Norman Lockyer Observatory Radio Group


  • Increase font size
  • Default font size
  • Decrease font size

Radio Meteor Detection 
By David Knight

Meteor: Any atmospheric phenomenon, but particularly a piece of matter from space rendered luminous by collision with the upper atmosphere.
Meteoroid:  A sand to boulder-sized particle of debris in space. 
Meteor trail:  The track of totally ionised gas (plasma) left in the wake of a meteoroid as it enters the atmosphere, dissipates its kinetic energy, heats up, and vaporises.
Meteorite: A fallen meteor remnant. A mass of stone or iron which has managed to reach the ground without vaporising.

The detection of meteors by radio is most readily accomplished by a method known as "forward scatter".  This technique usually exploits the existence of a VHF radio transmitter intended for some other purpose (such as radio or TV broadcasting) and which is preferably situated some way beyond the optical horizon so that the direct signal does not desensitise the receiving equipment.  The radio signal reflects mainly from the electrically-conductive meteor trail as it forms and dissipates, causing a brief signal to be heard on or close to the transmitter frequency.  The trails form in the ionosphere (i.e., the upper atmosphere) at a height of about 100 ±20 Km [].
     Direct reflection from the meteoroid itself  is not so readily detected.  Meteoroids are not necessarily reflective at radio frequencies, they are usually small (0.05 - 200mm  - see: Meteors in the Atmosphere) and they generally enter the ionosphere at supersonic velocities.  Thus the direct signal is usually weak; and the initial Doppler shift is  large, making it difficult to associate the signal with the transmitter.  Sometimes however, a Doppler shifted signal is observed to slew onto or across the transmitter frequency at the beginning of the detection event.  This is the reflection from the ball of plasma surrounding the meteoroid (as opposed to the trail left behind), and is known as the "head echo". 
     The term "radar" is sometimes used to describe the forward scatter detection method.  Note however, that 'radar' is an acronym for 'radio direction and range'; and so, although distance and direction information can be extracted from data aggregated from an array of receivers, a single receiver installation does not constitute a radar system.  A single receiver can only strictly report an estimate of the number of meteoroids which enter the ionosphere in the region illuminated by the chosen radio transmitter.  Other interesting aspects of the meteor strike can be inferred from the recorded signals, but apparently obvious information, such as the relationship between signal strength and meteoroid mass is complicated by issues such as signal polarisation, trajectory and transmitter coverage.  A discussion of meteor location and radiant determination from multiple head echo observations is given by David Entwistle.
     One advantage of radio detection is that it works when the sky is light or when the sky is dark but overcast.  By choosing a sufficiently powerful host transmitter, it also possible to record meteors which are too faint for the human eye even in the darkest and clearest conditions.  A figure of between 2 and 10 times as many meteors as can be seen by visual observation under ideal conditions is sometimes quoted; but this must depend on the transmitter power and radiation pattern.

The following links give useful background information:
Internationa Meteor Organisation: ; see particularly, meteor science pages: . . . . . . .

 Transmitter Choice
The chosen transmitter needs to be beyond the horizon.  Distances in the range 200 to 1000Km are suitable.  The transmitter must also operate in the VHF range, so that the direct signal is weak or absent in normal atmospheric conditions.  The frequency should not be excessively high however, because the received signal power is proportional to the cube of the wavelength [].
     A high-power transmitter is best.  Many broadcast transmitters radiate tens or hundreds of Kilowatts, but modulation has a generally deleterious effect on meteor-related information.
     If a transmitter is frequency modulated, the carrier signal is broadened by the programme information.  This has the effect of smearing out the Doppler-shift information in meteor echoes.  Hi-fi broadcasts in the 88 - 108MHz band use wide-bandwidth FM.  Doppler information is lost, and meteor detection is a matter of listening for snippets of signal from stations which are otherwise out of reception range.
     An analogue television signal is a mixture of amplitude and pulse modulation.  It appears in the frequency domain as a series of narrow spikes spaced at an interval equal to the frame refresh rate (25Hz in Europe and Russia, 29.97Hz in USA).  The picture information is encoded in the relative amplitudes of the spikes, but there are clean and constant signals on the specified channel frequency and at intervals equal to the line scanning frequency (15625Hz for 625 line TV).  This repeating 'comb spectrum' characteristic complicates the interpretation of Doppler information, but does not prevent it.  Unfortunately, most TV stations have switched to UHF, and there are few VHF analogue transmitters left.
     The ideal host transmitter is one which produces a strong stable continuous signal of very narrow bandwidth (a continuous-wave, or CW transmitter) on a clear channel and is never shut down for maintenance.  There are no ideal host transmitters; but at a distance of about 750Km from East Devon, there is one belonging to a space surveillance radar system called 'GRAVES' which has tolerable idiosyncracies.


'GRAVES' stands for: 'Grand Réseau Adapteé à la Veille Spatiale' (Large adaptive space-surveillance array).  It is a bistatic (separate receiver and transmitter) continuous-wave radar system used to track satellites.  The transmitter site is located at a disused airfield near Dijon, at the foot of the Alps, 202m above sea level.  Continuous wave operation makes a VHF allocation possible without causing an international outcry; the chosen frequency being 143.050 MHz, just outside the 2m
(144-146MHz) amateur band.   

The following links provide more information, and the transmitter array is pictured below:
Graves sourcebook (18MB pdf).
Note: According to the Graves Sourcebook, the ERP is "several Megawatts".  This is not consistent with claims that the American NAVSPASUR CW radar system, at about 768KW, is the most powerful in the world, although it may be that the American figure is actual rather than effective radiated power. [].
The Graves Radar, a VHF Beacon.  PE1ITR .

GRAVES transmitting array. Picture: © ONERA 1996-2006 .

Aerial view of the transmitting array. Location: Broyes-lès-Pesmes, near Dijon, France. 47º 20' 52" N, 5° 30' 54" E.
Elevation: 202m ASL.  Picture: © 2010, Google Earth, Europa Technologies, IGN-France, Tele Atlas.
Each antenna panel covers 45°, giving 180° coverage in a Southward direction.

The approximate GRAVES ionospheric illumination field, and the corresponding antenna direction from East Devon.
Underlying picture: © 2010, Google, Europa Technologies, Tele Atlas.
Based on a Powerpoint presentation by Andy Smith, G7IZU:
Note that this illustration should not be taken too literally.  Antenna radiation patterns do not have the idealised sharp cutoff shown, which means that the transmitter array will illuminate an area somewhat greater than indicated.

Meteor detection using the Graves signal

The graves transmitter is not a perfect radio beacon.  Firstly, the part of the spectrum in which it operates may be used by other services and falls near the second harmonic of a PMR channel.  Interfering signals are however distinguishable from meteor echoes if the receiver has a good enough signal-to-noise ratio.  Secondly, the signal is not truly continuous.  The antenna radiation pattern is switched on a timescale of 3.2 seconds or longer, and this can modulate long meteor echoes.  Also, when atmospheric conditions permit direct signal pickup, it causes an intermittent carrier signal to heard on the radar frequency.  Thirdly, the radiation pattern is much more tightly controlled than that of a broadcast transmitter, restricting the region of ionosphere from which echoes can be received.  Finally, the meteor echoes are relatively weak due to the short wavelength (λ=2.1m), there being a 14dB disadvantage relative to the 6m TV band because signal strength is proportional to λ3.
     In East Devon, due to the short wavelength and the long distance to the transmitter, it is necessary to use a good radio receiver.  Some commentators attach little importance to receiver performance, but 2m amateur radio transceivers of 1980s vintage typically had front end noise figures of around 14dB, and the practice of connecting antennas using tin-plated cheapernet RG58 cable can easily increase the overall noise figure to 20dB or worse.  Such an installation will give poor results, and worse still, the 1/f noise of a worn-out front-end transistor can combine with received CW signals to produce an output which can be mistaken for meteor echoes.
     Ideally, a mast mounted low-noise preamplifier with a good roofing filter will give the best scientific outcome.  Good test results were obtained however using receivers having <1dB noise figure and connecting to the antenna using low-loss cable to obtain an overall noise figure of <2dB. 
     The chosen observation method was to set the receiver to 'upper-sideband' mode (USB) and tune to 143.049 MHz.  In that case, signals at 143.050 MHz are heard as 1KHz audio tones, and Doppler shifts are measured relative to this 1KHz datum (which is nominal, depending on drift and calibration error).  Audio signals were converted into spectrograms using a computer and the venerable Spectrumlab software of DL4YHF: .

Initial tests were carried out at the author's home on East Hill, Ottery St Mary. The receiver was an Icom IC275H, recently serviced and thus in a known good state of calibration. The IC275 has a front-end noise figure of <1dB and is one of the best commercial 2m SSB transceivers. Two antennas were compared: a low-gain rooftop-mounted vertically-polarised antenna designed for 145 MHz use; and a high-gain (ca. 11dBi) horizontally-polarised Yagi, which was fettled specifically for 143.05 MHz using an MFJ269 antenna analyser, then mounted above rooftop height and pointed along the Otter Valley in an approximately SSE direction.  Antennas were connected directly to the receiver using low-loss coaxial cables, with a coaxial switch at the receiver socket for changeover.

Receiving antennas used for the initial evaluation (East Hill, Ottery St Mary, 157m ASL).  11dBi Yagi with Gamma match and 15m of Andrew
FSJ4-50 Superflex feedline. 

Low-gain vertical (j-pole) with RG213 feedline is concealed in the flagpole in the background.

The view in the SW direction is blocked by East Hill, so direct pickup of the Graves signal is weak.

Apart from an increase in signal strength when using the Yagi, the two antennas gave similar results.  

Meteor echoes have a characteristic sound, usually described as a "ping", when heard through a receiver in SSB mode.  Just because people say that these noises are due to meteors however, doesn't mean that they are.  As mentioned earlier, a receiver with a crackly front-end transistor tuned to a weak continuous signal also produces pinging noises.  Some kind of verification is needed.
     The basic test of veracity is statistical.  On a short timescale (ca. 1 hour), the entry of meteoroids into the atmosphere is random in time.  Such events follow a Poisson distribution [], whereas bad transistors and man-made interference produce outputs which are auto-correlated, i.e., the presence of signal tends to predict that there will be more signal to come.
     Outside of specific meteor-shower periods (i.e., presuming that the earth is not passing through a comet tail) the number of meteor strikes peaks at 06:00 (actual time, not 'daylight saving' time) and is at a minimum at 18:00.  This is because the Earth's orbital motion is in the direction of the dawn terminator, and so at 06:00 an observer is at the longitude which is collecting most dust from the Earth's orbital plane.
     The most convincing verification is, of course, visual correlation.  From East Devon, the ionospere above the Graves transmitter is at an elevation of about 5º in the SE direction.  At around 23:30 UT on the 13th of December 2010, meteors belonging to the Geminid shower were occasionally observed heading towards central France.  Standing outdoors away from light pollution, using a set of wireless headphones (operating on 863 MHZ) to listen to the receiver output, such meteors were found to become audible as they disappeared below the treeline.

Polarisation diversity
The polarisation of the received signal is unpredictable.  It depends on the polarisation of the transmitter, but is randomised by interaction with the plasma which reflects it.
     An experiment was conducted using two radio receivers and two antennas (the H and V polarised antennas discussed earlier).  The second receiver was an SSB Electronics LT2S 2m transverter (also having <1dB noise figure) used in conjunction with a Trio TS930 HF transceiver (tuned to 27.049 MHz USB to receive 143.049 MHz USB).  It was found that a good many of the weaker short pings (under-dense meteors) could be heard on one receiver or the other, but not both.  Stronger meteor signals of longer duration were found to oscillate between the two polarisations, giving a stereo effect.

Doppler shift
Calculating the Doppler shift:
frx = ftx (1-v/c) / (1+v/c)

The difference between the transmitter frequency and the received frequency, for objects travelling at non-relativistic velocities is given by:
Δf / ftx = (frx - ftx) / ftx = 2 v / c
where v is the velocity of the object relative to the source, and c = 299 792 458 m/s.

If the frequency shift (Δf) is 100Hz, and ftx = 143.050 MHz, then:
v = 100 × 299792458 / ( 2 × 143050000 ) = 104.7859 m/s
Note that v is proportional to Δf.   Thus we can simply memorise a figure: 1.05 m/s per Hz.
1m/s = 3.6 Km/h
Doppler shift is negative for receding objects ('red-shift') and positive for approaching objects ('blue-shift').  Doppler shifted reflection signals always move from the high-side to the low-side of the illumination frequency.  Hence, when listening in USB mode, audio signals which are descending in frequency are possible reflection signals from moving objects. 

The speed of sound in air is approximately: 20.0457√T m/s, where T is the absolute temperature (centigrade+273.15). At 0°C (273.15 K), the speed of sound in dry air is:
20.0457√273.15 = 331.3 m/s
Thus, for Graves, we have: 331.3 × 1.04786 = 347 Hz / Mach (approx.)

Experimental results and example spectrograms
Note that the frequencies marked on the spectrograms below are nominal.  The master oscillator of the IC275H has some thermal drift.  Also, the synthesizer tunes in 10Hz steps, but the frequency readout is only to the nearest 100Hz.  The frequency setting is generally within ±100Hz.  This is an evaluation.  An optimal installation would have its frequency precisely settable, and ideally would have its master ref. locked to a national standard.

East Hill
Preliminary experiments.

Meteor strike heard using the vertical antenna. The time on the computer clock is not accurately set at this point (probably within ±1 min). The darkening of the spectrogram either side of the strong signal is due to activation of the receiver's AGC system.  The initial signal shifting from high to low is the head echo.  The longer signal on the transmitter frequency is the specular reflection from the ionisation trail.  The long-duration signal at about 940Hz is probably the direct signal from the Graves transmitter.


Initial experiments with the Yagi antena.  The evenly-spaced vertical lines are multiples of the frame rate of a CRT-type computer monitor which was in use at the time.  The problem was cured initially by altering the screen refresh rate (a better solution is to use an LCD monitor).  While this was being investigated, a meteor-strike event lasting nearly 20 seconds occurred.  Switching of the Graves antenna field is probably responsible for the gap in the recording. The object was probably part of the Leonid shower (Nov. 10th - 21st, the tail of Comet Tempel-Tuttle. See:


A short duration meteor strike event. 
A diffuse band centred at about 1.1 KHz in the audio range does not change with receiver tuning, and so is a receiver audio response artifact.  A carrier signal drifting from low to high is a spurious emission from a recently switched-on computer (Dopler shifted signals always move from high to low frequency).

Clock synchronisation
Comparison with observations made elsewhere requires accurate clock setting.  Windows XP and Windows 7 operating systems have a built-in internet time client (double-click the clock and see the 'internet time' tab).  A more versatile synchronisation program, which runs on Microsoft OS' from Windows 95 onwards, is Dimension 4: .

Tests at the NLO
Having established that meteors in the Graves Ionospheric Illumination field could be detected from East Hill Ottery St Mary, the software was set-up on a laptop computer, and the receiver was taken to the NLO and connected to a steerable horizontally-polarised 6-element quad antenna (ca. 9dBi) mounted on the Radio Group's Stumech Tower.  The elevated location reduces ground noise, but increases direct transmitter pickup and interference from other signals.

Peaking in strength in the SE direction is an intermittently appearing signal consisting of groups of long pulses.  A commonly recurring pattern is 3.2s strong, 19.2s weak.  This is the direct transmitter signal, subject to atmospheric fading and modulated by an electrical steering sequence which causes the received strength to vary.

A meteor strike appears a few seconds before 20:51 in this spectrogram.  There are also a couple of possible under-dense events, but the direct signal and other clutter on channel makes interpretation difficult.  The 800Hz-spaced comb-spectrum signal is discussed below.

Visible in this spectrogram is a weak comb-spectrum with 800Hz spacing which undergoes a frequency step between 21:02 and 21:03.  After some investigation, this was found to be an artefact of the IC275H.  It is not affected by the receiver tuning, but it also fades in and out and is not present on all spectrograms.  When it is present, it remains when the antena is disconnected or shorted. 

One further test carried out at the NLO was to observe the effect of using a 25W NBFM transmitter operating on 70.45 MHz, with its antenna 10m away from the 2m receiving antenna.  Slight overall gain-variation effects (blocking) were visible in the spectrogram, but they were not seriously disruptive.  The IC275H however is a high-performance receiver.  A good roofing filter will be necessary for any meteor detection receiver operated at the NLO site, due to the elevated location and the abundance of PMR and telemetry systems in the surrounding region.

More experiments on East Hill

Here we see a possible meteor strike event lasting about 20 seconds. There is considerable Doppler broadening of the specular reflection, but no head echo.  The brightening of the trace at twice the audio frequency of the main signal is due to second-harmonic distortion in the audio channel.  The Graves frequency is otherwise completely free from interference.

Meteor strike.  A change of colour scheme improves the visual impact.

Meteor strike

In the spectrograms below, the audio level is given as a side bar.  The meteor events are often loud, but over a 2.4KHz band, they cannot be distinguished from interference clicks and pops by level measurement.



Noise test

This spectrogram shows the difference in noise level between the IC275 receiver with input shorted and with the Yagi antenna connected.  The background with the antenna connected is mostly thermal (ground) noise from East Hill. 

Spectrograms and audio recordings made during the 2010 Geminid meteor shower are given in a separate article.  In that article, the use of the full receiver bandwidth (up to 2.6KHz) gives head echo recordings up to Mach 4.9.

Detecting spacecraft
See: Graves Sourcebook (qv): p44, p51-56.
Some satellite detection events (so far unidentified) are shown below. Note the change to horizontal time and vertical frequency axes.

The descending tone passes through the Graves frequency (zero Doppler shift) at 22:12:43.  The gradient is -38.9Hz / sec at crossover.  Note that the Graves antenna points South, and the reflection becomes strong when the signal is red shifted.  Hence the satellite was travelling from North to South and passed the Graves latitude at 22:12:43 (± ca. 1sec)
mp3 audio clip 22:12:42 - 58 .
This recording was made during the 2010 Geminids meteor shower, and two under-dense meteor strikes also occur during the clip.
Recorded at the NLO using a 6-element quad antenna pointing SE.

In this case, the 3.2 second bursts indicate clearly that the object is being illuminated by Graves.  The descending tone passes through the Graves frequency at 03:49:15.  The gradient is -50.8Hz / sec at crossover.

mp3 audio clip 21:29:47 - 53 .
This is similar to a meteor head echo, but there is no echo from the meteor trail.  Maximum strength is when the object is approaching Graves.  Possibly a satellite in low Earth orbit.  Zero Doppler shift occurs at 21:29:50.

Moon reflections
See: Graves Sourcebook, p61.
see also: Website of PE1ITR / Graves .
When the Moon is South of Dijon at an elevation of about 20°, the Graves signal is seen to double.  The Moon reflection has variable Doppler shift and should appear delayed by 2.67s relative to any corresponding structure in the direct signal.

Aircraft reflections

Commercial jet aircraft cruise at about 250m/s.  Such an aircraft flying directly towards or away from the Graves transmitter would have a Doppler shift of ±262Hz.  For aircraft flying at constant velocity in other directions relative to the transmitter, the relative velocity will be less than the actual velocity and will vary with time, passing through zero when the aircraft is at a tangent to a circle centred on the transmitter.  Thus, typical aircraft reflections (subsonic) will lie within ±260Hz of the transmitter frequency; and, when the receiving mode is USB, will move from high to low in the audio range.
     There are NO aircraft reflections in the spectrograms shown above.  This might seem surprising at first, but bear in mind that the transmitter has a tightly controlled radiation pattern and will only illuminate aircraft flying a little to the South of Dijon.  The ceiling for commercial aircraft is about 40 000 ft (12.2 Km).  This height above Dijon is below the horizon for observers anywhere in the UK.  Hence aircraft reflections will only be seen in conditions favourable for long-range tropospheric propagation (typically in summertime), in which case, the direct transmitter signal will also be very strong. 

A spectrogram showing aircraft and moon echoes is given here: .

>>> Geminid observations .