Aircraft Scatter – What works? And what not?

[This article appeared as a cover story in the June 2022 issue of the specialist journal “Funktelegramm”. Publication here with kind permission of the publisher and editor-in-chief, Joachim Kraft, DL8HCZ].

That’s where the action is: Flight CAT405 flies between Hanover VOLMET and DK8OK and causes a characteristic Doppler trace in the VHF range – see mouse arrow. Receiver Winradio ‘Excalibur Sigma’, Antenna Dressler ARA 2000 [original], Software SDRC.

Again Godfrey, Dr. Westphal et al. have dealt with aircraft scatter – see here and here. And again they are catching fools with unscientific theses. Apart from the DARC, which enthusiastically welcomes these charlatanries and gives them – against all reason and physics – a lot of publicity, there is probably nobody who believes this blooming nonsense of people “who don’t know what they are doing” (physics Nobel laureate Prof. Joe Taylor, K1JT, on exactly this topic).

From different sides I was now asked to explain as simply as possible and directly comprehensible for everyone, what it is about this not so mysterious “airplane scatter”. And why the “thesis”, that one can trace aircraft movements over thousands of kilometers by WSPR log data, belongs to those stupidities, which are only believed by the “friends of the flat earth”.

Let’s start simple:

  • What happens when an airplane and a radio wave meet?

You can see this in the screenshot above. In the centre, the carrier of the aeronautical meteorological transmitter “Hanover VOLMET”, which transmits in AM with 50 watts (e-mail Deutsche Flugsicherung from 10.5.2021), runs in time from bottom to top. The window is only 400 Hz wide, so that one can just see the modulation sidebands on the left and right because of the 200 Hz audio highpass at the transmitter.

Clearly aircraft scatter: Doppler traces

But in the immediate vicinity of the carrier, there is all the more activity. You can see several Doppler traces of airplanes to the left and right of the carrier. They are caused by the radio waves of the transmitter being scattered by the metal hull of the aircraft and being received in addition to the original transmitter signal..

Since the aircraft is moving, the Doppler effect familiar from passing vehicles with a horn also occurs here: the Doppler frequency of the approaching object is initially above the frequency heard by vehicle drivers. As it approaches the pedestrian, however, he hears a tone that is first high-pitched and then falls until the vehicle is exactly at the same height with him. Then the observer and vehicle passengers hear the same pitch. It then drops again for the observer as the vehicle moves away. The so-called “Doppler frequency”, i.e. its shift with respect to the actually radiated frequency, depends on the (relative) speed at which the vehicle is moving.

Yes, and that’s exactly what happened when the Boeing 737-430 flew into the path of the radio waves between Hannover VOLMET and my location on its way between Copenhagen and Geneva. FlightRadar24 shows this nicely, see below.

Flight CAT405 crosses the route between Hannover VOLMET and DK8OK at 09:08:48 UTC on March 12, 2022.

If we compare the trajectory with the Doppler trace, we can trace the following:

The Boeing flies from north to south. So it enters with a high Doppler frequency, which then drops to exactly zero (“Zero Doppler”) when it crosses the direct line between transmitter and receiver. At this moment, it disappears in the carrier and comes out again in the lower sideband as it flies further south – the Doppler frequency drops because the Boeing is moving away.

For further consideration, our questions include the following:

  • What is the power backscattered from the aircraft?
  • By what amount does it increase the level of the carrier signal when it crosses the line between the transmitter and receiver?

First, let’s look at the situation in 3D with the transmitter, receiver and aircraft at 10 km altitude just above the “crossing point” in Matlab:

A transmitter (left), a receiver (right) a scatterer (top).

The path Hannover VOLMET -> DK8OK is the normal case. But a part of the energy of the transmitter is also “heard” by the metal hull of the plane and reflected back, or scattered.

The signal strength received directly at DK8OK and by the aircraft can be easily calculated. However, how much energy is scattered back from the aircraft depends on so many factors that one must resort to either measurements or modeling. The result is a diagram, quite similar to an antenna diagram. And a number, the so-called “backscatter area” or RCS (radar cross section).

Actually, this is not a single number, but a range depending on frequency, aircraft type, and relative direction to transmitter and receiver. For commercial and transport aircraft, the literature gives RCS values between 10 to 100 dBm2. The theory should not bother you here. Just this: the higher this number, the better electromagnetic waves are scattered back. Military stealth aircraft are mostly below RCS 10.

Stiff Kint (private communication: “Bistatic Radar Cross-Sections and the Detection WSPR Signals By Aircraft Scatter; March, 2022”) took the trouble to simulate the RCS for frequencies between 1 MHz and 120 MHz with the software NEC2C using a wire model of a Boeing 777-200ER. On none of the frequencies the RCS exceeds a value of 50. Mostly it runs between 20 and 40. For safety I nevertheless assumed a value of 50, performed the calculations and get the following picture:

The level ratios – see text.

Accordingly, the aircraft reaches an energy of -45 dBm2. This is converted into a signal of -28 dBm, which is received at a distance of 13 km with -96 dBm. It is about 36 dB below the original carrier. Does all this halfway agree with the actual observed levels? Despite some rounding and assumptions?

Theory – quite practical

By the Doppler trace we are here in the fortunate position to measure scattered and directly received signal in each case clearly and separately. A conversion of the above spectrogram into a number matrix facilitates these measurements, but also the determination of maximum and average values:

Here is the implementation of the spectrogram in Matlab. If we look closely, we also notice the pearl string effect of the Doppler signal, which is caused by phase changes.

Measurements then show that we are at about -65 dBm for direct reception in the peaks of the carrier, but the Doppler trace actually has a maximum value of -90 dBm. This answers the above question: Yes, the measurements show that the assumptions and calculation are basically correct.

“Level jump” due to addition of carrier and Doppler trace?

That leaves the next exciting question:

  • How large is the level jump of the carrier signal observed on its own at or near the “passage”?

To conclude by such “level jumps” rock-solidly on airplanes and even helicopters in thousands of kilometers distance is, after all, one of the pivotal points of the publications of Godfrey, Dr. Westphal et al.! [In his recent paper, Godfrey & Dr. Cotzee seem to detect even cars …]

Since we have measured scatter and original signal separately here, we only need to sum them uo and take the difference to the pure carrier.

So: -65 dBm (carrier) plus -90 dBm (scatter) and, after de-logarithmizing for power addition logarithmed again in dBm to measure the level increase of the carrier, we get -64.9863 dBm or:

an increase of 0.0137 dBm!

Gorgeous, as measurement uncertainties etc. are at least two orders of magnitude larger!

The fairy tale of the “trip wire

Godfrey, Dr. Westphal et al. now think that a shortwave signal forms a “tripwire” between transmitter and receiver: if an aircraft touches this virtual wire, the alarm sounds. And this shows up in an “anomalous” increase of the received signal level. Well.

Indeed, some kind of this concept does exist. But mainly in the range (far) above about 20 MHz. Radio amateurs use it blithely every day, supported by smart software like this one and this one. That it can be transferred to shortwave only in exceptional cases, I will be shown below. But first the concept “in action”:

It describes a connection between transmitter and receiver, which are outside their radio horizon. Thus the receiver can hear the transmitter either not at all or only with the smallest signal. But if an airplane (cruising altitude 10 km), the plasma tail of meteorites (approx. 80 – 130 km altitude), the space station ISS (400 km) or even the moon (examples here) can “see” the transmitter and receiver at the same time, then the scattered signal can be detected at the receiver.

Let’s take the French radar transmitter GRAVES as an example; Rob Hardenberg, PE1ITR, has produced a knowledgeable paper on it: “The 143.050MHz GRAVES Radar a VHF Beacon”. The transmitter on 143.050 MHz scans the sky for space debris with a high rms power using sector controlled antennas.

As the crow flies, the distance between GRAVES and DK8OK is 654 km, but the so-called “radio horizon” of the transmitter barely reaches 30 km. However, a meteor tail flaring up at an altitude of 80 km can see almost 1,200 km wide and thus has quite different dimensions in view: Casually, it can connect GRAVES and DK8OK via scatter, if it falls down at a suitable location. Unlike the above example with Hannover VOLMET, the carrier of the transmitter is then of course no longer visible. The Doppler signals on the other hand can be quite strong, see screenshot below from March 12, 2022 around 21:16 UTC. Mostly this is also the effect which connects two amateur radio stations outside their radio horizon via aircraft or meteor scatter.

Scatter at a meteor tail: Even if the carrier of GRAVES cannot be “seen” at a distance of 620 km, the meteor tail “sees” the transmitter and receiver at an altitude of about 80 km. Above the spectrogram of the short time flare, below the spectrum with the typical pearl string effect.

Of course, this is not necessarily a “trip wire” either, because it does not yet allow the position of the meteor tail to be determined.

However, to apply this concept generally to shortwave is only possible by completely ignoring the propagation mechanisms there: At least below about 20 MHz, a small carrier residue can still be found with sensitive methods even in the so-called “dead zone”. It propagates via “backscatter” caused by irregularities of the ionosphere. This effect can be simulated with the PropLab Pro propagation software also used by Godfrey, Dr. Westphal et al. – just as in 3D ray tracing the splitting of a single signal into several components that propagate along different paths. This in turn leads to Doppler traces due to the dynamics of the ionosphere as well as great circle deviations. Godfrey, Dr. Westphal et al. give these mundane matters a wide berth, however. At least here they will know exactly why.

The application of the concept of VHF propagation to shortwave propagation with its strongly fluctuating signals, which are also only collected every 110 seconds via WSPR log data, proves to be even more humbug. But let’s do that again: “Never Give a Sucker an Even Break” (W.C. Fields, 1941). So let’s take another look at least briefly at “Airplane Scatter on Shortwave”. There is, sure. But not as an Alfred E. Neuman [MAD] dreams of.

We take the AM carrier of the 500 kW transmitter of the Saudi Arabian Broadcasting Authority on 15,380 kHz, radiated from strong (+20 dBi) focusing curtain antennas (310°, Middle East direction) just outside the capital Riyadh, distance 4,360 km. The figure below shows in the spectrogram the carrier and a Doppler trace caused by aircraft scatter. Furthermore one sees the splitting of the carrier by magneto-ionic effects of the ionosphere, the thereby in immediate carrier proximity caused Doppler tracks plus meteorite scatter.

Two ionospheric hops – components of the spectrogram of the carrier of a powerful broadcast transmitter on 15,380 kHz.

If one looks at the matter in detail (see below), the landing approach of MH2090 at Hannover airport shows up more clearly in the now 12 Hz narrow window. Since this and the meteorite scatter have been brought out with more contrast compared to the previous image by reducing the dynamic range, however, the view of the multipath Doppler of the carrier signal is lost here.

The strong signal visibly illuminates MH2090’s landing approach.

For nine consecutive days, I recorded the 19-m broadcast band and checked literally hundreds of stations. Always the same story:

  • You need a strong signal from the radio transmitter, because only this is able to illuminate especially airplanes in the immediate region of the receiver and cause a Doppler trace.
  • Level fluctuations due to ubiquitous fading are orders of magnitude above aircraft tracks.
  • Meteorite tracks often pass closer to the carrier (WSPR measures within a bandwidth of 6 Hz!) than aircraft Doppler. That they are detected is therefore more likely than the influence of aircraft.
  • Small-scale effects within the ionosphere, unpredictable by statistical ionospheric models such as IRI-2007 (PropLab), alter the signal in terms of Doppler and amplitude, which cannot be even approximately accounted for by the coarse-quantitative methods used by Godfrey, Dr. Westphal et al.

And this is only a part of the ionospheric effects alone. Others are added, see below.

Why a strong signal is needed is shown in comparison in the following figure, taken at the same time with a weaker signal from Saipan (Marianas, 11,300 km away): no aircraft Doppler of the MH2090 landing approach in the immediate vicinity of DK8OK. The energy simply is not sufficient for scattering.

As you can see, you see nothing: Nothing to see at the much weaker signal of Radio Free Asia from Agignan Point/Saipan, 100 kW with curtain antenna towards Myanmar. The illumination is not sufficient to show the landing approach of MH2090.

But even with the strong signal, everything dissolves into noise if you only look at the level. Below the levels of the Saudi Arabian transmitter over two hours – in the upper representation in 1-second intervals, below then averaged in WSPR manner over blocks of 110 seconds each.

In the same way, the original WSPR software also measures the SNR:

The program measures the average signal power over the full transmission.  It measures the average noise power (per Hz of bandwidth) in whatever noise baseline segments (no signal present) are available.  It then scales the noise power up to what would be contained in 2500 Hz bandwidth, and calculates SNR = (average signal power)/(average noise power).

Information from K1JT, April 5, 2022

So only the lower plot is what the WSPR log data gives! And a “drift”, which is however little meaningful.

Above, the transmitter level recorded every second over two hours. Below, the level values summarized in 110-second blocks (“WSPR mode”), mean [average] 47.3 dB, standard deviation 2.5.

Godfrey, Dr. Westphal et. al. do, after all, make the lower recording of the level in 110-second blocks the basis of their argument. Since I can see exactly when an aircraft is visibly illuminated, by observing the Doppler traces in the spectrogram, I have marked these clear sightings in the diagram below – each individual sighting with three adjacent points.

Four aircraft Dopplers were clearly visible in the spectrogram on this day. They are marked with three blue dots each. The level curve underneath does not clearly indicate this at any point. On the contrary, there are only “false positives”.

What Godfrey, Dr. Westphal et al. are doing now is to catch the aircraft from “anomalous” level jumps alone by means of a hat trick that I do not quite see through. What exactly is “anomalous”, can be defined by oneself. But it should be a positive and visible level jump! If one follows the descriptive statistics, it should be “outliers” in their sense. I have checked the level curve according to several of these established methods: not a single detection of even a single aircraft. On the other hand, if the threshold is set accordingly, only false positives are reported – i.e. aircraft where there are none by any stretch of the imagination.

Have I now bored you enough with my long explanations about all too obvious things? Wait – three more things: What does it mean for the carrier signal when the scatter signal is added? Not much, as we already saw above with Hanover VOLMET. The graph below takes a carrier signal of -50 dBm (about S9+20dB on shortwave). The red line shows by how many decibels this carrier is raised when the Doppler signal is between -50 dBm (easy: +3 dB, since doubling) and -130 dBm. The result is very sobering: In the spectrogram, where Doppler and carrier signal can be measured separately, the Doppler signal is always about 20 to 50 dB below the carrier signal.

Moreover, a Doppler trace passes through the 6 Hz filter of the WSPR software within only a few seconds. And only in this time Aircraft-Scatter has the possibility to contribute to the total signal averaged over 110 seconds, from which the SNR is calculated. If one generously assumes 11 seconds in which the Doppler signal contributes to the total signal, one may subtract another 10 dB …

How then with the signals described above and the standard deviations seriously anything else than mumbo jumbo could be read out, nobody understands. Except the President of the ´ DARC, of course. And those who, gladly for money, wave incense for him.

Nothing happens: Even if a Doppler signal doubled that of the carrier, the 3 dB increase in the overall level would be lost in the normal fading.

Oh yes, second last point: “Doppler and drift”. The latter is reported in the WSPR log data. Can we make it short – who has ever looked at spectrograms of the WSPR range knows the sub-optimal quality of many WSPR transmitters. If not: the three screenshots below will convince anyone. Most WSPR transmitters generate the drift themselves. While Doppler shows up on shortwave by traces mainly next to the carrier, but does not shift the whole signal. Like just an unsteady transmitter oscillator … Not to be taken into account are also insufficient receivers …

“Again ye come, ye hovering Forms”: Twelve hours of WSPR signals on 20 m in the 400 Hz window. Solid benchmarks, on the basis of which one can see helicopters spiraling into the icy air of Antarctica via signal strength and drift over thousands of kilometers, look different.
A detailed look at about eight passes of 110 seconds each does not convey pure confidence either. Above all: nowhere is aircraft Doppler to be seen! The “crooked” signals are due to insufficient frequency stability of the transmitters …
… which is certainly particularly clear here in the area marked in red.

All my measurements, and here I come to the announced third as well as last point, were made with a professional SDR at almost 100 dB dynamic range on RF level and under careful qualitative analysis of all signal components at below 0.05 Hz resolution bandwidth. The WSPR log values, on the other hand, are noted at 2,500 Hz bandwidth, at audio level and from a very heterogeneous set of receivers and only quantitatively. Noise, interference, AGC effects and many more have an impact on this data, but can neither be clearly recognized, nor even calculated out.

The following screenshot thus shows what else can happen in the other 92 percent of the total captured bandwidth and what affects AGC, noise, etc. More than a helicopter taking off in Antarctica, that’s for sure …

2500 Hz bandwidth is seen by the receiver and WSPR decoder, 8 percent of which is taken up by the WSPR signals (light center), 92 percent by the rest. And within this bandwidth, this can have a more significant effect on the SNR values calculated from the audio signal than an airplane several thousand kilometers away – for example, due to the AGC.

Possibly Godfrey, Dr. Westphal (this one a former DARC chairman) et al. started their investigations with good intentions. However, they would have had to find out quickly that physics opposes their wide-ranging conclusions. Instead of writing a scientific paper, which could have been argued with, they preferred to go to the tabloid media without any technical knowledge. Irresponsibly, they raise the hopes of hundreds of relatives and the lay public, which their method cannot fulfill. In denunciatory manner they also instrumentalized the DARC for this attitude, which not only jumped over this little stick beaming with joy, but seems to prevent a professional discussion in specialized media by mobbing.

If this article contributes to a repression-free discussion with facts and arguments – as it should be a matter of course – my intention is fulfilled.

P.S.: Allegedly Richard Godfrey, a pensioner from Hesse, wants to apply for a patent for his GDTAAA method and therefore keeps a low profile with a disclosure. Many an unsuspecting person is already dreaming that the entire air traffic control will switch to this method, which will also make the previous investments in high-powered over-the-horizon radars with their smart evaluations superfluous. As far as we can hear, however, the experts are not even laughing about it, but are just shaking their heads at these (radio) amateurs.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s