodies evidently were found (3° 34.08'N, 30° 27.30'W). Also plotted in yellow is the "AF447 ultima reporte" location taken from the 6 June FAB PowerPoint presentation.

Interestingly, the SAR map shows yet another location for the last position of AF447 - the red airplane symbol about 27 km WSW of the "AF447 ultima reporte" point. The remainder of the text on the SAR map is difficult to read, but most of the red marks are labeled "Debris." Possibly the red marks inside the northern shaded box are radar targets reported by the R-99 aircraft. The ACARS 0214Z and Recovery locations lie inside a reported Debris area.

-rer47

-evidently you must login to see the attachment

@ref47

Nice work.  One point to bear in mind when you are working with different charting/mapping projections, is that just "warping" the images doesn't get the correct results.

The aircraft flies a Great Circle route, but in this case the the Way-Points on this GC track are plotted on a Mercator projection.  The course and distance from one WP to the next is shown in degrees Magnetic (deviation is about 17.8 degrees West at this location), and the distance between WPs is in Nautical Miles.

A Mercator Chart has the lines of Longitude drawn at a standard distance apart (determined by its scale) and the lines of Latitude are essentially the same distance apart at the Equator, but increase in distance by the Secant of the Latitude.  1 Nautical Mile = 1 minute of Latitude, so also varies though not in the same way.  A standard NM = 6076 feet (1852 metres), but at the equator is 6043 feet (from memory). 

Anyway, warping different projections to try and get positions to match is even more complex than that, as Satellite derived images require quite complex computer generated models to either place a Lat/Long grid on them, or warp them to some standard projection.

The final point I wish to make is that the "ultimo" 0214z position used by the FAB is that which has been calculated based on the last known position at INTOL 0133z and the ETA TASIL 0220z (not 0223z as has been reported elsewhere).

I suggest you read the substance of an email to Tim Vasquez headed
Sun, 07 Jun 2009 20:01:18 +0000     

http://weathergraphics.com/tim/af447/comments.shtml

and take note of the 0214z ACARS and 0214z Ultimo positions, they are different but strangely similar.

mm43

 

Some additional maps are here:

https://www.defesa.gov.br/UserFiles/Fil ... 6%2009.pdf

I have also not been able to confirm the coordinates given in the press for the last ACARS message.  Based on the above map, the coordinates used for the last reported location appear to be roughly: (2°57'45.71"N,  30°32'14.96"W) which is on the flight plan.

The search area does not exactly triangulate based on numbers given but is roughly: (3°35'51.00"N,  30°34'22.17"W).

This is promising in that the locations we are looking at are not an enormous spread.  The US Navy is sending a ship with sonar to search for the flight data and voice recorder pings.  I worked on the design of this sonar system and know it is extremely sensitive so there is a very good chance of locating the recorders.

mm32 & billb - Thanks for the comments.

I didn't want to put a lot of technical comment in the earlier posts, but your point about just warping maps is quite true. In this case, however, I georeferenced the northern portion of the SAR map by digitizing the lat/lon grid on the map. It was difficult where the map had been folded. In all there are about 55 control points, and I used Delaunay triangulation to pull the map back into plan view. Triangulation generally goes crazy outside of the control point array, so I just cropped that off. The resulting geotiff is pretty good (it's in UTM zone 25 S, WGS84) - at least good enough for this pixelated image. Once georeferenced, of course, it is easy to plot all kinds of lat/lon points on it. I think this map, as pixelated as it is, is the best reference we have for the search area - because it was actually in use by the SAR teams. The center of the "red airplane" symbol - presumably what FAB thought was the last ACARS message position - is located at: 3° 57' N, 30° 35' W.

Most of the rest of the FAB maps are not really maps but sketches drawn on a Google Earth base map. The one billb mentions is an example. If you get the PowerPoint version of this and just delete all the drawings you are left with an empty ocean. On this map, for example, TASIL is mislocated about 150 km to the NE (you can check this on Google Earth) and it is impossible to plot the distances given in the legend and come up with a single point.

There are exceptions. On the 06-06-09 FAB maps they actually plotted the "Ultimo reporte" location in Google Earth - so they must have had coordinates. By plotting the same airplane symbol and moving it around on the sea floor background you can eventually get it to match the FAB map exactly. So, the coordinates FAB used to produce "Ultimo reporte" on the GE map were: 3° N, 30° 21'W. And the apparent location of the first bodies recovered was plotted in GE at 3° 34.08'N, 30° 27.30'W, with the actual coordinates printed on the map (06-06-06 FAB PowerPoint).

FAB reported that AF447 checked in at INTOL at 0133Z, and when it left the coverage area of Brazilian radar (at 0148Z) it was at FL350 traveling at 453 kts. Air France told FAB that the 0214Z ACARS message was sent when the aircraft was about 100 km from TASIL. See:

https://www.defesa.gov.br/mostra_materi ... ERIA=33047

Thus between 0133Z and 0214Z at 453 kts, the aircraft would have traveled 310 nm, placing it at about 3° 11.5" N 30° 25.1"W - or about 23 km NE of the yellow "Ultimo reporte" point on the SAR map and 102 km from TASIL (a close match to what Air France told FAB).

According to Aviation Herald, FAB confirmed that the 0214Z location was N3.5777 W30.3744, but, as mentioned, I have never seen it in any official report. Please post if you find the source. The wording of the AV Herald report implies there is a lat/lon sent with each of the 24 ACARS messages, but I am not aware of any others that have been released/leaked. Why FAB has not used the N3.5777 W30.3744 location on their SAR map is unknown - perhaps because they know it is incorrect. But I think it must not be a coincidence that this ACARS 0214Z position and the Recovery location are only 10 km apart. I think we will need to wait for the initial official accident report to sort this out.

-rer47

I've been meaning to try Google Earth. The math about the miles is very helpful.  Everything about flying and sailing is just a tiny bit different than what we know on ground.

Was it poor weather that kept them from going NW of the different last known positions? I know I read something early in the week. If that's the case they couldn't help going other places first. I only had the New York Times interactive map and they seemed to have reasons to go every where but north. Even to the point of searching the entire flight path back to the island. And as result found ocean trash in their great need to find anything remotely debris-like. Not only are they working with a different "last known" they seem to have differing currents in mind vs. what was observed late in the week.

Admittedly, surface currents could be different than below. And they are knowledgeable of that sort of thing. They could be ahead of the game looking for the endpoint of things going on below. But if they all chose different starting positions this debris is going to slide on down to Davy Jones' domain with no one the wiser.

I'm beginning to think there's a book's material in the search efforts itself. And they're military entities. So has that something to do with how they thought through this search effort? I'm sort of thinking there was some target fixation on south and east.

After trying Google Earth with the points in the posts here and on the meteorological comments post mentioned above I do understand more why the SAR groups were using the southern point. But it's linear thinking. It's a starting point for the SAR search. I feel the true location isn't going to end up being on the flight line. Intuitively, from a storm's force, I know the line of flight is going to waver. The "ACARS" location is not out of the ballpark crazy(if it remains as a media invention and not being transmitted with the ACARS breakdowns); I now understand it's merely different math results. And possibly backed up by early debris finds before they got stuck in the shipping trash diversion.

@rer47

Thanks for your comments, and apologies for inadvertently try to teach you how to "suck eggs".

I have just come back to check on this threads latest postings, and some may find the following useful in understanding why the positions noted appear to be at odds with each other.

http://countjustonce.com/af447/

mm43

Thanks mm43, I was not aware of Carlson's analysis. I think the flight path provided by BEA now supercedes these kinds of analyses, as I discuss below.

By 11 June the Bureau d'Enquêtes et d'Analyses (BEA) published a reconstructed flight path for AF447. We must assume that BEA used the best data available to them for this reconstruction, including much that has not been released to the public. In my view this flight path supercedes the results of all prior analyses.

http://www.bea.aero/anglaise/actualite/ ... t.path.jpg

The final point is labeled "1st message" (1er message) and indicates the location of the aircraft at 0210Z when the initial ACARS maintenance message was sent. It is noteworthy that the points are labeled to the second (02:10:00), suggesting that this is a GPS position transmitted by the ACARS system separately from the widely-discussed ACARS maintenance reports. The flight path points suggest these position reports were sent every 10 minutes. When you look closely at the flight path it looks as though it is actual data, with closely spaced points (slower speeds) near Rio as the aircraft climbed to cruise altitude, and then some faster speeds south of NATAL.

On 17 June BEA published a press release that contains useful information:
http://www.bea.aero/anglaise/actualite/ ... 90617.html

From this we learn that:
- AF447 was in cruse at FL350
- The last ACARS position message was at 0210Z
- 24 ACARS maintenance messages were sent between 0210Z and 0214Z

Mention that the "last position message from the airplane was broadcast by the ACARS automatic system at 02h 10 UTC" lends support to the idea that ACARS position reports were sent separately (perhaps at 10 minute intervals) from the ACARS maintenance reports, since otherwise there would have been a position for the 0211Z to 0214Z maintenance messages and the 0210Z position would not have been the last one.

This press release also points to a pdf of a PowerPoint presentation:
http://www.bea.aero/anglaise/actualite/ ... ch.ops.pdf

This presentation contains several maps of the debris field locations between 6 and 10 June. In these maps the 0210Z location is labeled "last known position" (Dernière position connue).

Since BEA plotted these maps on a Google Earth base map it is possible to obtain approximate coordinates for the times that BEA plotted, and I have done this for the period 0120Z to 0210Z. The results are interesting. The BEA 0210Z location for the first ACARS message closely matches the location of the FAB "red airplane" symbol on the SAR map of 6-11-09 that I posted above. Thus, the Brazilian SAR teams were using the same location as the BEA. The approximate location of the 0210Z BEA "last known position" is 2° 58.7'N, 30° 35.8'W

Knowing position and time it is possible to reconstruct distance covered and ground speed for the final minutes of the flight:

0120Z
     77.0 nm, 462 kt
0130Z
     78.9 nm, 473 kt
0140Z
     78.0 nm, 468 kt
0150Z
     77.4 nm, 464 kt
0200Z
     77.2 nm, 463 kt
0210Z

Variation in the speeds and distances could well be due to small errors in reproducing the BEA flight path positions in Google Earth. The average speed between 0120Z and 0140Z is 468 kt, a close match to Barry Carlson's estimate of 467 kt referenced in mm43's post above, which was derived from the flight plan speed of Mach 0.82. The calculated speed between 0200Z and 0210Z of 463 kt is a reasonable match for the speed reported by FAB of 453 kt (presumably calculated from radar) at 0148Z when AF447 left Brazilian radar coverage:

http://www.fab.mil.br/portal/capa/index.php?mostra=3085

The 0214Z location of the last ACARS message reported by Aviation Herald (N3.5777 W30.3744) and attributed to FAB does not seem to have been confirmed elsewhere. If we continue with the analysis above we get:

0200Z
     77.2 nm, 463 kt
0210Z
     38.2 nm, 573 kt      
"0214Z"

The 573 kt speed is impossibly fast - it is about mach 0.98 at the altitude (FL350) and OAT (about -46C) of the flight. (If we assume a 10 kt headwind as calculated by Tim Vasquez, the resulting airspeed of 583 kt approaches mach 1) I think we must now assume that the "0214Z" position is incorrect and was not an ACARS position. The proximity of the "0214Z" position to the location of the first bodies discovered must be a result of the northward drift of the debris field as shown in the 12 june BEA PowerPoint maps.

By 0210Z AF447 had begun to depart from its planned flight path. At that time the cross track error was about 3 nm to the left of course, a turn of about 2.4 degrees (from the 0200Z position).

There has been much discussion of the role that overspeed resulting from erroneous airspeeds from faulty pitot tubes might have played in the crash. If the analysis presented here is correct, there was no meaningful increase in speed between 0120Z and 0210Z when the first of the ACARS maintenance messages was sent. If overspeed was a factor in the crash it must have occurred between 0210Z and 0214Z. The BEA 0210Z location falls well inside the massive storm system illustrated by Tim Vasquez, when AF447 was abeam the large, red "strong" radar reflector in his Figure 12. The BEA location at 0200Z, when the pilots evidently reported strong turbulence, falls just inside the southern limit of the storm system - shortly after AF447 entered it.

-rer47

While there is some astute analysis provided here on the crash location, my thoughts turn to snippets of what at least 2 commercial flights reported seeing the night of the crash.

Two pilots of an Air Comet flight from Lima to Lisbon saw a bright flash of light in the area where Flight 447 went down, the Madrid-based airline told CNN. The pilots have turned in their report to authorities.
"Suddenly, we saw in the distance a strong and intense flash of white light, which followed a descending and vertical trajectory and which broke up in six seconds," the captain wrote in the report.

"There is information that the pilot of a TAM aircraft saw several orange points on the ocean while flying over the region," he said.
But Brazil's air force said in a Tuesday statement that a French merchant ship — the Douce France — had conducted a search of the area where TAM reported seeing fire "without identifying any trace of the flight."

While the above information from these flights seems vague, does anybody know if the TAM (or other sightings - plane or ship) reported coordinates are similar to coordinates stated in this thread? The 'orange points' would seem as the closest to a 'smoking gun' as to the location, if the coordinates reported are indeed accurate (and what was seen wasn't some form of ITCZ phenomenon).

bolivia

So many things don't add up with the pitot tubes theory, mainly the chronological order of events, assuming that the ACARS messages are sequentially and chronologically correct. So far, there is no proof that they are otherwise.

#1. According to this thread, the pilots took a 2.4 degree heading change northward, at 0200z according to the location of the ACARS messages, which is roughly about the same time the aircraft hits a zone of stormy weather with strong turbulence. At this time the pilot sends a manual signal indicating that he is flying through an area of "CBs", but gives no warning to course deviation. At 0210z an ACARS message is sent warning the auto pilot automatically, not manually, has shut-off, followed by stabilizer, speed sensing, rudder limiter, navigational conflicts with flight computer, and flight management computer warnings. At 0214z, the last ACARS message is received warning loss of cabin pressure at a rate of 1,800 ft/min. The pilots were out of radio contact at 0130z, and out of radar contact at 0148z. Theoretically if they were planning on deviating from course, they would have planned it well-ahead of time and given notice to ground control of this flight path deviation. As far as I've been told, all pilots would report such a change given anything goes wrong, someone out there would at least have a clue of where to look for the downed plane. So why the unscheduled course change? Would bad weather be enough of a factor for the non-declared deviation and why didn't the pilot report the change along with his last known message of CB's at 0200z?

#2. If the pitot tubes did ice over they would have done so gradually, so that the plane would theoretically have increased or decreased in speed over a relatively long stretch of flight so that the auto pilot would have automatically shut-off well in advance before the 4 minutes alloted in the ACARS messages. The pilots would then have a reasonable amount of time to correct the problem or at least send a distress call alerting the other 11 flights on the same route to the predicament. From the location of the ACARS messages, the only speed change comes after the 0210z "systems deteriorating" messages. Could the plane accelerate fast enough during the four minute interval of messages at 0210z and 0214z when it looks like mid-air break-up was imminent? Also, its been pointed out that the plane would have to be relatively in a horizontal position for theses messages to transmit. So the mid-air break up would have to happen after the 0214z ACARS loss of cabin pressure message. Doesn't make much sense since by this time the plane is already descending at 1,800 ft/min. Unless mid-air break-up happened at a lower altitude, somewhere between 35,000 and sea level, meaning break-up was do to non-stability of the aircraft in a steep descent versus break-up do to speed increase along flight path.

#3. Deductive reasoning would allow for the theory that this accident unfolded instantaneously, without warning, and not over a large period of time as the pitot tubes theory would suggest. If the calculations in this thread are correct, the plane was not flying any faster or slower between known ACARS transmissions than normal, except for the last four minutes which would agree with the last ACARS message sent at 0214z that cabin pressure was dropping at a rate of 1,800 ft/sec. Meaning the plane was in a speed increased horizontal descent at last transmission. So within the last four minutes, the plane was in a moderately horizontal descent.

#4. So I ask other pilots out there, what could cause the auto pilot to automatically shut-off, and then speed sensors warnings to activate, then stabilizer warnings, rudder travel limiter warnings, navi-computer discrepancies, followed by loss of cabin pressure? If the plane did break-up in mid-air as it is now believed, isn't it more likely it did so on descent? If so, why was the plane in descent? What could have knocked this A330 out of the sky? If the pitot tubes did freeze-over causing systems failure at 0210z, wouldn't other alarms go off first? How could this A330 accelerate over 100 knots needed for mid-air break-up in a matter of minutes if not in a descent?

a). flying too fast or too slow. (the ACARS messages prove otherwise.)
b). faulty plane, weak structure. (there is no evidence so far to prove this on AF 447 or any other A#330)
c). pilot error in response to a systems failure. (could be. but it would have to be some serious pilot error to bring down a jetliner in less than 4 minutes. normally, pilots would struggle to correct errors or to correct effects from mother nature on the aircraft, and would have a reasonable amount of time to do so, unless the aircraft was non-navigational or incapacitated so that adjustments could not be made.)
d). something hit the plane causing it to become uncontrollable. (at 35,000 ft, what could do this to a plane? mother nature in form of strong winds, lightning, a freak hail storm, an unlikely meteorite, space debris, or maybe something not so uncommon, such as blue ice could hit the rear stabilizer, enough to render the stabilizer useless and hence a non-navigational plane.)

I still can't see how the pitot tubes can have sole blame for bringing down this aircraft in less than 4 minutes. Even if all three tubes were frozen over, a plane wouldn't just disintegrate without very skilled pilots knowing something was wrong ahead of time. I still think there had to be another variable, other than faulty pitot tubes and a big storm at night time. An unforeseen variable to which the ACARS messages, so far hold the only thread of proof.