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
Many thanks for pointing out the error in the numbering of the figures in Appendix B of my paper titled “Global Detection and Tracking of Aircraft as used in the Search for MH370” dated 1st May 2021. I have now corrected this error:
It might help to solve the quandary you highlight by completing the GDTAAA output for the MH370 flight path not just from 18:02 UTC to 18:14 UTC but until 18:30 UTC after completion of the SDU reboot at the 1st Arc and assuming the flight path defined in the UGIB report “The Final Resting Place of MH370”, dated 7th March 2020.
We can then put the two alternative hypothetical flight paths side by side for the time frame between 18:02 UTC and 18:30 UTC and see what position and progress indicators are observed.
If this first step proves helpful, this exercise could be extended to after passing the 2nd Arc at 19:42 UTC.
Here is the GDTAAA output for the next leg of the blind test of the WSPR technology from 22:12 UTC to 22:48 UTC.
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 traverses the Johannesburg Oceanic FIR from waypoint GEVIS. Out of these 19 GDTAAA outputs, there are 3 position, 8 progress and 1 hybrid indicators. At 7 positions there is no indicator.
Your Blind Test Data posted on 26 June 2021 at 15:01.
Point 21:32 – You have shown the point as a “Position Indicator”.
I have not done so in my “Overall Plot”.
The inherent uncertainties in the actual indicated position on the GDTAAA Output plot (26.320°S,28.470°E) need to be accomodated. The GDTAAA output (or your supplementary output) for estimates of aircraft speed this early in the flight, being 380.3 Avg.GS and 466 Inst.GS, may be misleading. The resultant estimates for IAS, without taking wind into account are higher than for both the next two two-minute samples, or periods. (Currently, I have no information concerning winds prevailing at the time of the flight.)
As the aircraft had just taken off, the assumed groundspeed required to meet the requirement to reach the indicated point at the requisite time would seem a little hopeful.
In other words, it seems unlikely that the aircraft reached the point indicated by the GDTAAA Output plot.
HOWEVER, had the point been shown as a region of uncertainty about the crossing of the red and green lines on the plot, then this may have been easier medicine to swallow.
I have classed the result for 21:34 as “IN DOUBT” because I can’t remember if and how you have justified using such a “glancing blow” as a progress indicator. As seen in the GDTAAA output plot for 21:34 it is being used more as a “direction guide”, or even a possible confirmation of previously assumed direction, than as a progress indicator. Similar comments apply to 21:50, 21:52, 21:54 and 22:10. I have refrained from classing these as “IN DOUBT” on the Overall Plot.
21:38 is certainly IN DOUBT, and deserves comment for more than one reason.
Firstly, and importantly, you have attempted to stretch the aircraft performance to breaking point to suit some imaginary target (or that is how it seems to me).The ground covered in the two minutes is 23.68 (and a little bit) nautical miles, as I have calculated it. I have searched for a typographical or numerical error without success. But that searching is not really necessary. All one has to do is to look at the GDTAAA Output plot for 21:38. The red dotted line on the chart is considerably longer than the previous two two-minute leg lengths shown on the plot. This step length (the red dotted line) is actually so long that I have the aircraft easily breaking Mach 1.
The other main problem with 21:38 is you seem overly keen for Mike Glynn, the pilot, to get onto an agreed flight route. The Red Dotted leg length ends up at a crossing point of one red line (a WSPR link withan SNR anamoly) and a standard flight route. In contrast, “Your mission should you wish to accept it” is to track the aircraft. You have already been advised that the route (overall) was not a standard route. Even though it is to be expected that the pilot will be flying along standard flight routes for most, if not all, the flight, the task is to use GDTAAA to track the flight. Of course, you may have simply have mixed up lines on charts.
22:04 – In this case, by just looking at the GDTAAA Output for 22:04, one might ask: “Why did he divert off the previously followed path here ?”. Looking further on it seems you had scouted ahead and found the path you took to be the best option. Or were you aiming at Waypoint GEVIS ?
22:08 – Here you deviated from the previous four minute track (two two-minute legs). Clearly you ended up on an anomalous WSPR link. But why choose that one ?
The Latitude value given in the plot for 21:46 would seem to be a repeat of the previous one. I have used 27.550°S in my Overall Plot. Both Latitude and Longitude given in the plot for 22:06 were not consistent with the data. I have used 29.136°S and 33.111°E.
Overall plot attached.
And as I send I see you have completed another batch.
There was a typo on the GDTAAA output.
I missed to update the latitude from the previous 27.342°S at 21:44 UTC to the actual 27.522°S at 21:46 UTC.
I have now corrected this error.
@Richard, Two comments:
22:20 – (32.170°S,35.033°E)
On the GDTAAA Output plot, and very close to the No Indicator position you have chosen, is a WSPR Drift Anomalous Link (green).
In many of your previous decisions concerning where the flight path was heading you would have made a slight diversion to this “green line” on the plot, and would have termed it a “Progress Indicator”.
This would have been consistent with your 21:34, on which I made previous comment concerning treating such a “glancing blow” as a progress indicator.
Why did you not do this with the green anomaly for 22:20. ?
Or, do you have specific knowledge that it was caused by something else. ?
In fact had you chosen to divert to the green anomaly the resultant point would have been almost midway between the Pre (22:18) and Post (22:22) points.
Or are you trying to make a point. ?
For 22:26 you actually did chose to make a slight diversion to a similar “green line”,
although, as you will see following, I now doubt this green line is really so similar.
You also chose to call this a “Hybrid” Indicator, by your use of “PRGPOS” in the file name.
I reviewed your previous two uses of “PRGPOS”.
The first, 21:46, I chose to call a “Position” indicator in my overall plot OP1, and I admit I did not notice your use of “PRGPOS” at the time (or more likely the import). The intersection of two adjacent close to parallel, but crossing, anomalous links might possibly indicate some influence by something close to their junction, such as the aircraft you are trying to track. Particularly at the next stepping stone (the blue dot on the GDTAAA Output plot) in the straight line extrapolation of the previous flight path (route) is within 8 nautical miles of the junction.
With hindsight, I might now be tempted just to call it another “Progress” indicator.
But, perhaps justifiably, you have given it it’s own type designation, PRGPOS.
The second use of your PRGPOS designation is at time 21:50. Here, another two close to parallel links, both drift anomalous, cross close to another stepping stone. You chose to call this PRGPOS. I just treated it as if just another Progress indicator in my OP1.
Returning to 22:26, and looking at the plot closely, what I just considered a slight aberration may simply be two virtually parallel drift anomalous WSPR links passing by another stepping stone.
I seriously doubt, if that is the case, that this type of occurrence should be considered any differently to any other possible “Progress Indicator”. (Although having two links similarly affected may increase the chance that the hypothetical aircraft in the vicinity and under attempt-to-track may have had some influence, and not something different altogether.)
My apologies again!
There was another typo in the labelling on the GDTAAA output at 22:06 UTC.
I incorrectly updated the latitude and longitude in the label, which should be 29.135°S 33.112°E.
I have now corrected this error:
In both cases the chart was correct, only the annotation was in error.
Many thanks for the constructive criticism and pointing out the two errors in the labelling of the GDTAAA output, which I have corrected.
There is a fundamental problem that the WSPRnet data is only accurate to 18 nm. At 500 knots an aircraft flies 16.7 nm in two minutes. So each point I plot can be 18 nm out in any direction. At a very worst case two successive points can be 36 nm apart, if the 18 nm error goes in opposite directions. Of course, the errors get smoothed out after a while.
I am plotting the points where the anomalous WSPRnet link data “thinks” the aircraft is, not where the aircraft actually is. If an aircraft moves 16.7 nm and there is an 18 nm error on top, then it makes it look like the aircraft moved 34.7 nm in two minutes, which is 1041 knots and obviously not possible.
To solve this problem I average the ground speed and track over legs between position indicators. The two minute markers will be erratic, but the smoothed flight path between position indicators marked with a dashed dark grey line is more representative of the leg flight path. It takes a while for the average ground speed and average track to settle.
The aircraft has no passengers and luggage and is therefore lighter than usual. The initial acceleration and climb is not unrealistic.
I am not assuming that the waypoints and flight routes in the departure and arrival areas were followed precisely. My flight path prediction is 12 nm north of the VOR RBV at Richard’s Bay for example. Once the aircraft is beyond waypoint GEVIS in the Oceanic FIR, I am not assuming any waypoints or flight routes at all. I am just hunting for the next position indicator.
I have introduced the dashed dark grey line between position indicators as a more reasonable representation of the flight path from which the leg ground speed and track are derived. I agree with you that making minor adjustments of ground speed and track based on progress indicators is problematic, but there are clues in the data which I do not want to ignore completely. Strictly updating ground speed and track only every 14 minutes on average at a position indicator seems to miss out on these interim clues.
My working hypothesis is that WSPRnet radio waves are affected by aircraft (in the case of the blind test a Boeing 747-400) in three different ways:
The Boeing 747-400 with a take off weight of 376 MT and a wing surface area of 525 m2 will reflect, deflect and scatter radio waves.
Four RB211-524G engines with 58,000 lbf thrust each, produces massive short lived turbulence behind the aircraft with a flow rate of 687 m/sec and an exhaust gas temperature of 526°C.
The air density changes in the exhaust due to exhaust particles and water vaporisation will scatter radio waves.
As a Boeing 747-400 moves through the air at around 500 knots it creates a wake turbulence that can last 3 minutes and be felt over 20 nm behind the aircraft. The wake turbulence descends at around 500 fpm behind the aircraft.
The air density changes in the wake due to water vaporisation will scatter radio waves.
The WSPRnet “footprint” of a Boeing 747-400 is quite possibly up to 20 nm long.
The “footprint” will increase when the aircraft is turning, climbing or descending.
On 29 June 2021 at 11:09, you wrote: “I have introduced the dashed dark grey line between position indicators”. Noted and appreciated. Helps in smoothing and perception when a straight flight path is being followed.
There was an error in the colour coding on the GDTAAA output at 22:06 UTC which was spotted by George G.
I have now corrected this error:
Here is the GDTAAA output for the next leg of the blind test of the WSPR technology from 22:48 UTC to 23:16 UTC.
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).
@Richard. Steaming along.
You are having to look ahead a long way. 40 mins and 20 markers between the last two positions.
Re the nominal climb of 20,000 ft. between 23:02 and 23:04, presumably spread over more, how are you determining altitude please?
I have position indicators at 21:32, 22:00, 22:12, 22:36, 22:48 and 23:16 UTC.
The average gap is 20.8 minutes.
GDTAAA cannot determine altitude. There were three parallel SNR anomalies at 23:02 UTC and my suspicion is that a step climb of 2,000 feet from FL310 to FL330 may have been in progress.
Oceanic waypoints are commonly defined simply by a latitude/longitude. At 23:02 UTC the position was close to 33S41E which could have been used as a waypoint. This might have marked a turn and/or a step climb.
@Richard. Thanks. The 22:48 escaped me. So 28 mins and 14 markers. Still a fair way. Where would you draw the line in such circumstances and start thinking about a possible turn?
Yes, I follow re the clue to a climb (2000 ft, not 20.000!) and/or turn, the 23:16 position indicating a climb, .
So as I understand it, the 31,000 before was based on what could be expected of that aircraft at that weight and speed and the 2,000 ft would be a standard step climb?
In the middle of the SIO with little other traffic, I am prepared to wait for a position indicator for much longer than in a high density traffic area. However 28 minutes is getting close to my limit.
I have little clue when step climbs were actually performed, but I expect at some point the weight and fuel schedule would dictate FL310 to FL330, then possibly FL350, FL370, FL390 and FL410. It depends also on the tail winds at the various flight levels as to when it is advantageous to make a step climb. It is possible to skip an interim flight level. It is also possible to make a step descent back to a lower flight level because of advantageous winds.
@Richard, Michael Glynn. Thank you.
As trials advance Richard your explanations do add to confidence in your techniques and their development while clarifying where they can be applied with confidence.
WSPR: Mid July on 2021.07.15 it will be one year since I started working on WSPR reception related to MH370!
In my humble opinion I think for just 12 months of work starting from scratch all of us as a community at mh370search.com made good progress on this sophisticated subject regarding WSPR and aircraft detection. Meanwhile Richard is the driving force on this WSPR data extraction and interpretation effort with regard to MH370!
Yesterday, 2021.07.05, 12:52 h UTC a Gulfstream jet G650ER, VH-LAL, departed from Sydney (SYD) westbound and passed over the Western Coastline of Australia at 17:12 h UTC between Margaret River and Cape Leeuwin heading over the SIO towards Saint-Denis (RUN) on the Island of Réunion arriving there at 01:30 h UTC on 2021.07.06 after 12 h 38 min flight time.
We had a SFI (Solar Flux Index) >90 (quite good for new solar cycle 25) and excellent WSPR reception between VK (Australia) with emphasis on VK6 (Western Australia) and ZS (South Africa) and vice versa as well as WSPR reception of VK6 TX signals in Réunion (WSPR RX: F61695) on 7 MHz (40 m). Other WSPR signal paths may exist but have been ignored for the moment.
Good practice round for WSPR detection trials. I also gathered live WSPR signals (decodes and non decodes) from KiwiSDR VK6QS close to Perth (PER) for app. 2 hours (Australia. Europe, Japan, South Africa, USA).
I would never have started with WSPR, if it had not been for your pioneer research.
WSPR: Flight VH-LAL just crossed the Western Australia coastline on its return flight RUN to SYD across the SIO!
@Rob, or Richard. My ‘basic user’ Flightaware access identifies the aircraft as a Gulfstream owned by Little Aviation, a Melbourne VIP charterer, though its track is undisclosed there. Flightaware notes, “This aircraft (VH-LAL) is not available for public tracking per request from the owner/operator.” I assume that would apply to its Aireon satellite tracking ID and destination data too.
You identified its registration, approach direction and destination as a flight to Sydney as it crossed the WA coast.
For the Indonesian vessel’s Cobham RSCU aircraft identification and tracking, apparently Flightaware with AMSA were the sources.
What other access did you need for VH-LAL identification etc at that juncture please?
Yes. I first tried Flightaware, Planefinder and got the same results that you mentioned.
After that I tried FR24 and it worked. Some interesting WSPR data that did not exist in 2014 such as the WSPR link from TX in India (VU), only at http://www.reversebeacon.net) at the time in 2014, and Antarctica (DP0GVN as RX) as well as VK6 (Western Australia). We had the best space weather in years on July 5th a little degraded July 6th, 2021. SFI (Solar Flux Index) between 91 and 94. New solar flares on 2021.07.03.
There is a WSPR receiver (RX) F61695 in Réunion, which normally is switched off during weekdays. This time we got WSPR data! Interesting data between VK as TX in WSPR and this receiver (RX) F61695 in Réunion.
These flights are “WSPR data milestones” in the SIO but produce tons of work to analyze the data. Live Kiwi SDR data from VK6QS close to PER have been gathered. So we have WSPR data from Réunion (FR), India (VU), South Africa (ZS), Australia (VK), Antarctica (DP0GVN) and other places. 2-way-WSPR links between ZS and VK and more.
For the AMSA data in Mid May 2021 during the rescue missions I have data from flightaware.com (flightplan etc.) as well as from FR24. For QFA114 from JNB to PER in November 2020 I also have data from both sources. Many departures and landings of aircraft from Réunion (RUN) have been detected from far away places. Initially I detected the AMSA Challenger flights on FR24 by chance looking for different things in the SIO area.
A friend of mine and I plan to work on the forward scattering issue of aircraft at HF to shine a little more light on the scattering subject that many people are more than willing to reject. We have many data now and will work on the theory as well if time permitts that time consuming endeavour. Summer time normally is prohibitive to “work on theories”!
@Rob. Your remarks re the VH-LAL source, developments and recent solar flare importance add to the picture thank you.
I was unsuccessful with FR24 but now have found those Reunion flights and, having upgraded, downloaded their details.
“A friend of mine and I plan to work on the forward scattering issue of aircraft at HF to shine a little more light on the scattering subject …”
Should you find the time, would you expand on, “Initially I detected the AMSA Challenger flights on FR24 by chance looking for different things in the SIO area.”
More specifically, what alerted you?
Also,in reversed courses (eg VH-LAL, the Cobham Challengers) from the anomalies can you distinguish approaching a receiver, about on the track, vs receding?
1. Okay on subjects VH-LAL, SFI and FR24
2. Interesting article on the subject is: M.A. Cervera et al: Climatological Model of Over-The-Horizon Radar, 2018, 14 pages, to be found at Research Gate and AGU100 containing RCS model (Radar Cross Section) of Boeing B777 at HF bands.
More to come….
3. I regularily monitor the WSPR link DP0GVN (TX) in Antarctica to ZL2005 in New Zealand (does not make too much sense in Antarctic winter) as well as
the Western coastline of Australia for westbound flights in FR24. So I detected the AMSA Challenger once Mid May 2021 and then numerous times, later found out that there were various and different Challenger aircraft. Data extraction from Flightaware and FR24 as well found press article on AMSA rescue mission on Indonesian fishing vessel days later. Search patterns were outside ADS-B range at that time.
Just yesterday I found the Bombardier Challenger 604, VH-XNC, RSCU440 again over the SIO Northwest of Perth (PER) flying meander flight patterns close to the coast line west of Joondalup. So this time it was within ADS-B range and one can see the flight pattern on FR24, 2021.07.10, duration 03 hrs 41 min. at least at 16:39 h UTC.
I do not yet know the purpose of this flight mission (Practice?). Landing in PER at 16:51 h UTC after flight duration of 03 hrs. 53 min.
4. I could if I would be able to retrieve the Doppler variation positive or negative. Normally that will not work over 2 min in real life. Having multiple hits in the time domain you can distinguish the track approaching or receding.
A very good HF WSPR receiver with GPSDO (GPS Disciplined oscillator) can do that (see also HamSCI 2021 and ionospheric Doppler shifts in the one digit Hz range, they also use it for sun eclipses and measuring height change of F2 layer).
2021.07.03 we had the strongest X-flare of the sun in 4 years and new solar cycle 25, so 2021.07.04 we had a SFI of 94 (still lower than March 2014) and luckily 3 active WSPR stations on Réunion Island (TX. FR1GZ; RX: FR1GZ and F61695)! You can etxract WSPR signals in the database http://www.wsprnet.org
I could monitor multiple take-offs and landings from and to RUN (Saint Denis) with TXs in ZS, VU, VK6, DP0GVN and vice versa which is monitoring take-offs and landings in PER and other places thousands of km away. Lots of data exist using http://www.wsprnet.org, FR24 and others.
Maybe Richard should also test GTDAAA on 2021.07.04 in the SIO and its boundaries. On 2021.07.10 WSPR links such as India (VU) TX to RX VK6 (PER) with Challenger VH-XNC, RSCU440 and others such as SIA 478 (SIN to JNB), SIA479 (JNB to SIN) and others, even from Rèunion TX, RX to far away Maui (KH6PR, AI6VN/KH6) mostly on 7 MHz (40 m).
Hope this is useful for you and others in the moment!
@Rob . Thank you for your response. I can see that much of the climatology of your reference is relevant to WSPR.
Its modelled OTHR range, while addressing that, supposes there to be ionospheric balloon use, transmission frequency control and a very large Rx antenna with polarisation control and amplitude taper.
Also, the marked effect of seasonality on detection hours-per-day of that Boeing 777 target model you mention is illustrated.
Then to add to the complexity, in the below, sporadic-E (Es) gets a mention, as a “thin, erratically occurring layer that forms in the earth’s ionosphere.”
With all that as part of/needed by OTHR, your aim to address how WSPR works looks challenging!
Thanks too for your general rundown on how you detected, and are detecting, AMSA Challengers, plus others; and how solar conditions have facilitated monitoring.
I have previously published the flight path of the blind test every two minutes from 18FEB2019 21:30 UTC until 23:16 UTC on my web site.
Here is an overview of the GDTAAA output from 18FEB2019 21:30 UTC to 19FEB2019 09:16 UTC.
I have compared the actual flight route taken with a great circle path between waypoints GEVIS and KAYTU.
I assume a more northerly flight path was taken for two reasons:
(1) The tail winds were more favourable further north.
(2) The flight route kept the option of a diversion to Perth International Airport (YPPH) open until around 19FEB2019 04:00 UTC.
I believe the flight eventually diverted to Melbourne Airport (YMML) and not to Perth International Airport (YPPH).
In the Oceanic FIRs I have only sampled the GDTAAA output at a low frequency.
Nevertheless I have found a number of position indicators that fit a consistent flight path.
Here are some examples:
I am not planning on publishing all the oceanic GDTAAA output.
Here is the arrival into Melbourne Airport (YMML) from 19FEB2019 08:40 UTC to 09:16 UTC:
Meanwhile I have received the detailed onboard computer data for the RNZAF MH370 SAR sortie by the Orion P-3C aircraft (NZ4204 ICAO:C87F0A) of 5 Squadron on 28th March 2014.
The NZDF Mission Start Date-Time included the briefing and started on 27th March 2014 at 21:43:25 UTC and the estimated time of departure from Pearce AFB was planned 28th March 2014 00:00 UTC.
The aircraft took off from Pearce AFB just before the planned time and the actual position on 28th March 2014 00:00 UTC was at 31.707754S 115.726430E overhead a point just north of Burns Beach on the coast of Western Australia 30 km north of Perth. The NZDF calculate the GS incorrectly in their .pdf attached, but this does not matter as the GS can be easily calculated from the distance travelled each two minutes.
This is good news regarding the sorties in March 2014 and your GTDAAA efforts for blind tests 2019! Maybe we can do more WSPR studies with these data, especially from 2014.
Topic 3 of yesterday: Never say never! I just stated that watching the Antarctic WSPR links does not make sense in Antarctic winter! On 2021.07.11 we had a medical(?) emergency or evacuation flight of a Hercules C130H, KRC626, NZ7004, towards McMurdo station from CHC, departing 10:30 h UTC and returning app. 02:00 h UTC at CHC on 2021.07.12. (see FR24 playback) and Flightaware.com flight protocol KRC626.
SFI lower again (app. 74); no specific data on WSPR Link DP0GVN TX to RX ZL2005SWL but 4 signals to RX VK7JJ in Tasmania and one to RX VK3DXE in MEL from TX DP0GVN (IB59ui) in Antarctica on 10 MHz (not 14 MHz as in Antarctic summer) between 22:28 h UTC and 23:08 h UTC on the return flight app. half way between Northern Antarctican shore line and Southern tip of New Zealand! These are the first and interim WSPR observations!
A blind test of GDTAAA was devised 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 has now revealed the data after I had published the GDTAAA output for the flight. Geoffrey Thomas kindly agreed to be the independent adjudicator of the WSPR Technology and GDTAAA Blind Test.
Mike also now reveals that when the aircraft departed Johannesburg Airport (FAOR) on 18th February 2019 at 21:29 UTC, it was cleared direct to oceanic airspace and flight level FL350. Mike explains that one of the engines had an oil leak which could not be repaired in Johannesburg. Originally it was planned to overfly both Perth and Adelaide, but in the cruise the oil leak worsened which necessitated a landing in Perth. Because of the likelihood of having to shut down that engine before reaching Perth, Mike elected not to climb as aircraft with asymmetric thrust are easier to control in thicker air. Therefore the aircraft remained at FL350 for the whole flight and there were no step climbs. Mike managed to land at Perth however without having to shut the engine down.
A link to Mike’s documentation of the actual route taken follows:
I have previously published the track of the aircraft from Johannesburg Airport (FAOR) between 21:30 UTC every two minutes until 23:16 UTC. After that time I only checked every 30 minutes for a position indicator and found two optional flight routes to Perth Airport and Melbourne Airport. The analysis of an arrival into Perth Airport showed only a few progress indicators, whereas the analysis of an arrival into Melbourne Airport (previously published) showed a much larger number of position and progress indicators.
This highlights the issue with GDTAAA that @David and @George have analysed in some detail, that where there are alternative paths that fit the WSPR data it is difficult to select the correct alternative. Mike Glynn also commented on my result that, “there is a lot of traffic in the Great Australian Bight as it sits in the Great Circle Route between Perth and Melbourne. Hence I’m assuming a lot of WSPR signals for you to wade through.” In hindsight I should have continued to analyse the GDTAAA output every two minutes, but this requires automation not only of the production process but also of the analysis process.
GDTAAA successfully tracked the aircraft over shorter distances but failed over longer distances.
Here is a comparison of the initial flight between predicted (grey dots) and actual (green dots). At 22:00 UTC I was 20.9 nm out, which is over my estimated precision of 18 nm. At 23:16 UTC I was over 50 nm out, although there is a position indicator within 14 nm of the actual position. I was obviously on the wrong WSPR path. Getting the ground speed right is the key and with a tolerance of 18 nm every two minutes, you have to average ground speed over at least 30 minutes. At 22:00 UTC I estimated 512 knots but it was actually on average 482 knots measured over the leg. At 23:16 UTC I estimated 521 knots but it was actually on average 516 knots measured over the leg.
Here is a comparison of the overall flight path. The great circle is marked in cyan, a typical flight route is marked with black dots, the predicted flight route is marked with blue dots and the actual flight route is marked with green dots:
Mike commented on the results as follows: “It is quite clear to me that Richard detected the aircraft north of Richards Bay and that any errors after that were probably attributable to guesses he had to make about things that he would not have had to guess about if was being advised of the probable and known aspects of the flight, such as;
1. Correct aircraft type and loading; It was a GE powered B747-400ER that had no load apart from fuel to go to SYD and a standard crew. Therefore the most likely cruise Altitude was initially FL350. The fact that we stayed at that altitude later on in the flight for aerodynamic reasons would not be known by an investigator and hence would not be revealed to Richard.
2. In any case, in real life there was a radar/ADSB ground speed and a readout of the altitude at the top of climb. This would have given him the initial cruise speed and the knowledge that the aircraft was flying at a higher than normal speed. If the feeling is that these parameters would have been available in real life, they should be made available to the investigator.
3. That the aircraft was tracking via an optimised route and not by published Air-routes.
4. That the SIO is pretty empty of other air-traffic but Southern Australia is not.
I got quite excited by the finding of the initial tracking north of Richards Bay. If this was a real life investigation and I was appointed an expert adviser on the 747 for this exercise then I would have told Richard all of the above to avoid him having to make educated guesses; which apparently may have led him astray after that.”
@George pointed out in a previous comment “you seem overly keen for Mike Glynn, the pilot, to get onto an agreed flight route”. @George was right and it turns out that the aircraft was given permission to fly direct to oceanic airspace. I failed to consider this possibility.
For me the biggest lesson learnt is the need to fully automate the whole process. There were around 1,500 WSPR links every two minutes during this blind test. There are around 2 x 1,500 x 1,500 = 4.5M possible intersections of the WSPR links every two minutes. Any two great circles will intersect twice on opposite sides of the globe. Out of these 4.5M intersections there will be several close to the predicted position of the aircraft. The aircraft might be turning and there may be several alternative intersections close to or within reach of the predicted position of the aircraft.
With the current partial automation, it takes 30 minutes to produce and analyse a single point in time. Production has been automated and takes 2 minutes per point. Analysis needs to be automated as well. A nine hour flight (Johannesburg-Perth) results in 270 GDTAAA outputs and takes currently 135 hours of production and analysis. When there are alternative routes to be checked, then this time increases significantly.
I am writing some new software to automatically find all points of intersection of anomalous WSPR links within a particular area. The software will also count the number of WSPR links intersecting at a particular point. An aircraft will follow a track at a given altitude above the earth’s surface. I use the WGS84 definition of the earth’s surface which is defined as an oblate spheroid. Radio waves follow a line of sight propagation path in free space. Radio waves can be reflected by the ionosphere, refracted by the troposphere, diffracted and scattered by objects and surfaces along the propagation path. When anomalies occur in the radio wave propagation, I assume the major part of the remaining signal after any reflection, refraction, diffraction and scattering continues on a perfectly spherical great circle path around the earth.
When WSPR anomalous links intersect at the predicted location of an aircraft, then it is possible that the metallic structure of the aircraft, its engine exhaust and/or its wake turbulence affected the signal propagation. The WSPR anomalous intersection may not be exactly where the aircraft is. The wake of the aircraft might have caused the WSPR anomaly. Points with 3 or more intersections are much safer in predicting the presence of an aircraft. Nevertheless, any particular point can be up to 18 nm in error. Only the average taken over a number of points and consistent with the aircraft performance is valid.
It was a mistake to believe that I did not need to track the aircraft in the blind test every two minutes. I was not a million miles out, but I discounted the route to Perth in favour of an alternative route to Melbourne. As Mike points out, the Southern Indian Ocean is relatively empty of other aircraft so checking for other flights is easy. For flights between Perth, Adelaide, Melbourne or Sydney there are a lot of other flights and I failed to check whether I had picked up another arrival into Melbourne.
A fascinating exercise and many thanks to Mike and Geoffrey for their help.
@Richard. While the trial has not verified GDTAAA outcome reliability, in your objective analysis you do describe useful lessons gained.
However the separation of the tracks quite early and the lack of any alert that subsequent position markers were false, do raise some doubts as to what level of effort will be needed to find a fix.
The major and apparently enduring underlying problem seems to be the 18 nm tolerance.
Automation analysis would allow 2 minute markers on long trips like this and more intensive assessment of alternatives but with that 18 nm tolerance requiring half an hour to establish ground speeds (and to establish confidence in track changes?), the accelerating number of alternatives the meanwhile, natural random variation of transmission density, the ambiguity of top hats (and even 3 anomalies meeting at a point?), reduction of anomalies in straight and level flights and added uncertainties during climbing and descending turns – with all that and OTHR in the background I now wonder whether reliable tracking will need extensive modelling and computer support.
But even if not, some time?
Concentration on more bounded and straight and level flights in areas of low traffic density could be seen as an MH370 priority though I am unsure that such simplifications will overcome some of those above issues, or would help with confidence.
Maybe too gloomy but then again maybe realistic?
The 18 nm precision is a major problem in my view, even when the GDTAAA analysis is also automated.
I am currently discussing with academic physicists the possibility of detecting the differences between the influence of aircraft metal, exhaust and wake on HF radio signals. If I could tell the difference between the three alternatives of WSPR link disruption, then I could narrow the 18 nm considerably.
In 2021 you have around 2,000 WSPR links every two minutes, 1,500 in 2019 during the blind test, but only 300 in 2014 for MH370 or MH371. The alternative paths in 2014 are significantly reduced.
The MH370 WSPR analysis in combination with the hourly check point from the Inmarsat satellite data is certainly more promising.
I have not yet I started with experiments changing the SNR or drift tolerances.
“I am currently discussing with academic physicists…..”
“The MH370 WSPR analysis in combination with the hourly check point from the Inmarsat satellite data is certainly more promising.”
“I have not yet I started with experiments changing the SNR or drift tolerances.”
Although the Blind Test was mostly a failure, a number of useful lessons have been learnt.
WSPR links vary from direct line of sight propagation over short distances to indirect sky wave ionospheric propagation over long distances to the other side of the globe.
The first error discovered in GDTAAA from the blind test results is the difference in the great circle calculations between an aircraft’s flight path which follows a WGS84 approximation of the earth as an oblate spheroid and the long distance sky wave propagation of radio waves which follow a perfect sphere in free space.
Radio waves such as WSPR links are subject to four categories of anomaly due to reflection, refraction, diffraction and scattering. Reflection is an abrupt change of direction of a wave that strikes the boundary between different mediums. Refraction is a deflection or change of direction in passing obliquely through the interface between one medium another or through a medium of varying density. Diffraction is a bending of part of the wave energy from the normal line of sight path. Scattering is removal of energy, change of direction, phase or wavelength.
Over the horizon radio sky wave propagation is predominantly at high frequency (HF) between 3 MHz and 30 MHz, but WSPR links have been successfully transmitted over shorter distances at frequencies between 0.1 MHz and 1296 MHz. Sky wave propagation is based on HF radio wave refraction by the ionosphere. The ionosphere consists of particles ionised mainly by ultraviolet radiation from the sun in various layers in total around 300 km thick and can be as low as around 50 km or as high as around 1,000 km above the surface of the earth.
The most important layer of the ionosphere for long distance radio wave propagation is the F2 layer because it is present around the clock, has the highest altitude and highest ionisation. The typical height above the earth’s surface is between 250 km and 275 km, but this can vary. Below is a link to a recent example of a global map of the F2 Layer maximum height in km:
Radio waves appear to be reflected by the ionosphere, but are in fact refracted. The distance achieved in a single skip or hop is around 4,000 km and the maximum single hop is around 4,800 km. There is a maximum usable frequency (MUF) and a lowest usable frequency (LUF) for any sky wave transmission. The critical frequency (fc) is the limiting frequency at which radio waves are no longer reflected back by the ionosphere but pass through into space. MUF = fc / cos ϴ, where ϴ is the incidence angle. The critical angle at a given frequency is the limiting incidence angle from the transmission antenna to the ionosphere at which a signal will be lost in space as shown in the following diagram:
The radio waves reflection off the ionosphere is similar to that off the earth’s surface and multi hop propagation paths are possible with overall distances of around 8,000 km, 12,000 km, 16,000 km and even 20,000 km. There are propagation losses through the atmosphere (mainly in the troposphere), absorption losses in the ionosphere and absorption losses on the earth’s surface. Due to these losses the theoretical maximum number of hops is five and this figure is only achievable if interim reflections are mostly on calm ocean (wave height ≤ 2 m).
The absorption loss at the earth’s surface depends on the whether the interim reflection point was on the ocean (calm or turbulent) or the land (smooth or rugged). The oceans represent 71% of the earth’s surface. The following link shows an idealised diagram of a theoretical model of multi hop WSPR propagation:
The presence or absence of an aircraft makes a difference of 3.4 dB in the received signal in this example in the theoretical model. The value from the empirical model for the same conditions was 3.355 dB.
A Boeing 777-200 ER presents a reflective and scattering surface area of the metal structure of 834 m2 . The wake behind the aircraft is up to 460 m either side of the aircraft and up to 20 nm behind the aircraft and presents a reflective and refractive surface area of up to 17 km2 . The engine exhaust is up to 230 m either side of the aircraft and up to 10 nm behind the aircraft and presents a reflective and refractive surface area of up to 4.25 km2 . An aircraft has a typical cruise altitude between 25,000 feet (7.6 km) and 41,000 feet (12.5 km) of say 10 km. At 500 knots an aircraft is moving in the horizontal plane at 257 m/sec and the wake is also descending behind the aircraft at 2.5 m/sec. The engine exhaust has a flow rate of 650 m/sec.
The wake is water vapour, whereas the engine exhaust is CO2, SO, NO, unburnt fuel, soot, metal particles and water vapour. The wake, exhaust and the aircraft will create WSPR SNR anomalies by reflection, refraction and scattering. The aircraft wake is 20,000 times the size of the aircraft structure and the aircraft exhaust 5,000 times the size of the aircraft structure. Drift anomalies will be predominantly from scattering and predominantly from the aircraft structure.
GDTAAA will be modified to include the number of hops between Tx and Rx locations, the category of earth surface at the interim reflection points (calm ocean, turbulent ocean, smooth land or rugged land) and the proximity of the predicted aircraft position to the interim reflection point.
I am most grateful to Lingxiao Liu, Tian Lu, Mingxue Gong and Wuyu Zhang for their paper titled “Study on the strength loss of Multi-hop HF Radio Propagation”:
You write: “Although the Blind Test was mostly a failure, a number of useful lessons have been learnt.”
Your blind following of a ghost aircraft was FAR from a failure.
It clearly showed more pitfalls that might become one who tried to blindly follow such a ghost.
I seriously thought you would have become “becalmed” and go back to the start and do a rethink and a re-start. When you “arrived” in Melbourne, it was a surprise.
The lessons are many,
BUT, the actual flight taken by Mike Glynn and his Boeing from Johannesburg to Perth will serve admirably as an example across the Southern Indian Ocean for you and your to-be-revised-and-improved GDTAAA to “cut your teeth” on.
As radio signals across the ocean necessarily will incur some variations of “indirect sky wave ionospheric propagation” the actual flight will serve admirably for a real flight to trial the improved GDTAAA. NOT only that, it will be effectively in a world five years after March 2014, with an increased WSPR activity to work with in comparison.
There should be much more chance that the Johannesburg to Perth flight interrupted any WSPR signals (or “links” between Transmitter and Receiver) during the flight. (Simply because of the presumed increased WSPR activity as well as the longer flight duration over the ocean)
Any chance that the final flight of 9M-MRO on Flight MH370 disturbed any WSPR signals might then be more reasonably estimated (and searched for).
The blind test was a failure in that I discarded the flight path to Perth in favour of a flight path to Melbourne with more position indicators. I failed to recognise that I had picked up another aircraft.
Automation of the analysis will allow me in future to take on long flights and complete the tracking at two minute intervals in a more timely manner.
The blind test was indeed a most useful learning curve and GDTAAA will see significant improvements as a result.
The ability to match a theoretical model of GDTAAA with an empirical model has also helped enormously.
I am convinced that WSPR can add to the flight analysis of MH370 based on the Inmarsat satellite data, 9M-MRO engineering and fuel data, Boeing 777-200ER performance data, MH370 floating debris material and drift analysis.
I plan to use the blind test actual track as a 2019 test bed and the AMSA SAR RNZAF actual track as a 2014 test bed. Mike Glynn has already started to prepare another blind test for some point in the future. Meanwhile there is a lot of software development and testing to do.
I have examined some initial points along the track of the blind test flight from Johannesburg in light of the lessons learned from this most useful exercise.
The first set of WSPR links are 1 minute after departure at 21:30 UTC. The original GDTAAA output had an error. I used the WGS84 oblate spheroid instead of a perfect sphere to map WSPR links. The original GDTAAA output at 21:30 UTC looked like this:
Analysing the original WSPR intersection close to the departure point looked like this:
Correcting this WSPR intersection moves the point 13.3 nm further north and is obviously not on the track of an aircraft departing to the south:
The global view of the corrected WSPR intersection reveals that this intersection did not happen, because 6 hops would have been required for one of the intersecting WSPR links and 7 hops for the other intersecting WSPR link. These are both beyond the maximum of 5 hops predicted by the theoretical model. The two intersecting links are CM98edFM19qh at 40 dBm at 14.097092 MHz and JO59csJN80nu at 30 dBm and only 0.475707 MHz. The frequency of the second intersecting link is too low to achieve such a propagation distance.
The second set of WSPR links are at 21:32 UTC and was originally deemed a position indicator. The original GDTAAA output looked like this:
The two intersecting links are DN70icFM18is at 37 dBm on the 14 MHz band and JO59csJP33wi at 30 dBm and only 0.475707 MHz. Again the frequency of the second intersecting link is too low to achieve such a propagation distance.
Correcting this WSPR intersection moves the point 19.6 nm further north west and is obviously not on the track of an aircraft departing to the south:
The global view of the corrected WSPR intersection reveals that this intersection did not happen, because 6 hops would have been required for one of the intersecting WSPR links and a distance of 9,684.2 km is not realistic for the other intersecting WSPR link at a frequency of 0.475707 MHz.
The next position indicator in the original GDTAAA output was at 22:00 UTC and looked liked this:
The two intersecting links are JM75fwJN48fw at 30 dBm on the 7 MHz band and FM08ppDN07ic at 27 dBm on the 14 MHz band:
Correcting this WSPR intersection moves the point 16.0 nm further south and away from the actual track of the aircraft. However another link FM08ppCM97gs at 27 dBm on the 14 MHz band comes closer and intersects with the first link JM75fwJN48fw at 30 dBm on the 7 MHz band only 5.2 nm away from the actual position of the aircraft:
The global view of the corrected WSPR intersection reveals that this intersection possibly did happen, because 2 hops would have been required for one of the intersecting WSPR links over a distance of 7,391.5 km with one interim landing on smooth ground of a river plain in northern Democratic Republic of the Congo and 4 hops would have been required for the second link over a distance of 13,646 km with interim landings all on the ocean surface:
Out of the lessons learned from the blind test I have decided in future to only use WSPR links in the 3 MHz to 30 MHz range and only above a transmission power of 10 dBm (where distances between 8,000 km and 20,500 km are feasible). I will discard any theoretical intersection points over 5 hops or 20,500 km from the transmitting station.
The WSPRnet transmissions over distances exceeding 19,000 km are usually at 37 dBm on the 7 MHz band and between countries like Spain and New Zealand or South Africa and Hawaii where multi hop propagation is with all interim reflections or landings on the ocean surface. It is interesting to note that WSPRnet links between typical DX stations in Spain and New Zealand take the long path at slightly more than 20,000 km with interim landings once in the Atlantic and twice in the Pacific, as opposed to the short path at slightly less than 20,000 km with interim landings in a lake in western Uzbekistan, the rugged highlands of north eastern Laos and in shallow waters just off the coast of Northern Australia.
On 20th July at 16.03 you wrote” “I discarded the flight path to Perth in favour of a flight path to Melbourne with more position indicators. I failed to recognise that I had picked up another aircraft.”
Your observation about the ‘failure’ of the blind test might help to remedy the global failure, to date, to clarify what happened to the passengers and crew aboard MH370.
Similar errors, mis-identifications, presumptions and calculations may have occurred in earlier parts of the search for the plane, or even within the military and civiian flight control and surveillance systems that were in operation in early March of 2014, locally, regionally and globally.
Welcome to the blog and many thanks for your thoughtful comment.
We have not found MH370 after 7 years and several underwater searches covering a total area of 278,000 km2. There are as many alternative solutions as there are analysts.
MH370 is either not in the area searched and we have been looking in the wrong place or is in the area search and we have missed the debris in difficult sea floor terrain, equipment issues, data processing or interpretation errors.
The Inmarsat satellite data, Boeing 777-200ER performance data, 9M-MRO engineering data, fuel range and endurance, floating debris drift analysis, hydroacoustic, seismic and chemical data analysis will not give you a precise crash location. Satellite imagery and over the horizon radar have not provided a definitive answer so far.
I am hoping to refine the WSPR data analysis to the point where some over the horizon radio reception anomalies taken together with the other data we have will help us to narrow down to a more precise crash location.
With regard to: “I am hoping to refine the WSPR data analysis to the point where some over the horizon radio reception anomalies taken together with the other data we have will help us to narrow down to a more precise crash location.”
Would there be any archived geomagnetic data that could also be used to add another layer to make the plane’s final airborne location even more precise, if more precisoin is required?
This is based on a guess that, as well as the three forms of radio signal interference you described, a fast moving big plane in a remote location might also have interfered very subtlly with the earth’s magnetic field in ways that may have been recorded.
As a start point I was wondering about the Brirish Geological Survey’s “World Data Centre for Geomagnetism”, based in Edinburgh.
Apologies for mistyping ‘precision’ as ‘precisoin’
I have documented all the results and lessons learned from the blind test of GDTAAA:
I have implemented the changes described in the GDTAAA blind test results and lessons learned document.
I have started to re-run the GDTAAA output for the blind test. Now there are two views for each two minute interval. Firstly the local view which shows the predicted flight path in dark grey and the actual flight path in light grey. Secondly the global view which shows the WSPR links from the Tx station (marked with a black dot) to the Rx station (marked with a magenta dot) both in short path (SP) and long path (LP) as well as the point of intersection with the predicted aircraft position. Each WSPR link used to detect and track the aircraft is defined by a 12 character Maidenhead code which is a concatenation of the 6 character Tx station Maidenhead code followed by the 6 character Rx station Maidenhead code. You will find the WSPR links in the WSPRnet database for 18th February 2019 at the appropriate time.
At 21:30 UTC there is one WSPR drift anomaly CM98edFM19qh marked in green:
At 21:32 UTC there is one WSPR SNR and drift anomaly DN70icCM97gs marked in blue:
At 21:34 UTC there is one WSPR SNR anomaly again at DN70icCM97gs marked in red:
At 21:36 UTC there is one WSPR drift anomaly at EN90fxCM98hq marked in green:
At 21:38 UTC there is one WSPR SNR anomaly again at DN70icCM97gs marked in red:
At 21:40 UTC there is no WSPR anomaly:
At 21:42 UTC there are three WSPR SNR anomalies again at DN70icCM97gs as well as at EM12kpDM13if and at JM75fwJN47wk which are all marked in red:
This is the first position indicator and allows an update of the average ground speed since 21:30 UTC to 473.24 knots. The instantaneous ground speed is 514.32 knots.
I previously reported the first position indicator at 21:42 UTC with three WSPR SNR anomalies at DN70icCM97gs, EM12kpDM13if and JM75fwJN47wk which are all marked in red:
I subsequently checked to ensure the combination of these three WSPR SNR anomalies was unique during the time frame of the blind test flight which departed Johannesburg at 21.29 UTC and arrived in Perth at 05:54 UTC. From the attached Excel spreadsheet of all WSPR links from these three combinations you can see that this position indicator at 21:42 UTC is unique from before the flight commenced at 21:20 UTC until after the flight arrived at 06:00 UTC:
At 21:44 UTC there is one WSPR SNR anomaly again at DN70icCM97gs marked in red:
The reason the DN70icCM97gs WSPR SNR anomaly reoccurs is that the final bearing from the Tx station in Longmont, Colorado, USA at 40.0722 °N 105.3539 °W is 121.11 °T and aligns with the track of the aircraft. The final bearing of the flight path from the position at 21:32 UTC at 26.2989 °S 28.4822 °E to the position at 21:44 UTC at 27.1175 °S 29.9962 °E is 120.71 °T.
At 21:46 UTC the aircraft reaches the top of the climb (TOC) and levels out. A new type of indicator occurs, which I have called a multiple indicator. There is a WSPR SNR anomaly between Tx station W0ENO in Longmont, Colorado and Rx station N6KOG in Tracy, California; there is a WSPR drift anomaly between Tx station W0ENO in Longmont, Colorado and Rx station KN4GDX in Warrenton, Virginia; and there is a dual WSPR SNR and drift anomaly between Tx station K2RAS in Longmont, Colorado and Rx station N6KOG in Tracy, California all at the same time.
The WSPR anomalies all come from the same Tx station location and overlap on the next two GDTAAA outputs. You can see the separation on the third GDTAAA output which is at a higher resolution:
When you zoom in on the WSPR links, the SNR and dual anomaly pick up the aircraft’s exact position and the drift anomaly picks up the wake of the aircraft about 2 nm from the centre line:
I again checked to ensure the combination of these three WSPR SNR, drift and dual anomalies was unique during the time frame of the blind test flight. From the attached Excel spreadsheet of all WSPR links from these three combinations you can see that this multiple indicator at 21:46 UTC is unique from before the flight commenced at 21:20 UTC until after the flight arrived at 06:00 UTC:
Call sign W0ENO is Longmont Amateur Radio Club (LARC) which has a transmitter on the top of the Longmont Justice Center at 1,514 m above sea level as well as a microwave link to a transmitter on Lee Hill 13 nm outside Longmont in the foot hills of the Rocky Mountains at 2,412 m above sea level. Call sign K2RAS is Ronald Schwartz, who is a member of LARC. The club’s facility on Lee Hill is pictured in the following link:
At 21:48 UTC there are no WSPR anomalies, but DN70icFM18cs which was an anomaly at 21:46 UTC is still a borderline SNR anomaly. DN70icFM18cs was a drift anomaly at 21:46 UTC but not a SNR anomaly at only a 0.65 standard deviation. DN70icFM18cs was no longer a drift anomaly at 21:48 UTC and also not a SNR anomaly at a 0.80 standard deviation. There are also two other WSPR links which are determined to be stable (no drift anomaly and no SNR anomaly outside one standard deviation) that intersect at the predicted aircraft position:
It is possible that I have set the criteria for an anomaly too high, but I do not plan to change the criteria since I want to make sure that I only pick up obvious anomalies.
If I were to accept the SNR anomaly at 0.80 standard deviation at 21:48 UTC the global view would be as follows:
At 21:50 UTC there are no WSPR anomalies, but FM08ppCM87vl is a new drift anomaly about 2 nm from the centre line of the predicted path of the aircraft:
The global view at 21:50 UTC follows a similar pattern as the previous anomalous WSPR links:
At 21:52 UTC there is a WSPR drift anomaly at DN70icFM18cs once again about 1 nm from the centre line of the predicted path of the aircraft. There was a drift anomaly at 21:46 UTC from the same WSPR link and the SNR level is hovering on the borderline at 21:46 UTC, 21:48 UTC and 21:50 UTC:
It appears that WSPR anomalies can pick up not only the aircraft at the predicted position but also the wake of the aircraft just behind and to either side of the centre line of the predicted track.
@Richard. Summary of the testing situation from Geoffrey Thomas:
You raise an interesting idea that a fast moving big plane in a remote location might also have interfered very subtly with the earth’s magnetic field in ways that may have been recorded.
I downloaded the 1-minute data recorded at 5 stations in the Indian Ocean region during the flight of MH370 from 07MAR2014 16:42 UTC to 08MAR2014 00:30 UTC.
Station TUN in Sumatra recorded a maximum total intensity of 51,839 nano Tesla for 11 minutes between 18:33 UTC and 18:44 UTC.
Station GAN in the Maldives recorded a minimum total intensity of 39,941 nano Tesla for 2 minutes between 19:05 UTC and 19:07 UTC.
Station CKI in the Cocos Keeling Islands recorded a minimum total intensity of 47,653 nano Tesla for 2 minutes between 20:52 UTC and 20:53 UTC and again between 20:55 UTC and 20:56 UTC.
Station LRM in Learmonth recorded a minimum total intensity of 52,892 nano Tesla for 4 minutes between 20:52 UTC and 20:56 UTC.
Station GNG in Gingin near Perth recorded a minimum total intensity of 57,889 nano Tesla for 15 minutes between 20:47 UTC and 21:02 UTC.
I have plotted these observations of the short time frame disturbances in the total intensity of the Earth’s magnetic field against the predicted position of MH370 at the time of the disturbances:
I have not yet resolved the horizontal and vertical components of the disturbances and your idea bears further analysis, in my view. It would also be useful to include the observations from the French station at Martin de Vivies on Ile Amsterdam for 07-08MAR2014 but I could not find a website where I could download 1-minute data for the MH370 timeframe.
The geomagnetic ‘big jet footprint’ suggestion, if there is value in the idea, may be able to be used to further refine, or to corroborate the WSPR and other projections.
With regard to the one minute data from the French station at Martin de Vivies on Ile Amsterdam that you mentioned, and given that this is a global effort to resolve an unexplained civilian airline incident, it may be that IPGP, which is a member of ‘Intermagnet’, can make the raw, unfiltered data available for the required time and date, if it is still accessible.
IPGP is The Institut de Physique du Globe de Paris – Université de Paris – a French governmental, non-profit research and higher education establishment located in Paris, dedicated to the study of earth and planetary sciences by combining observations, laboratory analysis and construction of conceptual analogical and numerical models.
INTERMAGNET has its roots in discussions held at the Workshop on Magnetic Observatory Instruments in Ottawa, Canada, in August 1986 and at the Nordic Comparison Meeting in Chambon La Foret, France, in May 1987.
A pilot scheme between USGS and BGS was described in the sessions of Division V of the International Association of Geomagnetism and Aeronomy at the 19th General Assembly of the International Union of Geodesy and Geophysics in Vancouver, Canada, in August 1987.
This scheme used the GOES East satellite to successfully transfer geomagnetic data between the two organisations.
INTERMAGNET was founded soon after in order to extend the network of observatories communicating in this way.
62 different institutes are now members of the INTERMAGNET consortium, and, since 1991, data have been contributed to INTERMAGNET from approximately 150 observatories.
INTERMAGNET is a member of the World Data System of the International Science Council, and it is closely associated with the International Association of Geomagnetism and Aeronomy.
One-minute resolution data time series are available from all IMOs: these are described as “definitive data”, as they are not subject to future reprocessing or re-calibration and therefore represent INTERMAGNET’s “gold-standard” data product for scientific and other uses.
Definitive data are therefore considered an accurate representation of the vector geomagnetic field and its time dependence at the location of each IMO.
Reported or raw, unprocessed data are reported promptly from each observatory (for some stations, within an hour of acquisition).
The one-minute resolution data are time-stamped to the start of each minute and are derived from faster sampled data according to digital filters that accord with the technical standards for one-minute data.
(the above information is from the Wikipedia entry for Intermagnet on 5th August 2021 at 14.50 UTC)
Flight QF63 from SYD to JNB over SIO
SFI just 75
Another analysis layer that could help to validate and enhance the accuracy of the WSPR / GDTAAA mechanism would be to include blended infrared / meteorological data.
This extract from a 1986 article by Stehen Corfidi and Henry Brandli for the US National Weather Agency describes how satellite imagery and weather data can be used to map contrails and distrails.
The quality of satellite imagery and weather data, plus the capacity to analyse it, has, fortunately, increased quite a lot since 1986.
“The heat of combustion of an aircraft or spacecraft’s fuel can under certain conditions evaporate existing clouds (if not too dense) and yield a distrail or dissipation trail.
Narrow distrails, the opposite of contrails, are often viewed by ground based observers.
Contrails, exhaust trails, ship trails, etc. have been seen on weather satellite imagery most vividly in clear skies.
Distrails, on the other hand, have probably been seen on spacecraft imagery which have not been published.”