New MH370 breakthrough tracking technology passes another validation test.
I have used an MH370 search aircraft from the Royal New Zealand Air Force on March 28th, 2014 to help validate this new technology called GDTAAA (Global Detection and Tracking of Aircraft Anywhere Anytime).
This new system is based on the Weak Signal Propagation Reports (WSPR) pronounced ‘whisper’ and promises to give a new search for MH370 a more precise location of the Boeing 777. This new analysis aligns with the satellite and drift modelling that points to an end point of MH370 at around 34.5°S near the 7th Arc.
This validation report is one of many planned to build a compelling body of work before the final analysis of the MH370 impact zone.
The report can be downloaded here
@Richard,
RE: GDTAAA AMSA VH-XNF Sortie4 14MAY2021 0500 UTC
Position Indicator Defined by Two Orange Drift Anomalous Links.
Quote: 30.935°S 103.890°E AMSA Search Aircraft
Quote: Avg. GS 349.8 knots … FL 220.0
Is this my possibly imaginary, presumptuous, Hercules on the way to drop more life rafts, but at the wrong time ?
I have just checked your 9-flight Sortie List (AMSA SAR D2S2 Rescue Aircraft and Sorties).
A97-442, RSCU211, flew Sortie #5 from Richmond to Pearce on 14thMay2021 04:08:12 through 10:43:48 UTC, a repositioning flight,
and subsequently flew a return Pearce to Richmond flight on Sortie #9 leaving at 05:28:28 UTC (13:28 local Perth Time) on the Saturday afternoon following the rescue.
(I still presume A97-442 did make the flight to take-over from VH-XNF, RSCU550, but did not make it into the nine sortie listing.)
______________________________
Your rule, or expectation, 7: I guess this may be something that gets revised based on experience as time goes on.
@Richard,
Like your 9.
As time goes on the size of the “kind of error bar” may become more valuable.
@George G,
I have the ADS-B data for the AMSA SAR RAAF Hercules C-130J A97-442 RSCU211 and included the flight to Pearce AFB arriving on 14th May 2021 at 10:43:48 UTC as Sortie5 in my list:
https://www.dropbox.com/s/uqjtqn6tji5oix9/GDTAAA%20AMSA%20RAAF%20A97-442%20RSCU211%2014MAY2021%20104348%20UTC.png?dl=0
I have the ADS-B data for the AMSA SAR RAAF Hercules C-130J A97-442 RSCU211 return flight and included the flight Sortie 9 from Pearce AFB departing on 15th May 2021 at 05:28:28 UTC.
It is perfectly possible that in between 14th May 2021 10:43:48 UTC and 15th May 2021 05:28:28 UTC a third Hercules sortie was flown, but this flight is not recorded by the ADS-B data for some strange reason, although all other flights were recorded. The Hercules RSCU211 Sortie 5 definitely did not overlap with the AMSA Challenger 604 jet VH-XNF RSCU550 in Sortie 4. RSCU550 arrived back in Perth at 09:47:00 UTC almost an hour before RSCU211 arrived at Pearce AFB coming directly from Richmond AFB. A third 6 hour sortie to the SIO by the Hercules would have required extra crew due to rest rules.
As far as the GDTAAA rules 1 to 9 are concerned, this is a draft version and subject to change as further validation tests are performed.
@Richard.
As I am seeing it, you two main proponents of long range aircraft detection by HF WSPR are reporting progress towards your hoped-for findings, with the expectation that you will be able to demonstrate on arriving at these, their validity and that of your methodology.
Reconciliation of those findings with the doubts of many with expertise and experience in HF propagation phenomena is a separate issue and does not of itself disqualify your potential findings in advance.
Assuming that to be fair, with that as context I make some observations, ask some questions and make some recommendations below ;
• In your rules 7a & b (to George G., 071034 June) you have turning, climbing or descending flights exchanging progress indicators for position indicators at 2 min markers, such that the latter lifts from 20% of those to 50%, that is from a bearing to two crossing, a fix.
As a stray, can you explain how climbing and descending could result in multiple bearings of anomalies? In those the vortex inducement from lift will decrease slightly if speed, sensibly, be maintained, and while there would be an increase with increasing thrust in a climb, why would there not be the opposite in descent?
As to whether that this progress-to-position effect would arise from pitch attitude change, even so, skip angle would disfavour either descent or climb.
As to heading influence, the effect of pitch on the two divergent anomalies would depend on their relative bearings, that ahead being more affected by pitch than abeam.
In that case, there should be some correlation of that with Doppler, drift.
But explicable or not there could be some clues as to ascent or descent or, possibly, neither if abeam if some pointers could be extracted from the above.
• Of more consequence, to reduce concerns that the indicators are just ‘consistent with’ or just correlate with WSPR anomalies, which could have other explanations, to me there needs to be a ‘blind’ demonstration of tracking from a suitable aircraft from a known start point, heading and time. The demo would need to cover anomaly density at both FMT as well as the SIO. In it, the actual track would be available but unknown to the tracker (ie you!). Most probably then it would need to be that of a military aircraft without ADS-B and preferably a P8 for similarity to the 777.
Or perhaps you have in mind a quite different demonstration that would satisfy the validity demands of science.
One could see your NW Sumatra track as unalerted but it does not have the same certainty, one possibility being track transfer to another aircraft, if they had been less than 2 mins apart. Another might be the old ‘coincidence’ of anomalies, though supposing the track does end in the SIO without major position indicator gaps, that, reasonably, would rule both out.
Pending such ‘ratification’ by science, would description of the basis for GDTAAA be best characterised as a theory?
• A suggestion. About these ‘major position indicator gaps’ might I suggest that as part of any tracking presented in validation you utilise 2 minute steps, and highlight all position indicators, or at least table the gaps in them? I note that your most recent 04:22 to 05:00 UTC plots do not make clear what the gaps are, though you described the overall average at your 070808 June.
A question does arise as to when such a gap should be considered excessive, though I think that would have to depend on whether it is on a straight leg and also the density of anomalies.
• Further on that and about the above ‘coincidence of anomalies’ possibility I am unclear as to which of the anomalies in the 04:22 to 05:00 most recent sequence are ‘progress’ hits, if any other than at 04:22 and 04:40 UTC, ie whether they include the several close shaves.
You will see how there can be questions about coincidence when you look at the random close shaves highlighted below on the 05:00 plot below. https://www.dropbox.com/s/h5pprmncr31tl35/Illustrative%20of%20anomaly%20ambiguity.jpg?dl=0
Also about the progress hits, another suggestion is that on any validation track you use the SNR/drift anomaly colours in place of grey for hits and a new colour for a position marker.
• An observation about your comment at 070808 June, that at 04:48 the GDAS wind was 33 knots from the west, farther out at sea level it was of similar strength but from 060˚ so if the 33 kt from ahead continues along the track we can expect to see quite some shear in the descent.
@David,
GDTAAA is a work in progress. We are in the middle of a number of validation tests and therefore any attempts to develop a set of rules for the GDTAAA process must also be seen as a work in progress.
My goal is to understand the capabilities and limitations of GDTAAA and then to assess whether the capabilities of GDTAAA add value to the search for MH370, when taking into account the limitations of GDTAAA.
There may be some new insight(s) into aircraft detection using LF, MF, HF and VHF WSPR data that many expert and experienced radio amateurs have not previously considered. It would be unscientific to write off my findings without justification, just as it would be unscientific to accept my findings without justification. I am making a number of observations and at the same time seeking explanations for these observations.
Meanwhile I value and welcome your observations and questions.
When an aircraft turns, climbs or descends the cross-sectional area of both the aircraft and its wave vortex as presented to a set of radio transmissions will change. Rule 7 defines an average expectation based on observations to date. The aircraft cross section and wave vortex cross section are both subject to a large volume of literature in the context of radar detection and tracking, but comparatively few studies have been made with regard to LF, MF, HF or VHF radio transmissions.
You make a number of important points and suggestions on how to further refine the expected effect on radio transmissions of turns, climbs and descents. At the moment I am focused on collecting observations and establishing patterns. These observations and patterns require explanations using theoretical physics. I do not claim to have all the answers, only some ideas.
In general terms, I see a parallel to the early analysis of the Inmarsat satellite data and discussions about the Burst Timing Offset (BTO) and Burst Frequency Offset (BFO). Ray tracing, transmission paths, propagation, attenuation, path loss, reflection and refraction of WSPR radio transmissions are directional and distance issues akin to BTO. Doppler shift, frequency modification, scatter and frequency drift are relative velocity issues akin to BFO. In the case of WSPR the process can be influenced for example by the solar cycle, in the case of BFO the process can be influenced for example by the solar eclipse of the Inmarsat satellite. In both cases, there may be additional influences that are not immediately apparent but can be identified and accounted for. Some analysts are still arguing that we are not modelling the eclipse correctly in using the BFO data.
I invite you to set up a blind test. Select an aircraft and a start location (but not a departure airport, as that gives too much away) and a start time and ask me to detect and track the aircraft. I not willing to interrupt the current planned validation test series, but I am happy to add such a test to the list. The AMSA RNZAF Orion P3 flight on 28th March 2014 was blind apart from the departure airport being Pearce AFB and several locations in the search area taken from the GPS data in the photographs supplied under a Freedom of Information initiative by Brian Anderson. The NZ Defence Force are currently searching their archives for the flight path data having identified the mission start date as 27th March 2014 21:43:25 UTC and mission stop date as 28th March 2014 10:26:18 UTC. How much of the mission time includes pre-flight briefing and post-flight debriefing is still being clarified. AMSA have different times for the ETD and ETA.
The MH370 FMT path using GDTAAA avoids following flight routes. The chances of MH370 being confused with another aircraft crossing the track of MH370, obviously not keeping to flight separation rules and also not following official flight routes, at the same time as MH370 are negligible and can be discounted in my opinion.
GDTAAA is a hypothesis, as it is a supposition or proposed explanation made on the basis of limited evidence as a starting point for further investigation. GDTAAA may become a theory, when it becomes a supposition or system of ideas intended to explain something, especially one based on general principles independent of the thing to be explained. (Definitions taken from the Oxford Dictionary of English).
GDTAAA rule 8 precludes making any large changes and only allows small changes with a justification. Your concern of “major position indicator gaps” is moderated by both progress indicators and a track consistent with aircraft performance and operational conditions.
GDTAAA rule 9 already implements the first part of your suggestion to document the small gaps.
Your dropbox depiction of ambiguity is misleading as you are using WSPR anomalies at 05:00 UTC against previous position indicators at 04:22 UTC, 04:40 UTC and 04:48 UTC. The only valid comparison is with WSPR anomalies at the time of the position indicators.
As you can see in the linked chart at 05:00 UTC the closest alternative position indicator was rejected:
1. The blue dot is the predicted position of the aircraft (no gap identified, no adjustment required).
2. The red dot is the closest alternative position indicator, but is rejected as it is 7.96 nm distance away (two minutes flying time at 400 knots equates to 13.3 nm).
https://www.dropbox.com/s/xsv6gayz0v0phzy/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200500%20UTC%20Rejected%20Alternative%20Position%20Indicator.png?dl=0
If there are two close alternative position indicators, I follow both for the next 2 to 6 minutes and you quickly see that one flight path is viable and the other flight path is not viable. I have the advantage of already having run GDTAAA for the next 30 minutes and seeing what happens when the AMSA Challenger reaches the area close to the Indonesian fishing vessel D2S2.
It was obviously a mistake not to publish all two minute charts output from GDTAAA. My mistake was in thinking that in a long cruise flight out to the search area, publishing every two minutes of progress is boring for readers. My apologies!
I will make the following changes going forward.
WSPR links will be marked with lines in blue for dual anomalies, red for SNR anomalies, green for drift anomalies and yellow for stable links. Stable links will normally not be shown to reduce clutter.
Present aircraft predicted position will be marked in a large light grey marker.
Present aircraft actual position will be marked in blue based on a position indicator, red based on a SNR progress indicator, green based on a drift progress indicator and white based on no indicator. The actual aircraft position will be in a small marker and will not completely overlap the predicted aircraft position marker. If the two markers are concentric that means there was no adjustment to the aircraft’s predicted position and no update to the aircraft’s actual position.
Previous two minute markers will be marked in dark grey, if based on a position indicator, otherwise in light grey.
@Richard,
@David,
Richard writes: “If there are two close alternative position indicators, I follow both for the next 2 to 6 minutes and you quickly see that one flight path is viable and the other flight path is not viable.”
It is coming along.
As time goes by and you proceed on the one flight, you (Richard) are answering questions glaring us in the face, such as what you just wrote, quoted above.
All this step by step explanation will help in providing some background to your, or GDTAAA, decisions concerning Selection and Discrimination. The glaring question “Why did you choose that one ?”, concerning single “anomalous” links as progress indicators, or crossing links as position indicators, will always arise, so please be prepared to be able to answer them all. What I have just written, pretty well “off the cuff” indicates to me that for each significant point you need to have a database entry with a simple explanation. Perhaps you are already prepared for this.
I’m not sure why you have chosen green now for drift anomalies, other than for example for me personally, I found it difficult to discriminate between your shade of orange and your red links.
“Previous two minute markers will be marked in dark grey, if based on a position indicator, otherwise in light grey.” THIS will be a significant improvement.
Independently, I have been following your tracking of VH-XNF Sortie #4 on a simple orthogonal plot using various coloured dots or symbols to indicate various step results including what I may consider questionable, or for which I may have a separate question.
One thing I have been impressed by that is the dots in your tracking of VH-XNF Sortie #4 do not wander from side to side as much as I had expected.
“The red dot is the closest alternative position indicator, but is rejected as it is 7.96 nm distance away” I’m not sure that you have made it clear that this means 7.96 nm away FROM the continuing flight path to the predicted position, as I take it to mean, and not just simply 7.96 nm away from the blue dot predicted position itself. Please confirm/deny.
You invite David to set up a blind test.
Independently I had been going to suggest the unknown C130J flight as being a good candidate for you to search, find and track, SO LONG AS NO ONE GIVES YOU THE FLIGHT PATH DATA BEFOREHAND, as we know from the AMSA Media Release: “two Royal Australian Air Force P8A Poseidon maritime surveillance aircraft and one C-130J Hercules, who were also able to deploy life-rafts to the vessel”.
Of course there is also the second P8A “hovering” around somewhere.
I, personally think that the blind flight, or flights, should be May 2021 flights in the same environment as the VH-XNF flight.
@David, back to you.
@Richard,
I highlighted SO LONG AS NO ONE GIVES YOU THE FLIGHT PATH DATA BEFOREHAND here so anyone reading here might not accidentally hand it over beforehand. (Path and Timing)
Properly Reading what is on your linked chart 05:00 UTC I see that it is actually the distance from the blue dot (position predicted from both speed and track) to the red dot which is 7.96 nm.
But Avg.GS=349.8 and FL=220 still better suit a C130 than the Challenger being tracked.
…… already in descent ……
@George G,
In my comment yesterday to you at 16:17 CET I stated: “The Hercules RSCU211 Sortie 5 definitely did not overlap with the AMSA Challenger 604 jet VH-XNF RSCU550 in Sortie 4. RSCU550 arrived back in Perth at 09:47:00 UTC almost an hour before RSCU211 arrived at Pearce AFB coming directly from Richmond AFB. A third 6 hour sortie to the SIO by the Hercules would have required extra crew due to rest rules.”
On 14th May 2021 at 05:00 UTC the RAAF Hercules C-130J registration A97-442 call sign RSCU211 was 51m 48s out of Richmond AFB on the other side of Australia and could not possibly be descending along side the AMSA Challenger 604 registration VH-XNF call sign RSCU550 in the middle of the SIO.
Please look at the FlightAware ADS-B data I published in full in my comment 26th May 2021 at 13:28 CET.
@All,
Continuing the validation test with the AMSA Challenger SAR jet (registration VH-XNF and call sign RSCU550) en-route to the search area to help the Indonesian fishing vessel D2S2 (Sortie 4), here is the output from GDTAAA from 05:02 UTC to 05:14 UTC.
Please note that the annotation scheme has been updated as follows: WSPR links will be marked with lines in blue for dual anomalies, red for SNR anomalies, green for drift anomalies and yellow for stable links. Stable links will normally not be shown to reduce clutter.
Present aircraft predicted position will be marked in a large light grey marker.
Present aircraft actual position will be marked in blue based on a position indicator, red based on a SNR progress indicator, green based on a drift progress indicator and white based on no indicator. The actual aircraft position will be in a small marker and will not completely overlap the predicted aircraft position marker. If the two markers are concentric that means there was no adjustment to the aircraft’s predicted position and no update to the aircraft’s actual position.
Previous two minute markers will be marked in dark grey, if based on a position indicator, otherwise in light grey.
https://www.dropbox.com/s/tlov1ugf6frerpv/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200502%20UTC.png?dl=0
https://www.dropbox.com/s/g30clhs85b10di4/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200504%20UTC.png?dl=0
https://www.dropbox.com/s/eg1xitoik3slnxs/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200506%20UTC.png?dl=0
https://www.dropbox.com/s/komk1ykk69l3ld5/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200508%20UTC.png?dl=0
https://www.dropbox.com/s/5eexs04vpyc1f5r/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200510%20UTC.png?dl=0
https://www.dropbox.com/s/voyhr8oafh4cv9u/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200512%20UTC.png?dl=0
https://www.dropbox.com/s/zbuq27cy11lf8z0/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200514%20UTC.png?dl=0
The Indonesian Fishing Vessel D2S2 is drifting in strong winds and its position is also being updated every 2 minutes.
@Richard,
Your data links, 8 June 2021 at 17:06, for 5:08 and 5:14 UTC.
At 5:08 UTC the plot text at the bottom shows a longitude of 103.980°E whereas the plotted point is more consistent with 102.980°E.
Similarly,
At 5:14 UTC the plot text at the bottom shows a longitude of 103.225°E whereas the plotted point is more consistent with 102.225°E.
Additionally at 5:14 UTC the plot text at the bottom shows a latitude of 30.825°S, whereas were the longitude actually 102.225°E as above then the corresponding latitude would be very close to 30.864°S.
The lat/long of 30.825°/103.225° is the plotted position from when at 05:06 UTC.
https://www.dropbox.com/s/axrdgnu0hjjixwc/y%20%20-3.877E-02x.pdf?dl=0
@George G,
Well spotted! The typos in the two labels have been corrected:
https://www.dropbox.com/s/komk1ykk69l3ld5/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200508%20UTC.png?dl=0
https://www.dropbox.com/s/zbuq27cy11lf8z0/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200514%20UTC.png?dl=0
That is what comes of bypassing the automatic process and manually updating the code to the new annotation system after having generated under the old annotation system. Going forward I will have updated the template generator to the new annotation system.
@Richard. Re your 081405, thank you for your comprehensive response, which adds to understanding of your work and its underpinning.
I will look into establishing a suitable blind test for later, a non ADS-B flight.
Re hypothesis vs theory, I was thinking of Darwin’s, which yet remains a theory.
Yes I understand that my Dropbox pointers on your 05:00 UTC plot preceded the timing of the position there, its anomalies being at that time. My intent was to use those as just random, as described; illustrating that random indicators could hit or narrowly miss the markers in any track.
This is to do with apprehensions of the general risk of indicators arising from random anomalies, though I do understand that you are confident that this would become apparent subsequently – you note that, the consequent flight path will not be “viable”.
In the example that follows that comment you have the advantage of being able to discriminate by knowing the general heading of the flight path, though of course that does not undo the confidence in being able to discriminate without that.
I join George G. in thanking you for designating the progress and position indicators as the plots proceed with a continuing record of the position indicators. The replacement of orange with green does help the clarity.
Thanks too for your 081706’s continuing plots. However, I notice (line 4) you list them as extending to 05:12. I was about to thank you for the bonus 2 mins since in fact they extend to 05:14. However since 05:10 is missing and the markers of the last two are consistent with earlier spacing, I take it that those should be captioned 05:10 and 05:12, the change of annotation has intruded yet again?
@David,
You stated: “However since 05:10 is missing and the markers of the last two are consistent with earlier spacing, I take it that those should be captioned 05:10 and 05:12, the change of annotation has intruded yet again?”
I do not follow.
05:10 is not missing and shows the aircraft at 30.845°S 102.735°E FL120.
05:12 is correctly captioned and shows the aircraft at 30.854°S 102.490°E FL100.
05:14 is correctly captioned and shows the aircraft at 30.863°S 102.225°E FL080.
@Richard. Apologies. I missed it even after sending out a search party. That correlated with 05:14 being an extra. My mistake.
@David,
No worries!
@Richard,
@David,
I still think the Hercules C130J flight which I still presume took place from Perth out west over the water, would be a very good candidate for a blind test.
At this time we don’t know it’s details, time or course detail.
All we know, if it occurred, is that this was sometime between arrival early evening local Perth Time on Friday 14th and taking off Saturday afternoon for the return journey to Richmond. (Richard, these are Sorties 5 and 9 in your list in your comment 26th May 2021 at 13:28 CET.)
I first suggested that VH-XNF left the immediate vicinity of the rescue to allow the Hercules in to drop more life rafts
“(As almost an aside, and purely speculation, I have wondered if this was a farewell pass before flying out of the immediate vicinity, so that the Hercules could arrive and drop more life rafts.)” on 22 May 2021 at 15:11 CET.
That was a serious comment at that time, but, NO EXCUSE, I had totally neglected to properly account for the various time zones (ringing around in my head).
In my more recent comment:
“Is this my possibly imaginary, presumptuous, Hercules on the way to drop more life rafts, but at the wrong time ?” on 07 June 2021 at 14:00,
I was being totally tongue-in-cheek, due to “GS 349.8 knots” being approximate C130J cruise speed.
(Yet, I still hadn’t woken up to my time and time zone confusion.)
My apologies for this mis-attempt at a bit of levity.
______________________________________
Seriously, a blind test searching for, finding and tracking a May 2021 flight would be preferable to a 2014 flight in my view as there should be more data due to more WSPR stations than in 2014.
AND the tracking of VH-XNF on Sortie 4 should provide some “groundwork” for comparison.
______________________________________
@George G,
I accept your proposal to find, detect and track the RAAF Hercules C-130J (registration A97-442, call sign RSCU211) sortie to the Indonesian Fishing Vessel D2S2 somewhere after its arrival at Pearce AFB on 14th May 2021 at 10:43:48 UTC and its departure back to Richmond AFB from Pearce AFB on 15th May 2021 at 05:28:28 UTC.
I will first complete the validation test based on the AMSA Challenger VH-XNF sortie 4.
For my next validation test I have decided to keep to my original plan to base the test on the 29th March 2014 (Day 22) RNZAF Orion P-3C sortie.
Then I will run the blind test on the Hercules.
Then I will look at the Lufthansa 747-800 over Germany and finally MH371 on 7th March 2014 from Beijing to Kuala Lumpur.
Schedule noted. Good.
@All,
Continuing the validation test with the AMSA Challenger SAR jet (registration VH-XNF and call sign RSCU550) the aircraft now arrives in the search area to help the Indonesian fishing vessel D2S2 (Sortie 4), here is the output from GDTAAA from 05:16 UTC to 05:24 UTC:
https://www.dropbox.com/s/53f170x6rrxexl5/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200516%20UTC.png?dl=0
https://www.dropbox.com/s/1kaooyq4wnroeqw/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200518%20UTC.png?dl=0
https://www.dropbox.com/s/jbohfwyyf7l41wx/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200520%20UTC.png?dl=0
https://www.dropbox.com/s/9hm2chzobipwgv3/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200522%20UTC.png?dl=0
https://www.dropbox.com/s/0reucuc2zlit5gf/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200524%20UTC.png?dl=0
You will see at 05:16 UTC there are two possible position markers. The closest position marker in green can be reached at a ground speed of 450.96 knots, which is within the maximum cruise speed of the Bombardier Challenger 604 at 468 knots. The more distant position marker in red can only be reached at a ground speed of 556.81 knots, which is well above the maximum cruise speed and is therefore discounted.
@Richard,
Comments on 0516 through 0524 GDTAAA Output
as presented by your – Richard on 10 June 2021 at 16:54 –
0516: You explain why you chose one possible position marker from the other, based on the realistic GROUND speed of 450.96 knots for one compared to the other which requires a ground speed of 556.81 knots. Please note that if there was a wind from 060 degrees of 90 knots then this would be below 468 knots airspeed. But, please, would you expect the search aircraft to still be at cruise speed at what you have as 600 feet altitude.?
More to the point, in my viewpoint, you now need to begin to explain why you think that either of the possible position markers actually represents the position of the aircraft. What is your reasoning other than it may be probable.?
You have also introduced “Inst. GS”, in addition to “Avg. GS”. I’m not sure that you have yet explained “Inst. GS”, or I may have been lax and missed it. I’m sure it will help to define it in context.
0518: The red dotted possible position shown is a “progress indicator” as you have previously defined. But, in similar vein to the previous question above concerning 0516, have you any reason for this being a “possible” other than that it is simply possible and it reasonably fits with the expected behaviour of the aircraft, or more exactly, it’s crew.?
0520, 0522: No additional comment or question.
0524: How do you justify the red dot as a possible or hypothetical position.?
In accordance with “established procedures” (my terminology) the Grey Two Minute Marker you have also indicated (and presumably at the same continuing previously established speed) should be the only possibility One (You or I or anyone) should even consider.
If “One” were to introduce the fact that the aircraft was in a left turn then this might be reasonable, but you need to explain why you chose the crossing point (crossing of the link with “anomalous” SNR with your chosen flightpath. Or, perhaps, more to the “point”, why did you choose the implied turn rate that you have ?
It does imply that you are using other data, or information. Perhaps you have investigated alternative flightpaths, and are using the next one or two (or three) points (two minute samples) which imply the crossing point indicated by the red dot. (Without having yet informed us, the readers)
0524: Question 2: You have the aircraft now at 100 feet above the ocean. What justification have you for this.? Or to put it in other words, what evidence do you have.?
0524: Overall Comment; Were it not for your selection and placement of the red dot, in lieu of what might be implied by, or in accordance, with “established procedures”, then the other questions above would have waited til later.
@George G,
The assumed altitude at 05:16 UTC is 6,000 feet (FL 060.0) and at 05:24 UTC is 1,000 feet (FL010.0).
As I have said before a limitation of GDTAAA is that it does not provide altitude. The wind speed and direction changes sometimes quite significantly with altitude and may possibly allow an estimation of altitude, assuming a constant TAS and using changes in the GS. The wind at 05:16 UTC at 6,000 feet was 34 knots from 345°T.
Inst. GS is an abbreviation for instantaneous ground speed and is calculated across the two minute leg.
A progress indicator can only provide a possible position.
A position indicator can provide a probable position.
At 05:24 UTC I assume the same ground speed and change the track to match the progress indicator line.
@Richard,
Update on – Comments on 0516 through 0524 GDTAAA Output
as presented by your – Richard on 10 June 2021 at 16:54 –
And subsequent to your comment/reply of 12 June 2021 at 09:36:
Please accept my apologies for mis-read of the altitude indication.
For 0516 I misread FL 060.0 (6000 feet altitude) and miswrote as 600 feet.
For 0524 I misread FL 010.0 (1000 feet altitude) and miswrote as 100 feet.
Thank you for the correction.
Update on 0516:
Your more recent comment giving wind conditions at 6000 feet further explains your rejection of the the more distant position marker in red which not only requires a ground speed of 556.81 knots but also an unrealistic airspeed.
Thank you for defining instantaneous GS in context.
Update on 0524:
Now understood that you used an assumption of continued flight at previous ground speed of 379.7 knots and changed track to meet with the potential progress indicator line (the red SNR anomaly link) at a position consistent with that continued same ground speed.
My only comment is that there may be a slightly increased uncertainty upon that progress position compared to other previous progress positions based on continued flight along previous track.
Any such increase in uncertainty is yet to be evaluated.
@George G,
A key part of the GDTAAA process is to look ahead and extrapolate the current flight path continuing at the assumed ground speed and assumed track. When that look ahead presents viable future position or progress indicators that is good news. When that look ahead leads to a dead end, then you have to ask the question did the aircraft turn, climb, descend, speed up or slow down. For example, it is perfectly possible that an aircraft turns, climbs and slows down at the same time.
There may be an increased uncertainty when there are neither position or progress indicators and anomalous WSPR links fail to pick up the aircraft, but there is absolute certainty that the aircraft did not just disappear.
In my view, it is perfectly reasonable that the aircraft is not picked up every two minutes by anomalous WSPR links, especially when in straight and level flight. It is not a failure of the GDTAAA system or process, when we have to wait a few minutes for the next progress or position indicator.
@All,
Continuing the validation test with the AMSA Challenger SAR jet (registration VH-XNF, call sign RSCU550, Sortie 4) the aircraft having arrived in the search area now reaches the location where the first AMSA video and photographs of the Indonesian fishing vessel D2S2 were taken. Here is the output from GDTAAA from 05:26 UTC to 05:46 UTC:
https://www.dropbox.com/s/abea6orxmfe0dqh/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200526%20UTC.png?dl=0
https://www.dropbox.com/s/17183ye3jiqxruf/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200528%20UTC.png?dl=0
https://www.dropbox.com/s/hcvgdy2tx0ak9wu/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200530%20UTC.png?dl=0
https://www.dropbox.com/s/6h4yd9p4togtjhg/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200532%20UTC.png?dl=0
https://www.dropbox.com/s/uy3d1dzuuioqms5/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200534%20UTC.png?dl=0
https://www.dropbox.com/s/nr4w0jqro5k1vav/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200536%20UTC.png?dl=0
https://www.dropbox.com/s/jc9tnhxr4ud8aoo/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200538%20UTC.png?dl=0
https://www.dropbox.com/s/lw6sews7yrft8y7/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200540%20UTC.png?dl=0
https://www.dropbox.com/s/u4jp7z2lkfxf1vn/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200542%20UTC.png?dl=0
https://www.dropbox.com/s/csrsslyy27lp2ub/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200544%20UTC.png?dl=0
https://www.dropbox.com/s/snx7rca06lrvwxj/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200546%20UTC.png?dl=0
The first photograph was taken by the AMSA Challenger jet at 05:43:51 UTC and the position of the aircraft is shown on the GDTAAA output at 05:44:00 UTC. The position error in the GDTAAA output is 5.3 nm using the “established procedures”.
At 05:44 UTC there is neither a position indicator nor a progress indicator so the previous ground speed and track is assumed. The assumed track of the aircraft is 275°T, whereas the actual track is 296°T. Using a look ahead to the next GDTAAA output at 05:46 UTC and retrospectively correcting the GDTAAA output at 05:44 UTC the track is 288°T, which is only 8° different from the actual track.
https://www.dropbox.com/s/nt1yoo26zq9ghvo/GDTAAA%20AMSA%20VH-XNF%20Sortie4%2014MAY2021%200544%20UTC%20Corrected.png?dl=0
A position error of 5.3 nm and track error of 8° is a very positive result from this validation test.
@Richard. 05:28 doesn’t work, for me at least. Would you re-post that please?
@Richard. No need, now does.
Dropbox anomaly.
@David, @All,
Global internet traffic increased 40.3% due to Covid-19 and large Dropbox files can be a problem to download. GDTAAA output files are on average < 500 KB and normally not a problem.
Two hours of the full WSPR data is 23.8 MB and that can be a problem to download. I have decided to limit WSPR data to a maximum of 1 hour of the anomalous data. The following link to the Excel data is for the time period 05:00 UTC to 06:00 UTC on 14th May 2021 and covers the recent GDTAAA output for Sortie 4 of the AMSA SAR flights to D2S2 and is 4.2 MB if you wish to download the file:
https://www.dropbox.com/s/0khbstkf4me0dqv/wsprspots-2021-05-14-0500-0600%20UTC%20Anomalous.xlsx?dl=0
@Richard. Challenger VH-XNF’s track on approaching the stricken vessel that you have detailed raises some issues as to whether, firstly, any alternative track utilising the same anomalies are feasible and secondly to what extent the track might have differed had there been no AMSA video of the arrival.
As I understand it, ground speed within realistic airspeed bounds is deduced from GDT-AAA derived track, having subtracted wind; that is it is an output, not an input. So derivation of projected track is iterative with speed acceptability, sensible speed changes increasing flexibility as to 2 minute track options, those selected being the most likely.
So on the first issue above I have prepared indicative possible alternative tracks to yours, indicative because they are by hand and also speeds variously could be different from those I use.
My start point of 05:26 is where you show an alternative so I started from that but the start point could be elsewhere.
In doing this, I take no notice of the position of the vessel, as if this were a blind trial, so intended therefore to plot track while ‘unaware’ of that position and its timing.
About altitude, you have observed that GDT-AAA cannot find that, though anomalies are more likely in descents and ascents. I have speculated that anomaly density might reduce at low level and the combined effects of that with those of the descent you postulate might suggest a descent to low level. However on running through your plots from the descent start to flight at low level, that is not obvious to my eye. Indeed the anomaly density at low level at 05:46 tends to belie it.
Without knowledge that the aircraft would have descended, and without a GDT-AAA clue that it had, the track from the current start of the descent, somewhere between 04:48 and 05:00, would alter I expect if the aircraft stayed at altitude, ground speed varying with wind changes from 32 knot westerly you mentioned earlier and also with any selected airspeed changes consistent with an unalerted flight.
To investigate that would entail redoing around 25 subsequent markers to see what the track would have become without the descent. That is beyond me and I do not suggest it is warranted at all though I expect it to be unlikely to result in the same track as during descent.
Putting that aside and looking into alternatives to the descent track of your analysis, for simplicity as to speed, arbitrarily I utilise the distances covered in your corresponding 2 minute increments. Of course these could be adjusted as appropriate for optimisation but that is unnecessary I think for these purposes.
In taking no notice of the vessel’s position in all this I have felt if unnecessaryagain to follow alternative leads right through to the video start time but you will see that several at least are likely to differ from your conclusion as to the final position error from comparing that to the vessel’s at the same timing.
Related and adding to those two issues of my first paragraph, markers at the apex of sharp turns can be realistic if a turn curve could be fitted though them and their neighbours. I believe that can be done if the leg adjoining either side is not straight, or close to, for more than one marker. However, my outcome being indicative I do not observe that constraint.
One more relevant feature is that naturally a course change reversal, first in one direction then markedly in the opposite, can be seen as sub-optimal in the choice between alternative tracks, so that once there is a course change there will be an inclination to favour future course changes in that direction, as I have done and you might well have. However this risks overlooking a deliberate course change reversal.
The below illustrates workings but becomes a jumble. For the summary please go directly to the last.
https://www.dropbox.com/s/rvkmd91e40yd5ic/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0001.jpg?dl=0
https://www.dropbox.com/s/6aw6ji4z14p0bdc/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0002.jpg?dl=0
https://www.dropbox.com/s/dgg7zto3x90lpiu/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0003.jpg?dl=0
https://www.dropbox.com/s/1kq4fkmywh5iz7l/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0004.jpg?dl=0
https://www.dropbox.com/s/pi8qtstffa9w36q/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0005.jpg?dl=0
https://www.dropbox.com/s/2k2spgizssin5iv/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0006.jpg?dl=0
https://www.dropbox.com/s/lkfoiaw83h1isxv/Alternative%20tracks%20approaching%20stricken%20vessel%2C%20VH-XNF%2C%20early%2014th%20May%2C%202021_0007.jpg?dl=0
@Richard. Please add an ‘h’ in front of the second last URL.
@David,
Done!
@David,
I am not sure I correctly understand what you are trying to say and believe you have proven or disproven in your latest comment.
Let me feed back to you what I think you are saying and ask you to correct me where I have misunderstood.
1. Blind Trial.
You state: I take no notice of the position of the vessel, as if this were a blind trial, so intended therefore to plot track while ‘unaware’ of that position and its timing.
This is clearly not a blind trial. We know where the vessel is and we know where the AMSA SAR aircraft is at certain times from the photographs and videos provided by AMSA.
I have never claimed this is a blind trial.
You appear to be trying to make it into a blind trial, when that cannot be done in my view without appearing to be artificial or contrived.
2. Alternative Paths.
I have explained how I deal with alternative paths. I follow both alternatives and “look ahead”. Sometimes just a two minute look ahead is sufficient, sometimes four, six or eight minutes are required. In the case of MH370 following the military radar trace from the last civilian radar point at 18:00:51 UTC along the assumed flight path taken from the Beijing Lido slide of the military radar data, I looked ahead twelve minutes from 18:02 UTC to 18:14 UTC. There is no WSPR detection at 18:02, 18:04, 18:06, 18:08, 18:10, 18:12 or 18:14. The GDTAAA output is shown in Appendix B of my paper titled “Global Detection and Tracking of Aircraft as used in the Search for MH370” dated 1st May 2021. My alternative flight path however immediately detects and tracks the aircraft with multiple progress and position indicators. Since aircraft do not disappear, I conclude that the alternative flight path that I have shown in my paper is correct.
Your alternative path in the case of the AMSA SAR Sortie 4 fails at the latest in your slide 4 after 6 minutes at 13:32 UTC. Aircraft might not be detected every two minutes along a flight path but to assume an aircraft disappears without further trace is also unacceptable.
3. Best Fit or Worst Fit.
You show multiple branches at each point in time and include all the alternative paths, whether they are a best fit or not and whether in the “look ahead” they show a consistent flight path or not.
I agree that any point in time there are multiple alternatives.
Some alternatives can be discounted because they are beyond the aircraft’s performance capability.
Some alternatives can be discounted by a “look ahead” and soon run dry with no further detections.
Some alternatives can be discounted because they do no correspond to a consistent flight path.
I followed the best fit consistent flight path and end up with an error of 5 nm and 8° of track.
You appear to be saying, that if you follow the worst fit inconsistent flight path, then you will end up with a much greater error. I agree! But isn’t that obvious? What do you think that proves? In my view, that proves nothing.
4. GDTAAA Proof of Concept.
You appear to conclude that this test has not proven or validated the GDTAAA concept. I agree!
What I am saying is that this test, which is clearly not a blind trial, shows it is possible to detect and track an aircraft in straight and level flight with a medium degree of position accuracy (excluding altitude) of under 18 nm and on average at around 5 nm. Further it is possible to detect and track an aircraft in turns, climbs, descents, accelerations and decelerations with a low degree of accuracy, assuming a consistent flight path and allowing retrospective changes following a look ahead process.
5. Next Steps.
I am currently planning a blind trial, but I cannot say much because I do not know much (I am blind, so to speak). The test has been set by Mike Glynn a former Qantas pilot and uses a flight that operated without passengers, that you will not find in ADS-B archives. Since Mike Glynn was the pilot of the flight, he has the flight track data and will publish the data only after I have published the GDTAAA output for the flight.
Geoffrey Thomas has kindly agreed to be the independent adjudicator of the WSPR Technology and GDTAAA Blind Test. Geoffrey Thomas is four time winner of the prestigious Royal Aeronautical Society (RAS) Aerospace Journalist of the Year twice each in the ‘Systems and Technology’ and ‘Business’ categories. He is the editor of Airliner Ratings .com and has run several articles on the new WSPR Technology.
@David, @Richard, @RobW
David, you wrote: “I have speculated that anomaly density might reduce at low level”.
I had even speculated the opposite, in the mistaken consideration whether or not an aircraft wing area would present a bigger interruption to downward travelling radio waves, about to bounce of the surface of the ocean, if the aircraft was at low altitude.
Referring to the linked attachment, it becomes clear that “altitude”, in context of normal aircraft traffic, will not have any effect.
https://www.dropbox.com/s/o7qtfdkvk2kd41a/-%20the%20above%20chart%20is%20Fig.33.pdf?dl=0
@George G,
Johannesburg Airport is at a high elevation. Whilst watching the candidate aircraft for the blind test sitting on the take off runway at an “altitude” of 5,530 feet I was looking to see that there were no WSPR anomalies in the vicinity.
Take off was at 21:29 UTC. At 21:24 UTC and 21:26 UTC indeed all was quiet, but at 21:28 UTC I picked up both a SNR and Drift anomaly intersecting at exactly the location of the aircraft. It is possible that the engine exhaust caused the WSPR anomalies especially if the aircraft had wound the engines up and started to roll.
In general, GDTAAA has a severe limitation in that it cannot determine altitude.
@George G. Were there some sort of ground effect could we expect that to cause a SNR anomaly?
Looking at Richard’s 05:44 UTC and its adjacent 05:46, the Challenger passing the vessel, there was a marked increase in the latter in drift anomaly density out to beyond 75 miles for some reason.
In that 05:46 there was an increase also in SNR (in the blue) anomaly density near the aircraft as it passed. Possibly the aircraft was lowest as it passed the vessel, and banked. Thus if caused by the aircraft it could have been from either or both.
Or else the increase in both types was random, rendering further speculation useless.
@Richard. “I have never claimed this is a blind trial”. No, I understood that. My second issue raised was, about the track approaching the vessel, “…to what extent the track might have differed had there been no AMSA video of the arrival”, in other words the difference had it in fact been a blind trial.
I was trying to figure how it would have gone and how much your descent timing and track depended on advance knowledge of the vessel’s position. As I discussed, apparently there were no GDTAAA-sourced cues to its timing at least.
So if we can generalise from this particular example, it is difficult to see how in a blind trial such a descent would be detected in any flights.
I am left with the impression that turns mixed with descents will be a hard nut to crack.
My other issue was,”…whether, firstly, any alternative track utilising the same anomalies are feasible. Should have read ‘is’ but that aside, the answer to me appears to be yes.
You say, ““Your alternative path in the case of the AMSA SAR Sortie 4 fails at the latest in your slide 4 after 6 minutes at 13:32 UTC. Aircraft might not be detected every two minutes along a flight path but to assume an aircraft disappears without further trace is also unacceptable”.
I think you misunderstand. It simply has not been detected at that two minute mark. I draw an arc depicting its travelled distance while “missing”. When it reappears I can then retrospectively place it on that arc. A like “disappearance” is at your 05:32 UTC. Instead of using an arc you use a dead reckoning position. The only difference is that you suppose retrospectively that that DR position was valid. True it will be if the course remains straight. In your case it doesn’t, though you work from that position even so.
Re your, “You appear to be saying, that if you follow the worst fit inconsistent flight path, then you will end up with a much greater error. I agree! But isn’t that obvious? What do you think that proves? In my view, that proves nothing.”
I showed that even while being constrained by arbitrary speeds, there were alternatives from that one sample point. I see that at your 05:32 you adjusted speed to fit an anomaly.
My markers were just indicative. There would be at a lot more from other points. My examples and the others could be optimised. It is not obvious to me that all would end up being “worst fit”.
Besides, because we do not have the details of the actual track beyond ADS-B range we do not know how accurate your replication of it is, even though the end point is just 5 miles off.
I agree that what I have proved is nothing and that wasn’t my intent. .
“You appear to conclude that this test has not proven or validated the GDTAAA concept. I agree!” That is a consequence, yes, though not an aim. The test proved worthwhile all the same, being part of validation testing.
As to the most welcome Mike Glynn blind test/trial, I hope it incorporates turns and of course that it is successful, thereby implicitly endorsing your major findings of the MH370 turns north of Sumatra together with, in advance, the track that follows that, yet to come.
Detecting and tracking descents such as some possible in MH370’s final moments looks to be a separate issue.
@David,
I agree with you what you say and it is a fair assessment of the capabilities and limitations of GDTAAA.
We can be fairly confident that GDTAAA can detect and track an aircraft in straight and level flight. It appears that turns produce more WSPR anomalies than straight flights, so detecting a turn is not that difficult. Detecting and tracking an aircraft whilst both turning and climbing or descending is much more difficult and the limitation of GDTAAA in not being able to determine altitude forces a reliance on detecting any associated changes in ground speed resulting from a turn, climb or descent.
As a result, GDTAAA will be of limited value in determining the flight path of MH370 at the end of flight, if indeed the aircraft was in a spiral dive. GDTAAA would be able to detect a long glide, but we know from the analysis by Mike Exner and Don Thompson of the right wing control surfaces that have been found as floating debris, that it is quite possible that one or more control surfaces separated from the aircraft before the crash. It is even possible that the entire right wing separated from the aircraft before the crash. In either case, without certain right wing control surfaces or without the right wing, there was no long glide.
@Richard. OK thanks.
@Richard. In my 07:58, 6th para down, ‘A like “disappearance” is at your 05:32 UTC’.
Correction for the record, that 05:32 should read 05:34.
@David,
I think we agree that when there are multiple alternative flight paths we should follow each flight path until we reach a “dead end”.
You raise an important point on how to handle the absence of either a progress or position indicator and how to determine when an alternative flight path has reached a “dead end”.
We have observed that for any given GDTAAA output every two minutes along an assumed flight path based on an assumed ground speed and assumed track, it is reasonable to expect that there might be a single GDTAAA output without either a progress or position indicator.
Occasionally it is possible to have two such GDTAAA outputs in a row without either a progress or position indicator (using 2021 WSPRnet data).
Currently I consider a 2021 flight path alternative a “dead end” after three successive GDTAAA outputs in a row without either a progress or position indicator. With the much lower WSPRnet volumes in 2014, I extend this rule to four in a row before concluding a “dead end”. In general I start questioning any flight path after two successive GDTAAA outputs without any indicator.
@Richard. Yes, in six of my seven alternatives, otherwise replete with position and progress indicators, there were six with dead ends (I called misses).
As I mentioned in your case, at 05:32, you treated a dead reckoning substitute position as real when positioning the indicator that followed.
Instead, I drew an arc, alllocating a position retrospectively when the following indicator was known. Indeed in one there were two consecutive dead ends. In that case having drawn the first arc I fitted that (and its successor, the same length track between each), once an indicator was apparent for the third.
Obviously the indicator position, whether after one or two dead ends, had to be compatible with their positions. In all my cases that proved quite easy.
Too easy in fact; and therein lies a problem. Such a dead end provides an unwarranted flexibility in placing the indicator follower. It is too easy to find a suitable place for that when one can now juggle with the two legs preceding. So a dead end increases uncertainty as to whether, when its following marker is “placed” on an indicator and it is now retrospectively fixed on its arc, those positions are indeed valid.
The worst part is that with successor legs now being dependent on the accuracy of those placements that can lower confidence in the whole outcome.
When there are two consecutive dead ends that provides even more flexibility, supposing of course that there is not overstretch in reaching the indicator finally and that the resulting track is coherent.
Your use of the DR in the track is based on a supposition, which is only likely to be true if the track remained straight. If it doesn’t, as in your case, the same problem arises. Why, retrospectively suppose that the track did lead through that? It introduces the same sort of uncertainty that I have attributed to my arc approach.
With any dead ends included I do not see how you can have other than reduced confidence in the whole outcome, with two, or even worse three, then even worse, unless, again they are all on a straight course. But even that conclusion would be questionable.
@David,
It appears that we are going around in circles trying to track an aircraft going around in circles! 🤪
Do you agree with following rules:
1. In straight and level flight we can have a miss now again as long as there are frequent position indicators to confirm the ground speed and the track and frequent progress indicators to check the ground speed.
2. In a turn it is safe to assume that a bank angle of maximum 20° is used. For example, at 500 knots ground speed, the maximum turn is around 90° in 2 minutes.
3. We usually observe a position indicator or at least a progress indicator during a turn.
4. A turn can frequently lead to alternative flight paths that all need to be checked.
5. Alternative flight paths can sometimes be discounted as beyond the aircraft performance capability or ending in a “dead end”.
6. Any remaining alternative flight paths must be checked until a position indicator is found. If no further position indicator is found then the alternative can be discounted.
7. GDTAAA is unable to determine altitude and tracking may not be reliable when an aircraft is turning at the same time as climbing or descending.
@Richard. 2-5 and 7 yes, though about 3, the 05:32 dead end you depict as being in a turn was not observed..
1. Yes, so long as you know what ground speed is. Depends on wind effect, so course, and piloting selections. Also with a dead end, where ground speed is known even though dependent on course, where in the arc from the previous indicator that was. It was not necessarily at the DR and its successor is not necessarily on the same track as it was.
6. Yes but not the reverse. If a convenient and plausible position indicator is in fact found having worked from a DR or assuming a straight line from a previous indicator, no guarantees that it would have hit that position indicator had the missing indicator not been misssing.
@Richard. Sorry, cross eyed again for 05:32 please read 05:34
@Richard, @David,
Richard says: “I think we agree that when there are multiple alternative flight paths we should follow each flight path until we reach a “dead end”.”
Comment: Certainly. Can’t see how otherwise you can be “sure”.
Richard, I’d be careful about naming “recent” observations as “rules” until you have more information; i.e. have further added to your database.
“Alternative flight paths can sometimes be discounted as beyond the aircraft performance capability or ending in a “dead end”.”
This one is, however, a basic tenent. But it applies FOR an aircraft in NORMAL flight and under NORMAL control.
Richard, your 3. – – “We usually observe a position indicator or at least a progress indicator during a turn.”
This brings us to two quandaries:
Firstly: refer to my comment above about prematurely labelling recent observations as “rules”;
Secondly: I am still finding it quite remarkable that you are actually finding what appear to be indications from “links” which may possibly have another reason for them to be “anomalous”. The basic question will always arise, methinks, HOW and WHY do you think they are relevant ?
You will eventually need to be very clear in your answer, regardless of how much database information is gathered.
You say: “It appears that we are going around in circles trying to track an aircraft going around in circles! 🤪” (which irony I very much appreciate), but I very much hope that you will proceed through to the end of flight.
As “we” follow you through your process, more pitfalls and posiitives are being uncovered.
Richard, on 14 June 2021 at 15:12, and I think at at least one other time you have written: “Since aircraft do not disappear”, or something similar.
In the larger context perhaps this may be an unfortunate turn of phrase.
David, you ask: “Were there some sort of ground effect could we expect that to cause a SNR anomaly?”
I very much doubt it, but we are learning some sensitivities of the weak radio signals to factors probably not previously considered significant (RobW ?).
I, personally, have no reservations, for example, in considering that thermal effects and associated circulating eddies, such as Richard mentioned at Johannesburg in his comment on 15 June 2021 at 08:33 have noticeable effects.
For the Record, I have not yet fully reviewed the recent charts and comments.
@George G,
You state: “As “we” follow you through your process, more pitfalls and positives are being uncovered.”
Pitfalls and positives! Limitations and capabilities!
Oh really! Or is that just our perception?
If you like irony, try this: We are discussing WSPR noise and drift with our own noise and bias, if you catch my drift. 😂
A good read on the subject of noise and bias is The New York Times Bestseller:
“Noise: A Flaw in Human Judgment” by the Nobel prize winner Daniel Kahneman (Author of Thinking, Fast and Slow) co-authors are Olivier Sibony and Cass Sunstein.
In my view, the proof of the pudding is in the eating.
You can only say something is a success after it has been tried out. You cannot say something is a failure if you have not tried it out.
@Richard. My remarks at 14:08 were coloured by limitations evident in the descent analysis. However constraining your 1.to straight and level flight as in fact you did, and I overlooked, I agree with that, in the circumstances you describe.
It is more understandable that in straight and level there will be occasional position indicators absent – and progress indicators are no real substitute.
Careful we must be not to chuck out the baby with the bathwater.
Your 6. continues to pose the difficulty I have raised. In level flight after a turn, looking at alternative flight paths, when there is no position indicator found I believe that in
looking for one that fits subsequently that should be done in conjunction with retrospectively placing the missing marker so that the combination are consistent in both speed, (on a distance arc) and track with what has gone before, and what comes after.
Hence my use of arcs in place of your dead reckonings. Indeed treating DR’s as valid could deceive as to whether then hitting a position indicator (or near enough) based on that is realistic.
Of course in any retrospective ‘fitting of the facts’ there is the risk that that there might be a second position indicator which would fit but yet in so doing would make the track inconsistent with a supposed continuing turn. Thence that possibility can be overlooked when in fact the turn had stopped (or even reversed, though that could well be most unlikely in MH370’s case).
Related, noting your, “3. We usually observe a position indicator or at least a progress indicator during a turn”, I think it is comparatively easy to fit with/utilise a progress indicator if the track and it meet at an acute angle, and do not mean much.
Moreover I think that though there are unavoidable risks in coping with a no-position-indicator marker, to rely on a progress indicator between two no-position-indicator markers is risky, two consecutive no-position markers being worse again. Here I am not talking about small turns that are consistent before and after with the retrospective fitting and which might escape generating an anomaly, but, invoking you 3., in the sort of turns that you see MH370 making while off Sumatra, including the FMT.
While it could be argued that confidence in the subsequent track could retrospectively restore overall track confidence, the trouble is that subsequent markers though based on anomalies, could be construed as popping up by chance. Indeed that is a risk in all of this tracking, that it can be seen reasonably as possibly by chance.
So my hope is that the blind trial you envisage includes MH370 type turns, in level flight or, if not, another does.
Changing to a related topic, because we both acknowledge that tracking of turns in a descent is unreliable I will not belabour that with further observations about the Challenger descent track except to observe that your inputs started its descent at between 04:48 UTC and 05:00. If that about matches reality, all the track following the first turn after that, at 05:06, is in doubt; and indeed there is a speed surge late on plus various hefty turns without accompanying position or progress indicators.
What I now raise is whether the Challenger pilot would have left the vessel abreast some 16 miles away during that descent, then extended that to 37 miles distant. At that point, at a nominal 1000 ft., apparently he turned partly back to it. Presumably though he had located the vessel by radar just before turning towards it, when some 70 miles away. Finally then at that low level he then approached the vessel for the first time though finishing cross wind.
What I have in mind is that is that in fact he had arrived before that first video. I say that because there was a liferaft in the water already in that video, drifting across the bows of the vessel (supposing the surface wind was from 060˚).
By my estimation of the arrival time of the RAAF P8, from the Flight Radar track you posted, it would not have arrived by then, so most likely the video started after the Challenger VH-XNF had first arrived and then dropped it. According to the AMSA media release it dropped several.
The two that were secured to the vessel as per the video following, some 2 hrs plus later, look to be of a different type so i imagine the P8 had dropped those before then.
Finally, about progress indicators I think they could be described better as ‘possible progress indicators’, but a mouthful. Hence some acronyms for brevity might be PI’s, PPI’s and NI’s (the last for ‘no-indicators’)
@David,
Once again an excellent analysis and a fair assessment of the capabilities and limitations of GDTAAA.
I agree that the key is having regular position indicators. A progress indicator or even a non-indicator between position indicators is easy to resolve. I also agree that a progress indicator between two non-indicators is meaningless.
After a series of non-indicators, I believe we should roll back to the previous position indicator and then look for an alternative forward path. If we eventually find a series of position indicators then we can have confidence that the alternative flight path is a probable candidate.
Progress indicators in between position indicators can only be regarded as a possible flight path as the assumption is that there was a straight and level flight between the position indicators.
For example in the blind test the first 10 indicators after take off comprise 3 position indicators, 5 progress indicators and 2 non-indicators. The flight path taken after take off can be determined, because there is only one point with a valid alternative that leads to a dead end confirmed by 4 non-indicators in a row.
@Richard. Thanks. 3rd para, yes, good.
@George G. Re your comment to Richard, “I very much hope that you will proceed through to the end of flight.” I imagine that there would be a straight climb out, which circumvents any problems associated with turning while climbing, though the descent at Perth might be suss.
Your, “I have not yet fully reviewed the recent charts and comments.” Look forward to that.
@David, you say on 17 June 2021 at 10:39,-: “there is a speed surge” (during the descent of VH-XNF). I interpret that as being due to a wind change between 16,000 and 14,000 feet altitude as might be implied by my table of Average Ground Speed, from Richard’s GDTAAA output plots.
You also say at 10:46,-: “I imagine that there would be a straight climb out”. My bet is on a climbing turn based on an inference I have made previously.
I have some residual questions concerning the tabled speeds output from GDTAAA, but only after further (simple) analysis and comparison will I make any queries.
I had made pen and paper notes as I was first downloading and reading Richard’s last batch of plots 0526 through 0546 and have made more notes since, but you and he have batted these about enough for a while.
As time goes on I intend to conduct or initiate a dissection of at least one point, not necessarily yet “on the board”. But you and Richard and time have been reducing the number of moot points.
I have been plotting the path simply without regard to keeping latitude and longitude distances to the same scale.
My attempts at discrimination between the various types of points or indicators are just to keep some sort of overview. The two linked plots indicate what I have been playing with. They may help me to keep some sort of perspective.
https://www.dropbox.com/s/4inlr70zd8ru1fq/OPlot.pdf?dl=0
https://www.dropbox.com/s/c2h6f3sulufq5q3/Zoom1Plot.pdf?dl=0
@George G,
Nice plots! Much better than the GDTAAA output.
During the tracking of the AMSA Challenger 604, I should have paid more attention to the wind speed and direction and how it was changing over time, position and the assumed altitude. Since the Indonesian fishing vessel was in the middle of a storm, you would think that is obvious! It could not have been easy for the pilot of the Challenger 604 either.
In running the blind test the wind plays a dominant role. As GDTAAA does not tell me altitude, I have to make an assumption that the initial cruise altitude was FL310. I expect that there will be step climbs later during the flight. Along the flight path followed by the blind test the winds are very strong and it is essential to frequently check for changes in the wind speed and direction.
@George. Thanks. My “alternatives” plot could have done with your presentation skills.
@Richard. In your page 17 ‘Discussion’ of the MH370 turns around Sumatra and subsequent in the below you say, “The flight path is complicated with a number of turns and changes of speed “.
In this initial paper you make no mention of altitude or ascent/descent. Noting the GDTAAA weaknesses identified with that, it is relevant to the reliability of the tracking of the turns you identified and also to the nature of a suitable blind trial now.
Did you see those turns as being level please?
https://www.dropbox.com/s/tvtfw14tfyfljie/Global%20Detection%20and%20Tracking%20of%20Aircraft%20in%20the%20Search%20for%20MH370.pdf?dl=0
@David,
I assumed that MH370 maintained the same cruise altitude throughout the FMTs in my initial paper titled “Global Detection and Tracking of Aircraft as used in the Search for MH370” dated 1st May 2021.
@Richard. OK, thanks.
BTW while I used the termed that a GDTAAA ‘weakness’, your ‘limitation’ would have been more appropriate.
@Richard, @David,
On 17 June 2021 at 17:16, I wrote: “I have some residual questions concerning the tabled speeds output from GDTAAA”. I have since, from scratch, produced a speed check for my simple needs.
Richard, I now can see (with AMSA VH-XNF on Sortie #4) you started off doing with “Avg.GS” as would be expected.
But after having introduced “Inst.GS” you seem to have thrown the baby out with the bathwater. That must be confusing for others not persistent enough to follow your attempts at keeping up with changes in flight profile.
In your reply you did say that you should have paid more attention to wind conditions and how they were changing over time. You also said that “In running the blind test the wind plays a dominant role.” I was hoping you would cut your teeth a little more on your previous schedule, finding more about the process on the other validation testing you had planned (including completion of AMSA Sortie #4). It is probable the process still needs tweaking, and this tweaking would seem “easier” with flights where you might be able to ask and research about the flight when or where a challenge is found.
@George G,
I had hoped to have received some feedback from the RNZAF with the flight details for the AMSA MH370 surface search sortie on 28th March 2014 before running a similar scenario based on the RNZAF sortie on 29th March 2014. Unfortunately we are still waiting for the data.
GDTAAA measures ground speed and track, but I now run a consistency check on Altitude, Air Temperature, Weight, Fuel, Mach, TAS, Wind Speed, Wind Direction, GS and Track.
Inst. GS is now taken over the leg from the previous position indicator and Avg. GS since the start of the flight. With a precision of 18 nm, the Inst. GS taken over a 2 minute leg may not be indicative of the aircraft’s true progress.
GDTAAA cannot determine altitude so an initial Flight Level of FL300 or FL310 is assumed depending on track, but I do check for evidence of step climbs. A move to a flight level 2,000 feet higher can be accomplished by most aircraft in 2 minutes, so once again it may be difficult to detect within the GDTAAA precision.
Meanwhile, I have started work on the blind test, but that will take some time to complete.
Many thanks for your patience with the failings of GDTAAA, which remains a work in progress subject to continuous improvement.
@Richard,
Good to hear (or read) “I now run a consistency check”.
Methinks the pilot would be aiming to maintain a consistent IAS during the various stages of the flight of VH-XNF, even if the Challenger may be capable of meeting some challenging requirements.
@Richard. Answer to a basic question if you would.
You have said, “The position accuracy of the order of 18 nm is largely due to the use of the 6 character Maidenhead Grid for both the Tx and Rx stations (not GPS latitude and longitude co-ordinates) and the great circle path modelling based on a WGS84 datum surface using an oblate spheroid model of the earth’s surface.”
In quantifying the former, also you said, “The six character Maidenhead grid does not give the precise location of a TX or Rx station but is accurate to within 5.6 nm (10.4 km).”
About that 5.6 nm in particular, in my dabbling with alternative tracks between position indicators, using progress indicators as stepping stones, I took no account of the flexibility that might offer with both – and also no-indicators, i.e. lack of both.
Had I, the options for alternatives would have multiplied. I imagine the same would apply to any tracking.
While many might be discarded as unrealistic, others too might be left aside if apparently inconsistent with likely track.
However that might prejudice careful examination of the apparently ‘unlikely’, which might include such as an unexpected diversion or turn reversal. One of these might be overlooked?
Also, since denial of aircraft presence is another tool of GDTAAA as we have discussed, could it be concluded that there was no aircraft evident because a track apparently became discontinued, when by allowing those tolerances it might be on track still?
@David,
The worst case for a single anomalous WSPR link is an error of 18 nm. In most cases a single link will be more accurate, but I agree that there is a lot of room for error and progress indicators should be used with caution both in determining the presence or the absence of the target aircraft at a particular time and position. At 500 knots in two minutes a target aircraft will have flown 16.7 nm, so the granularity of the WSPR data is of the same order of magnitude as the precision.
Fortunately, aircraft do not usually change their track by any significant amount every two minutes. If the aircraft is in LNAV mode and heading for the next waypoint in the flight plan, then the ground speed and track will be stable across a number of two minute WSPR link legs.
A flight plan can be described in short form in a few lines. It will contain departure and destination airports as well as waypoints and flight routes. At each waypoint the air speed (TAS or MACH) may be defined along with the flight level. A long form flight plan will take up several pages and at each waypoint additional information such as air temperature, wind speed, wind direction and fuel remaining will be forecast.
Position indicators where two anomalous WSPR links intersect at an angle are much more accurate. Better still are position indicators where three anomalous WSPR links intersect, where the error will be under 2 nm. In my view, position indicators are the key to determining the track of the target aircraft, but on average you will only observe a position indicator every 14 minutes. Progress indicators can be very helpful, but equally they can be very distracting. Progress indicators are observed far more often, but equally as often as no indicator at all.
There is a hybrid progress-position indicator where two or more anomalous WSPR link progress indicators run close together in parallel. On one hand the anomaly is confirmed by multiple links, on the other hand the subsequent links add no new information with regard to ground speed or track of the target aircraft.
When the target aircraft position is predicted two minutes further on based on an assumed ground speed and track and at that point there is no indicator, it is a fallacy to assume that the target aircraft must have turned. In the absence of any other information, I let the flight path simulation run for several two minute legs looking for a position, progress or hybrid indicator.
I have observed that when the target aircraft is in straight and level flight there are fewer WSPR anomalies. The contrary is also true, that when the target aircraft is turning, climbing or descending there are usually a larger number of WSPR anomalies.
If you predict the flight path of a target aircraft for twenty minutes in two minute steps at the current ground speed and track and find no indicators, then I would assume there was an error in the tracking at some point. I would only make major adjustments in the ground speed and track based on position indicators. Fine adjustments in track based on progress indicators where the assumed ground speed is matched are also reasonable in my view.
@Richard. My thanks for that explanation and expansion. There is a lot of acquired skill/judgement involved.
As it appears to me, the outcome is error resistant but with no error-free guarantee, particularly with aircraft presence denial.
Software sorting may be feasible but would need careful judgement of the results?
You say, “I agree that there is a lot of room for error and progress indicators should be used with caution both in determining the presence or the absence of the target aircraft at a particular time and position.” Can I take it that that caution would apply to position indicators also, including those that cross at a point, since that is quite likely coincidental, cocked hats being the expected norm?
Also you say, “Fortunately, aircraft do not usually change their track by any significant amount every two minutes. If the aircraft is in LNAV mode and heading for the next waypoint in the flight plan, then the ground speed and track will be stable across a number of two minute WSPR link legs.”
Coming to your ‘working hypothesis’ of the north Sumatra turns and FMT, they would not be part of a flight plan or expected normally I would have thought. So I assume that projected tracks petered out and thence you searched for alternatives? You then were able to determine there was just the one, while allowing those error boundaries?
At some later stage it might be worth looking into whether GDTAAA ‘denies’ aircraft presence on the UGIB alternative route, including radar points you disregard?
@David,
You wrote: “At some later stage it might be worth looking into whether GDTAAA ‘denies’ aircraft presence on the UGIB alternative route, including radar points you disregard?”
This is the quandary we are in, if we were to believe the previously agreed (military) monitored radar track through to 18:22:12 when “The last Malaysian Military Radar capture was at 18:22:12 UTC about 10 NM after passing waypoint MEKAR in the Malacca Strait.” as written in the UGIB report “The Final Resting Place of MH370”, dated 7th March 2020.
In Richard’s “Global Detection and Tracking of Aircraft as used in the Search for MH370” dated 1st May 2021, he states in the first paragraph of “Results” on Page 5 of 24, the following:
“Figure 3 shows two WSPR links at 18:00 UTC that cross at the position of MH370 determined
from the last civilian radar data point. The flight path was traced along the assumed flight path
towards waypoint VAMPI and beyond, but unfortunately there were no further WSPR detections.
The resulting set of graphics are shown in Appendix B.”
In Appendix B, Richard has shown us that he followed the previously determined track of MH370 along through Waypoint VAMPI onto Flight Route N571. For seven two-minute periods, 18:02 through 18:14, he did not find a WSPR indication for any one. So he concluded a dead end to this track. Note the corresponding GDTAAA charts are Figures 31 through 37, (even though Figs 35, 36, 37 are mis-numbered 31, 32 and 33).
A quandary.
@Richard. P.S. A cocked hat is not a cocked hat only when its spread exceeds error margins?
@All,
A blind test of GDTAAA and the WSPR technology was devised by former Qantas pilot Mike Glynn. Geoffrey Thomas kindly agreed to be the independent adjudicator for the test.
The start point was Johannesburg Airport and the intended destination was Sydney Airport, but the aircraft had to divert to an unspecified alternate airport. The aircraft had mechanical failures and could not be repaired in Johannesburg, so was being flown without passengers.
Mike Glynn gave the following start point on the runway at Johannesburg Airport:
· Takeoff Position – 26° 8’ 0.49” S 28° 14’ 18.00” E.
· Altitude – 5,530 feet.
· Date-time UTC 21:29Z 18 February 2019 (only accurate to 30 seconds unfortunately).
Before beginning the test, I checked the GDTAAA output before the take off between 21.24 UTC and 21:28 UTC:
https://www.dropbox.com/s/m162u0z3znuljkx/GDTAAA%20Blind%20Test%2018FEB2019%202124%20UTC.png?dl=0
https://www.dropbox.com/s/9ar0kt1z1d9472d/GDTAAA%20Blind%20Test%2018FEB2019%202126%20UTC.png?dl=0
https://www.dropbox.com/s/jn9xs780660ley1/GDTAAA%20Blind%20Test%2018FEB2019%202128%20UTC.png?dl=0
All is quiet at 21:24 UTC and 21:26 UTC. There is a single anomalous WSPR link at 21:28 UTC near the target aircraft’s position.
Here is the initial output from GDTAAA from 21:30 UTC every two minutes until 22:12 UTC:
https://www.dropbox.com/s/7xil9prs23k27q1/GDTAAA%20Blind%20Test%2018FEB2019%202130%20UTC%20N.png?dl=0
https://www.dropbox.com/s/ps1hio4167hsvda/GDTAAA%20Blind%20Test%2018FEB2019%202132%20UTC%20POS.png?dl=0
https://www.dropbox.com/s/cz4bwff5mg70kg2/GDTAAA%20Blind%20Test%2018FEB2019%202134%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/x48h94cwgw6zltq/GDTAAA%20Blind%20Test%2018FEB2019%202136%20UTC%20N.png?dl=0
https://www.dropbox.com/s/lazti7b2pp6kwyh/GDTAAA%20Blind%20Test%2018FEB2019%202138%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/gljeiyr551szt90/GDTAAA%20Blind%20Test%2018FEB2019%202140%20UTC%20N.png?dl=0
https://www.dropbox.com/s/vegfas4hkizx2kq/GDTAAA%20Blind%20Test%2018FEB2019%202142%20UTC%20N.png?dl=0
https://www.dropbox.com/s/geixh3mhal0w8bc/GDTAAA%20Blind%20Test%2018FEB2019%202144%20UTC%20N.png?dl=0
https://www.dropbox.com/s/2kekqvtd99mw4uv/GDTAAA%20Blind%20Test%2018FEB2019%202146%20UTC%20PRGPOS.png?dl=0
https://www.dropbox.com/s/s7rtbtouzlqatr0/GDTAAA%20Blind%20Test%2018FEB2019%202148%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/999287ntucklizb/GDTAAA%20Blind%20Test%2018FEB2019%202150%20UTC%20PRGPOS.png?dl=0
https://www.dropbox.com/s/jrw3st2unjeeu6i/GDTAAA%20Blind%20Test%2018FEB2019%202152%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/nvddth20gixvv1j/GDTAAA%20Blind%20Test%2018FEB2019%202154%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/jfmxnbwui8ogblz/GDTAAA%20Blind%20Test%2018FEB2019%202156%20UTC%20N.png?dl=0
https://www.dropbox.com/s/tl7qnf8slpobhd7/GDTAAA%20Blind%20Test%2018FEB2019%202158%20UTC%20N.png?dl=0
https://www.dropbox.com/s/g0cp80fu8oa5wqj/GDTAAA%20Blind%20Test%2018FEB2019%202200%20UTC%20POS.png?dl=0
https://www.dropbox.com/s/rcb4j4wxwrylj4u/GDTAAA%20Blind%20Test%2018FEB2019%202202%20UTC%20N.png?dl=0
https://www.dropbox.com/s/s7e84kto9f5jdec/GDTAAA%20Blind%20Test%2018FEB2019%202204%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/hgtrk6e65hep1hi/GDTAAA%20Blind%20Test%2018FEB2019%202206%20UTC%20N.png?dl=0
https://www.dropbox.com/s/nziof7bgqr1j282/GDTAAA%20Blind%20Test%2018FEB2019%202208%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/73mwft1ej64sehr/GDTAAA%20Blind%20Test%2018FEB2019%202210%20UTC%20PRG.png?dl=0
https://www.dropbox.com/s/kw9ejzdgrzflqfr/GDTAAA%20Blind%20Test%2018FEB2019%202212%20UTC%20POS.png?dl=0
The file naming convention includes the date-time UTC as well as a suffix denoting the type of indicator (N = No Indicator, PRG = Progress Indicator, POS = Position Indicator, PRGPOS = Hybrid Indicator).
The flight route appears to follow a “Direct To” waypoint APDAK or AVAVA, then passes waypoint ETMAL and follows flight route UZ8. The flight path passes just north of the RBV VOR (Richard’s Bay) and follows the flight route UQ5 to waypoint GEVIS, where the aircraft exits the Johannesburg Regional FIR into the Johannesburg Oceanic FIR.
Out of these 22 GDTAAA outputs, there are 3 position, 8 progress and 2 hybrid indicators. At 9 positions there is no indicator. This underlines the issue raised by @David as to possible valid alternatives, that may have been missed. However, if you only take the 3 position indicators and ignore the rest, you would still conclude that the flight path passed over waypoint AVAVA, passed north of the RBV VOR and passed close to waypoint GEVIS.
The flight route followed is further north than a typical departure toward Sydney via waypoint APMAT and the TGV VOR and exiting to the Oceanic FIR at waypoint ANVED or further south at 3142S03247E. This raised the question as to why a more northerly route was selected. Obviously flight routes do not always follow the great circle path exactly. Apart from diverting around areas of bad weather, it has long been known that aircraft can benefit from the jet streams. For example, flights from Los Angeles to Tokyo follow a great circle route (typically 11 hours) and flights from Tokyo to Los Angeles follow a jet stream route (typically 9 to 10 hours):
https://www.dropbox.com/s/dbf36qezw2aubvp/Los%20Angeles%20-%20Tokyo%20-%20Great%20Circle%20vs%20Jetstream%20Routes.png?dl=0
In the case of the blind test, flights from Johannesburg to Sydney benefit from strong tailwinds from the southern polar jet stream that circumvents the globe at around 50°S. However, the jet stream on 18th February 2019 was quite different to the same day in previous years, and was much further north for the first half of the flight path. If I am right about the departure flight route, then this may explain why the flight path entered the Oceanic FIR at waypoint GEVIS further north than usual.