WSPR & Propagation [MH370] – an Experiment

Completely unintentionally, my last blog on WSPR and MH370 had led to more of a social psychology experiment than a technical science discussion. I expressed my doubts whether it is possible to recognize aircraft scatter from the historical WSPR data by “unusual signal changes” without essential knowledge of further circumstances.
As a reminder: WSPR works with weak transmission power at modest antennas in a rhythm of 110.6 seconds. Apparently this average value is noted and made available as SNR at the receivers.

I objected that practically all other influences on the signal on its way from the transmitter to the receiver (“channel”, with refractions both at the dynamically in three dimension, plus density, changing ionosphere and at the ground) exceed those effects by far, as they are to be expected by airplane scatter. I proved this with 3 x 10’000 level data and 30+ Doppler tracks.

The main proponent of the theory, that the proof is possible against all those odds, reacted with a juicy complaint to the German amateur radio association DARC, in which he argued exclusively personally, but not technically-scientifically. A behavior even more bizarre than trying to prove his actual thesis. The DARC immediately jumped over the stick held out to it, and published few hours later – apparently without or against better knowledge – a sweetish-mendacious “press release“, in which a so far not by technical-scientific papers noticed employee praised that as only beatifying truth.

[Auf Wunsch einiger deutschsprachiger Leser erfolgt in einem weiteren Blog eine Erläuterung dazu.]

WSPR vs. high-resolution data

For all those, however, who are interested in technical contexts, this blog answers a still open question from my last blog:

  • What is the smoothing/generalizing influence of the evaluation of mean values over 110 seconds – which is how the WSPR logbook is supposed to work – on the mapping of the actual signal changes?

Let’s simply test it
For this purpose I have analyzed on September 22nd, 2021 with the professional SDR Winradio Sigma between 07:00 and 12:00 UTC the broadcasting station CRI Kashi, which transmits continuously on 17’490 kHz with 500 kW towards Europe – 5’079 km, two hops. My antenna is a professional active vertical dipole antenna with 2 x 5 m long legs, namely MD-300DX.

With the software SDRC 17’930 level values were noted in dBm/Hz, every second. The FFT analysis was performed with a high-sensitivity resolution of 0.0122 Hz, resulting in a process gain of 53.1 dB compared to the data from WSPR, measured in a 2’500 Hz wide channel. Assuming the carrier power of the transmitter to be 250 kW and the gain of the transmitting antenna HR4/4/.5 to be typically 21 dBi, this results in an additional gain of 47 dB compared to a WSPR transmitter of 5 W on a dipole, which is already strong by its standards. Thus, the total gain of this experiment is 100 dB compared to a WSPR signal. If we assume that signals with SNRs between -20 and -30 dB can still be evaluated, the gain is still a robust 70 to 80 dB. Thus, if aircraft scatter were to be detected on a WSPR signal, it would be even more striking with this factor.

The spectrogram of the five hours’ recording see below, followed by an explanation of the annotations (as with all screenshots: double-click to get the original resolution):

Kashi’s carrier over 5 hours, shown within a window of ±50Hz and a resolution bandwidth of 0.0122Hz at an dynamic range of 90 dB. Explanation see the following text.

The spectrogram reveals a couple of different strong influences to level and frequency of the carrier. Most prominent is the Doppler shift by a moving ionosphere, plus the split-up into o- and x-rays due to the magnetionic character of the ionosphere. You may simulate it with PropLab 3.1, but only in 3D mode. Aircraft Doppler is very weak. It has been verified as such by a different spectrogram with better time resolution, not shown here. You see also some Doppler from meteorite’s plasma in the vicinity of the carrier.

The level of the carrier can be seen from the following screenshot at a time resolution of 1 second, enriched with some statistical data:

Levels over 5 hours. Mean = -44.02 dBm, Standard Deviation 7.872, Range 58.81 dBm. Max/min: -21.97 dBm/-80.78 dBm

The next screenshot shows the whole 17’930 datapoints, split up into consecutive groups of 110 each. This should simulate the the 110.6 seconds which the WSPR logbook boils down to one SNR value plus on “drift” value. To read this contour map:

Vertically you see 163 columns. Each column contains the levels 1 … 110, and 110 x 163 = 17’930 total level values. For the first time, you can see here the dynamic within a column of 110 values each.

Contour diagram, showing all 17’930 level data, grouped to 163 blocks of 110 data points each.

So far, we retained the level information of all 17’930 data points. What happens if WSPR boils them down to chunks of 110 seconds only? This question answers the next screenshot:

What is lost by boiling down the 17’930 level values at 1 second’s distance to 163 chunks of the mean of 110 values? This screenshot shows the answer.

If it still isn’t understood that information which simply are not palpable in the 110-seconds’ chunk cannot be “interpreted” as this or that, a zoom-in must convince you:

Same as above, but zoomed into. WSPR logbook will keep only the Chunks. So all information has to be derived from just the red line! Imagine that you don’t have any more information – no “Raw”, no “Spectrogram”.

Looking at the both screenshots above: are you still sure to see any faint details (refer to spectrogram on top of this blog) like any Aircraft Doppler just from the chunks? You have also seen that the “drift”, shown in the WSPR logbook, may have manifold sources, ionospheric Doppler prevailing.

Stanag 4285 & PSKSounder – a better mode

There, of course, is a way out of this dilemma: since many years free PSKSounder provides an excellent tool to extract many more information from STANAG 4285 signals, see the following screenshot:

PSKSounder shows relative “time of flight” of a Stanag 4285 signal. Here with FUV, French Navy in Jibuti. You see that the structure of the spectrogram of the signal at the right has it source in two strong and different paths of the signal. Their times of arrival differ by about one millisecond. This procedure is very sensitive and is also used to reliably reveal meteorites and – aircraft!

Finally two things: The path between two stations does not always have to be exactly reversible – that is, if two stations are equipped exactly the same, it is very likely that a different signal will be detected in each case. And if the black box of MH370 should indeed be found in the area supposedly designated by WSPR, it is due to many things, but certainly last of all to WSPR.

After which methods it might be tried nevertheless, one can read already now in Grete De Francesco’s “The Power of the Charlatan”, Yale University Press, New Haven/USA, 1939.

One comment

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