Monday, December 18, 2023

Meteor scatter experiment with the 2023 Geminid shower

In this post, I'll outline how I used an amateur radio beacon 500kms away to find the peak of the 2023 Geminids meteor shower and what else I noticed.

Introduction... Back in 2022, a new radio beacon called GB3MBA was put on air on the 50 MHz band and its primary purpose was carrying out meteor scatter experiments. Located near the town of Mansfield in the centre of England, the beacon runs 75-watts into a small Moxon antenna which beams straight up.

Unlike most amateur radio beacons, this one is basically a carrier that is on nearly all of the time which is exactly what is needed for meteor scatter experiments. The main coverage area is the UK itself but as I'm 502 kms from the beacon, I'm close enough for it to be of use.

I expected that most of the signals that I would hear would be from meteors burning up over Wales and would be arriving at my location about 20 degrees above the horizon.

Experiment Objectives... This is what I wanted to do...

  • Carry out an experiment as a 'proof of concept', take measurements, gain experience and see how I could improve on it.
  • Take 'hard measurements' with real data as opposed to just forming an opinion.
  • See if I could plot and find the peak of the 2023 Geminid meteor shower as a form of 'citizen science'.
Methodology & Equipment... For the previous few days before the peak, I was listening to meteor bursts from the beacon to establish the best method to carry out the experiment.

Radio... I was using an old Kenwood TS-690 amateur radio transceiver which still works fine and is pretty sensitive. I tuned it to 50.408 MHz and set the mode to CW. It drifts a few Hz with the changes of temperature in the shack but it was fine for this experiment. I just needed to be careful to make sure I was always on frequency.

Antenna... In the run up to the experiment, I used a 2-element Yagi in my attic beaming 120-degrees (east-south-east) to get ready. This was beaming about 45 degrees away from the beacon and being indoors, it wasn't optimum. Just before the experiment, I built a vertical T2LT (flowerpot) half-wave antenna and mounted it about 6-metres above the ground. It had a clear view of the sky towards Wales and the beacon.

Software... I used the SpectrumLab programme to look at the audio spectrum so that I could see any bursts and pings of signal. Under the 'Quick Settings', I used Slow Morse Reception ("QRSS") and I then selected 'Slow CW, 30 sec/dot'. 

I knew it wasn't practical to look at a screen for hours on end. At this setting, the audio spectrum is about 30 Hz wide and it takes about 1h 20m for the signals to drift from right to left across the screen. What this meant in reality was that I could look at the screen every 60-80 mins, count the signals on the screen and record the number of pings and bursts.

What to record?... I knew whatever I did, it had to be consistent for the observing period. I decided to use the number of signals seen in a 30-minute period. If it was a long burst or a short ping, I counted it as one signal. 

This raises the obvious question... should I count the number of pings and bursts, should I record the strong signals versus the number of weak signals? The problem is where do I draw the line? I knew I'd see long bursts, short pings and everything in between. It was the same for the signal strength... what's a 'strong signal'? 

I decided I just wanted 'hard data'. Was there a meteor scatter signal present on the screen... Yes or No. It's simple and there's no ambiguity.

Time period... For the experiment, I counted the number of meteor scatter signals from 17:30 UTC on the 14th of December to 22:00 UTC on the 15th of December 2023...a period of 28 hours and 30 minutes. This time frame coincided with the predicted peak of the Geminid meteor shower.

Results... This is the chart that I generated for the 28.5 hour period.

The recording period starts at 17:30 UTC on the 14th of December when the Geminid constellation and radiant point would be just above the horizon. I'll break down the chart in more detail below but the key thing to note is the peak at about 20:00 UTC on the 14th of December.

I had intended to stop recording early in the morning of the 15th but I was wondering if the peak was really the peak? Would there be another peak 24-hours later when the radiant point would be at the some position in the sky? As you can see, there was no peak 24-hours later. I had actually recorded the peak of the Geminids on the 14th.

Results - Time Frame - 17:30 UTC 14th Dec to 05:30 UTC on the 15th of December... The peak, the dip and the highest radiant...

The first 12-hours is shown above in more detail.

The start... I did a search on Google and it said that the constellation Gemini would rise above the horizon in London at 17:23 UTC. I'm 8.5 degrees west which means it would be 33 mins later... 17:56 UTC. The mid point might be say 17:39 UTC? I found another website which said that the radiant point would rise around 16:07 UTC??

This is further complicated by the fact that we're really looking at a time when meteors from radiant point in the Geminids would be hitting the atmosphere 100kms about Wales? I think 17:30 was a good enough time to start.

The peak... Between 19:00 and 20:30 UTC, I had counts of 15, 19 & 21 for each of the 30-minute intervals.

This is a screenshot I took during this period.

First, I'll explain what you are looking at. The vertical axis is a 30 Hz wide segment of the audio spectrum. The screen is moving from right to left and the white dotted lines are the 5-minute time intervals.

If I was hearing the beacon signal directly, it would be a solid yellow line running from right to left across the centre of the screen.

What you are looking at are the pings and bursts of meteor scatter signals which are spread out across the audio spectrum due to doppler shift.

Prior to the shower, I had seen that the peak was expected at 19:30 UTC on the 14th of Dec. Looking at my screenshot, I think the peak of the 2023 meteor shower occurred between 19:55 and 20:20 UTC on the 14th of December.

The dip??... I did make a note of how weak the reflections were from 22:30 and 00:00 UTC and this can be clearly seen in the chart above. 

As the meteor shower radiant moves slowly across the sky, the angles between the transmitter, receiver and the trails of the meteors will change. At some points, the meteor trails will be parallel between my location and the beacon and at other points, the trails will be perpendicular.

I suspect the dip is due to the geometry of the meteor trails but I'm not sure. It could also be just a statistical error. I would need to repeat the observations on several days around this time period to see if it repeats.

Maximum elevation... At about 01:30 UTC, the radiant point was supposed to be at its highest elevation which was about 71 degrees above the horizon.

Before the experiment, I was wondering if this mattered? Would there be some sort of peak when the radiant peak was at its highest point in the sky? The answer seems to be no, there is no peak.

Results - Time Frame - 01:30 UTC to 11:00 UTC on the 15th of December... The daily morning peak...

When it comes to random meteors entering the atmosphere, the best time for signals tends be around 6am local time. This is because the speed of the earths rotation is added to the speed of the meteors resulting in more energy to dissipate.

This enhancement can be clearly seen in the chart above. It increases after 02:00 UTC and decreases after 10:00 UTC. For me at least, it seemed to be an eight hour window.

The converse is also true. At 6pm local time, the speed of the earth's rotation is subtracted from the original speed of the meteors.

This means that the peak of the 2023 Geminids at about 20:00 UTC happened just two hours after 6pm for me which isn't a great time.

As I was looking at the original chart with the 28.5 hour display, I realised that I was looking at two peaks but both were completely different. One is a fixed peak and the other is a variable peak.

The fixed peak was the peak of the Geminids at about 20:00 UTC on the 14th of December. That's fixed, the peak is the peak.

The morning peak however will depend on where you are. Somewhere in the world, the peak of the Geminids meteor shower will coincide with 6am local time. That turns out to the western half of the Pacific. Anyone in say Hawaii, Alaska, Korea, Japan, east China and the Philippines should have seen some very good meteor bursts and pings. That is of course conditional on having the radiant point above the horizon and that will depend on the location.

Results - Time Frame - 09:00 to 22:00 UTC on the 15th of December... This is the last 13 hours of the plot in more detail...

The radiant point for the Geminids sets about 10:00 UTC and a drop in numbers can be seen. The signals from about 10:00 to 17:30 UTC are random meteors and are not Geminids as the radiant point is below the horizon.

You can see it was particularly poor from 16:30 to 17:30 UTC. Then the Geminids radiant point comes above the horizon and the numbers increase again.

Other observations... Aircraft Scatter... I've seen plenty of aircraft scatter on signals before but I was really surprised to see it on a signal from 500kms away.

These can be very easy to miss. On the QRSS 3 setting, these appear as almost horizontal lines due to the larger vertical scale and faster scrolling speed. On QRSS 30 as used above, the slant becomes a lot more obvious.

As you can see, the doppler shift is in the region of 5-7 Hz. When I was seeing these signals appear, I would check FlightRadar 24 and sure enough, there was often a large plane to the west of Aberstwyth on the west coast of Wales. This is around the halfway point from the beacon to my location.

This suggests to me that while the GB3MBA beacon is 'beaming straight up', there is plenty of RF heading towards the horizon as well.

You'll notice as well that the trails last for about 90 seconds which should be enough for digital modes like MSK144 or possibly FT4? I'd have no doubt that a lot of those one off FT8 decodes on the 10m and VHF bands that people get are due to aircraft scatter like this.

Additional Notes...

  • I couldn't tell if any meteor heard was a Geminid or not for the time the radiant was above the horizon.
  • Despite all of the 'streaks' visible on the screen, the signals were actually pretty weak. Many were not audible to my ear and I'd say the strongest signals were probably something like 519. 
  • The process that I used was very labour and time intensive. I literally had to set an alarm to wake up every 60-80 mins overnight and take the measurements.
  • My resolution was 30 mins which may have limited the resolution in that I couldn't tell if two very close bursts were one or two.

In Conclusion... I think the experiment overall was a success. I was able to see the peak of the 2023 Geminid meteor shower which was from 19:55 to 20:20 UTC on the 14th of December.

I'm sure there is room for improvement but I think this is a useful 'citizen science' project which can be done by anyone with a good receiver and antenna. This method is also easy enough that it's not restricted to those from a radio background. 

Anyone with a serious interest in astronomy and meteor research could do this and consider it as an additional tool. This system works regardless of the weather. While 'visual' amateur astronomers need clear skies, the 'radio' based amateur astronomer can carry on regardless.

It might make an excellent project for an astronomy or radio club and maybe give a presentation to students as part of a STEM outreach programme

I think it proves that GB3MBA is indeed a very useful beacon for meteor scatter research and I think the challenge now is for them to make more people aware of it.

You can tell from the length of this post that this was a deep dive down the rabbit hole for me. As soon as I had one question answered, I thought up of two more additional questions.

The future??.. I'd see these as areas for improvement.

  1. My antenna was a vertical half-wave which is omni-directional. I think something like a 3-element Yagi would give me at least 6dB more gain, reduce noise and result in more pings and bursts.
  2. My experiment was for 28.5 hours with a resolution of 30 mins. I think in the future, I would use QRSS 3 with a finer resolution of 10 minutes and plot the period around the predicted peak in more detail.

More information...

For more information on the GB3MBA beacon, go to

There is also the BRAMS beacon in Belgium on 49.970 MHz which might be of interest...


PE4BAS, Bas said...

Very interesting post John. Personally I would not take the efforts you made. I would try to automate the process and analyze after the event. One question: why is GB3MBA using a directional antenna pointing upwards? Why not just an omnidirectional like a vertical? 73, Bas

John, EI7GL said...

Hi Bas. The primary purpose of the GB3MBA beacon is to do meteor research in the UK. They point the antenna straight up so that receiving stations in the UK can get good signals.

Have a listen on 49.997 MHz to the BRAMS beacon from Belgium and you should get some good pings in the Netherlands.

One possible problem with automation is if it will be as accurate as the human eye? It's possible to listen to some beacons with FT8 or PI4 but the number of decodes would be very small.


Anonymous said...

Hello John,
good old ARGO-SW allows automatic capture, provided your receiver is stable enough you don't have to get up several times at night.
On Dec.14th evening I decided spontaneously to check out meteor scatter as the sky was conpletely cloudy so no chance to see any meteor trails. I hooked a 1W/28MHz QRSS transmitter built from existing homemade modules to an EFHW vertical at about 18:30z.
Given the low effort I was pleased to see fragments of my signal on G0MQW's grabber about 500km from here within less than half an hour - the best bursts lasted almost a minute.
Stayed up late night, but results were best well before midnight.

John, EI7GL said...

Thanks for the info Peter. I used ARGO in the past for QRSS but forgot about it once I started to use SpectrumLab.

I must have a look at it again for the meteor scatter signals.

John, EI7GL