RADIO METEORS

RADIO METEORS - Observations using simple amateur equipment

New and definitely Under contruction - last update 29th Dec.

Introduction:

Recognise this scenario? A major meteor shower is in progress, and outside the sky is completely overcast. It's frustrating to say the least. This prompted me to look into methods of using radio waves to monitor meteor activity - at least this can be done when the cloud gods are angry, or even in broad daylight!

Using radio reflection is nothing new, such methods have been used for many years as a way of estimating the gross number of meteor strikes, and the daily 'Radio Count' is often quoted as a measure of the intensity of a meteor shower and to plot it's timecourse. The only problem with this is that without further analysis the gross number of radio reflections can be influenced by things other than meteors - ionospheric reflection, aurora, even planes and satellites. Without direct corroborative data it's also difficult to correlate radio reflectivity with visual appearance. Any radio reflection due to meteors is dependent upon an ionised plasma 'cloud' being generated in an opportune location (geographically with repect to the transmitter and your receiver) so it only works when 'looking' in a certain direction. Of course, if you have the ability to 'look' in more than one direction (by simultaneously tuning to more than one transmitter) then you can get greater sky coverage.

Compared to traditional methods of years gone by, today's would-be radio meteor observer is able to take advantage of new technology and freely available software. Fast computers are relatively inexpensive these days and appropriate software featuring complex analysis routines can be downloaded off the 'net. Precisely tunable (and stable) receivers are also available which don't cost an arm and a leg. You probably already have a suitable 16-bit A-D converter in the form of your sound card (with it's audio-in socket). With these tools it's possible to do much more than simply count transient signal spikes, it is now possible to use the computer to analyse the signal in real time using FFT algorithms to generate dopplergrams, and such data can provide much more information about individual meteor events. After spending a couple of weeks accumulating thousands of individual traces the thought occured to me that perhaps these dopplergrams could be used to differentiate between broad classes of meteors based on the characteristics of their ionisation trail. The objective therefore, was to develop a classification system based on these traces and look at the distribution of meteors within each.

Methods:

The basis for making amateur radio meteor observations is that you have a receiver tuned to the frequency of a distant (1000Km or more) transmitter, one which you cannot recieve the signal from by ground propagation. The distance to this transmitter and it's direction will determine the area of the sky within which reflected signals can be monitored. Within this defined area, any meteor entering the atmosphere and vapourising will leave an ionized plasma trail (of a size dependent upon the meteor's mass and velocity), and it is this trail which reflects the radio signal down to your receiver. This method is called 'Forward Scatter Radio Observation' and is different to the 'Back Scatter RADAR echo' method used by professional radio meteor observers. Back Scatter involves transmitting a signal (RADAR) and recording the echo bounced back to the originating position.

The type of transmitted signal best suited to this purpose (for generation of dopplergrams) is an AM carrier wave, within a frequency range of around 10MHz to 70MHz. The reasons for this are complex and links are provided below to articles which provide more extensive techinical explanation of these limiting factors. Suffice it to say that we will be using a receiver tuned within this frequency and using a fairly narrow banwidth in SSB (single side-band) mode.

With the receiver thus tuned, the resulting audio output is digitised and fed into a personal computer (conveniently done using the computer's sound card). The digitised audio is then presented to a FFT (Fast Fourrier Transform) analysis program, where the entire audio spectrum is converted to a single line of a graphical display (Power Vs Frequency) at 0.5 second intervals. As this graphical display slowly scrolls across the screen (right to left) the intensity of the received signal within a range of audio frequencies produces what is called a 'waterfall display', with time as the X-axis. As an example, a carrier wave producing a 1 KHz sine wave (i.e., a pure audible tone of 1KHz frequency) would produce a continuous straight line across the display centered on 1000Hz (as shown on the left-hand scale in the images). When the receiver is tuned to a specific radio frequency (in SSB mode), the transmitter's carrier wave will appear on the display at a point which is dependent upon the audio frequency of that wave. However, various conditions will directly affect this - notably the tuning stability of the receiver.

As an analogy, if a hypothetical radio 'mirror' were placed stationary at a point high in the sky where meteors were being observed, then what you would record would be the reflected carrier wave at it's original audio frequency of 1000Hz (in our example). In contrast, if the mirror were moving towards the receiver at a constant velocity then the signal would appear at a higher frequency. Conversely, if the mirror were receding then it would appear at a lower frequency. The degree of frequency shift would depend on the velocity at which the mirror was moving. This is the Doppler effect - the change in frequency caused by signals eminating from a moving object. On the waterfall display this would appear as a vertical shift against the left-hand scale (which is in Hz), a line moving up towards the top of the display for higher frequencies, and moving downwards for lower frequencies. The faster the mirror moves then the more extreme would appear the doppler shift. If the velocity were not constant (i.e., if the mirror was alternately moved towards and away from the receiver with a changing velocity) then you would see a corresponding curved line.

If we now substitute the mirror analogy with a real meteor ionisation trail we can see the sort of factors which can affect the appearance of the trace on the display. Movements of the carrier wave straight up and down represent changes in velocity. A line which extends below the nominal carrier frequency represents an object which is moving away from us (reduced frequency), and conversely one that is above is of higher frequency and therefore moving towards us. A long horizontal line represent little change in velocity but instead describes a persistant trail - one that continues to reflect the signal for a significant length of time (the time scale is across the bottom of the image). More often than not a real meteor trail produces a complex trace which is the result of several interacting factors, a deeply penetrating meteor will spread it's trail through multiple layers of atmosphere, and often these layers will be moving in different directions at differing velocities thus leading to elegant curved lines. These curves describe the wind speeds in the upper atmosphere, although of course the speeds involved are much lower than that of the impacting meteor. The overall brightness of the trace is analagous to it's reflectivity*angular size up to the saturation level of the reflector (i.e., that which achieves maximum signal strength). To reflect the signal most effectively the layer of plasma must be orientated in the most advantageous plane (and this will vary), whilst it's physical size (reflective area covered by the plasma trail) will also strongly influence the strength of the received signal. It can be tempting when looking at the more complex traces to believe they represent some sort of visual image of the meteors trail. This is incorrect - and it must be remembered that movement in the Y-axis is actually a change in velocity. Only the X-axis is the same as for visual observations (time - equivalent to the persistance of the trail).

Equipment:

The basic setup is a radio receiver capable of reception within the 30-70MHz band, a suitable antenna, and a computer and software to record the data. Whilst the equipment is simple it should meet certain standards stated below.

The receiver needs to be low noise (good s/n ratio) and frequency stable. It must also be capable of working in the SSB mode (single side bands, SSB, USB, CW). Most receivers sold into the amateur (HAM Radio) market meet these criteria but they also frequently offer facilities you don't need for this job. If you already have such a receiver then fine - use it! - if not then consider purchasing a model at the lower end of the price range but from those catering for the HAM enthusiast. Whilst there are cheap portable radios which sometimes offer SSB as a value-added facility (usually on their top-range models), although cheaper than a decent HAM receiver the problem here is that they may not be stable enough to remain precisely tuned on the right frequency. That will cause your display to drift up and down. By all means if you have such a receiver then give it a try, you may be pleasantly supprised. I use an iCOM PCR1000, which is a computer-controled receiver (no tuning knobs!) and appears to do a good job. I paid about Ł300 pounds for it new and it represents the most expensive bit of hardware.
The antenna needs to be sited well away from your computer to avoid interference. I use a 'desktop' Discone type (quoted as being sensitive within 25-1300MHz range), but it is placed in the loft some metres away from the receiver to reduce interference. Yagi types are also excellent provided they are sensitive at the reception frequency. A directional antenna may be of value if you are targetting a single transmitter, but as noted below an omnidirectional antenna may be able to pick up more than one transmitter at a single tuned frequency - so you can 'look' in more than one direction at once.
The computer needs to be fast enough to generate the FFT waterfall display. Any modern computer with a CPU faster than 1GHz easily meets this criteria, slower machines may be pressed into service if they can be dedicated to the task. The computer also needs enough disk space to store accumulated data, two weeks worth of data on my machine takes about 350Mb disk space - when it gets too much I just archive it onto CDROM (though with nearly 1TB of capacity I'm not that worried :-)).
The software requires a soundcard to get the audio data from the receiver. Any 16-bit Soundblaster will do the job, the only advantage of new cards is that they may offer somewhat higher sampling rates. However, I doubt even an old SoundBlaster card would represent a limitation.
The software itself, and the one I use, is a free-ware program called Spectrum Laboratory, it can be downloaded from here: http://www.qsl.net/dl4yhf/
This is a brilliant piece of programming but it can be a bit daunting upon first examination. It was not written specifically for the task we have in mind so many of the more complex features are redundant. If you need a copy of the configuration files I use (as working examples) you will find them in the links section.

Apart from various bits of connecting wire that about covers the hardware and software. One factor more difficult to control is the amount of local radio interference you may encounter. Computers (especially computer monitors) emit all kinds of radio interference and unless you are careful with placement of the receiver and antenna it will be a problem. Nevertheless, it is possible - my computer room has 3 base units, 5 monitors, and 1 portable TV plus sundry electronics (printers, etc). I find the TV is the worst offender, but one of the PCs (in a plastic case) is also bad and cannot be used when I'm making recordings. I turn most of the monitors off unless I'm actually using them. Random interference from external sources is also possible and there's not a lot you can do about it. However, even if it should be a problem initially it's usually possible choose a different radio frequency that avoids the worst of it.

Choosing a radio frequency:

This might well be the most tricky problem, mainly because you are aiming to tune your receiver to a transmitter you cannot normally hear! When I first tried it I was randomly trying different frequencies hoping meteors would appear on the screen. This is not the best way forward. Next I tried searching the 'net for frequency lists of European transmitters and suitable candidates (in the Netherlands) seemed to be located at or around 42.250.000MHz. No luck. Finally I noted that Andy Smith (see links) was using 48.249.100MHz in USB (upper side-band) mode, so I gave that a try - and it worked! From this base frequency I could identify at least 3 transmitters which were producing radio reflections at different audio frequencies and were thus separable. I still intend to experiment with other frequencies but right now it works so well I think I'll wait until a meteor shower 'quiet' spell.

Results:

Counts:

This is the 'standard' radio meteor data which counts the raw number of events recorded over a given period (usually daily or hourly). See the links section for information on IMO and RMOB online data. Although I have collected data for the 2003Geminids, and currently the 2003Ursids, I have not compiled the results yet. I'll post them here when it's done.

URSIDS 2003:

Gross Count 13min bins. This plot does not differentiate between the intensity or duration (i.e., relative brightness of individual meteor reflections), all are counted as '1'. The odd bin size (13min) is due to the width of the screendump. (Data recorded looking NE - Tx in Sweden).

Gross Count 52min bins. Same data as above, re-plotted into 52 minute bins to smooth the plot.

Differential Count 13min bins. Same data as above differentiated to show short and long duration reflections: 'Pings' (less than 5sec), '5sec' (between 5 and 20 second train) and '20sec' (train longer than 20 seconds).

Differential Count 52min bins. Again, same differentiated data but plotted as 52 minute bins.

QUADRANTIDS 2004:

This 24hr plot covers the peak of the shower (3rd to 4th January).

Using two transmitters to look in two directions simultaneously (North East and South), this 24hr plot shows the meteor activity over the night of the Quadrantid peak. The graphic at the bottom shows the height of the radiant above the horizon changing with time.

Meteor trail dopplergrams:

I have to say at the outset that much of the following is personal conjecture, and at the moment there is little supporting evidence by way of corroborative visual observation for my interpretation of these traces. Most comments are simply a descriptive (though I would like to think logical) extrapolation of what we see in the trace taking into account the possible influence of the high atmosphere (winds) and possible varied characteristics meteroic bodies themselves (mass and composition). There are a large number of unknowns: The velocity and lamination of high altitude winds, the line-of-sight effect of a meteor where the trajectory is unknown, the changing illumination angle of the plasma trail by the transmitted radio signal, the height of the radiant at the time of the observations. These (and other possible effects) can profoundly affect shape and duration of the plasma trail. It would certainly help to have a number of different viewpoints geographically spread to minimise (and possibly to define) the effects of these unknowns. Whilst there may well be alternative explanations I offer for the characteristics of the traces described below, until such an observering program is attempted and evaluated it's true value also remains unknown.

Here is my list of the most commonly observed traces I have recorded, the characteristics of which may be interpreted as described in the text. Example traces of each grouping are provided below, and clicking on the thumbnail will display more examples of that type.

A small 'dot' on the display. Likely to be caused by a very small body which did not penetrate very far into the atmosphere. The ionisation trail was of very short duration (a fraction of a second). A very large number of these are seen, indeed - out of about 1000 records (each a 13minute recording) I don't think I have seen any without at least one of these 'pings' being present. Most likely these are radio observable only - no visual counterpart. Distinguishing Characteristics: Point-like trace, no evidence of doppler shift, no persistant trail.

A short horizontal line. This is likely to be a somewhat larger mass, but again it did not penetrate very far - likely exploding almost immediately. It's (presumably) somewhat greater mass was able to produce more ionisation however, so it's trail persisted a little longer (a second or so). Given the high number recorded (compared to overall visual estimates) these would likely not be seen visually either. Distinguishing Characteristics: Single horizontal linear trace (abitrarily less than 30sec), no evidence of doppler shift.

'Tadpole', a short horizontal line with a bright spot at start. Possibly due to a medium mass body exploding immediately on entry, contains enough material to produce a plasma trail lasting several seconds. Something like a bright 'ping' with a tail. Distinguishing Characteristics: Single horizontal linear trace with the addition of a brighter 'head'. No evidence of doppler shift.

A horizontal line, but bright and lasting many seconds (possibly minutes). This must be a larger mass, and one which did not penetrate very deep into the atmosphere but exploded almost immediately. The large amount of plasma generated remained within a high altitude atmospheric layer of limited depth. Distinguishing Characteristics: Single horizontal linear trace (abitrarily greater than 30sec), no evidence of doppler shift other than the line may be broad (again arbitrarily, 10Hz bandwidth). No evidence of multiple parallel lines.

Vertical line. This is a rapidly decellerating object (the vertical spread represents a large doppler shift), but it did not explode immediately. The object penetrated some way into the atmosphere before it was completely vapourised. No horizontal spread means that again the plasma trail lasted only a brief moment. An alternative explanation may be that it's a question of angle of approach. A shallow trajectory passing directly overhead would explain the doppler effect (although from directly overhead it might not be expected to reflect the signal to my receiver?). It is also possible the body is made of something more dense/durable (possibly metallic) which facillitated seep penetration. The fainter examples of these I would not expect to be observed visually. Distinguishing Characteristics: Single vertical linear trace, marked evidence of doppler shift. No evidence of a persistant train.

A diagonal line. I don't see many of these, but it could represent a low-velocity object, the apparent slow decelleration (doppler shift X time) possibly enhanced by a flight path which is at an obtuse angle to my observation point. Whatever the reason, it was slow enough such that the velocity change took a measurable amount of time from initial entry to final vaporisation. Distinguishing Characteristics: Single diagonal linear trace.

Bright, 'L'-shaped traces - brighter at the bottom and lasting several seconds. Most likely this was a penetrating object which then exploded, the vertical part of the trace is the decelleration, the horizontal bit the plasma fireball after it exploded and resulting in the extended trace. These are fairly common and probably represent the faintest of visually observed meteors. Distinguishing Characteristics: A combination of vertical trace (beginning) followed by a single persistant horizontal train. No evidence of multiple parallel horizontal lines.

A bright (short) curved line lasting several seconds (10 or more). Probably of larger mass than the 'Pingers' above, and it's first entry would also have been a straight line on the trace. The plasma train is persistant enough to show the doppler effect of high altitude winds. It didn't penetrate very far though, most of the plasma trail remaining within a relatively short depth of atmosphere. These are perhaps of medium visual brightness (mag 4-3) based on their numbers. Distinguishing Characteristics: A continuous 'C'-shaped curve, or two (maximum) horizontal lines connected by a continuous curve at the beginning of the trace.

Complex curved and bright line lasting 10's of seconds. These can often look like curved letter 'E' where each horizontal bit represents part of the train being spread out by differing layers of high altitude winds. Similar to the previous example but more deeply penetrating. I would again expect these to be observed visually. Distinguishing Characteristics: A continuous 'E'-shaped curve, or Three (possibly up to 5) horizontal lines all connected by a continuous curve at the beginning of the trace.

Bright traces which look like 'pyramids', lasting 10 seconds or so. Possibly the originating body was of relatively large mass which progressively distintigrated (not exploding immediately) on impact, the detached pieces vapourising to form plasma as it does so. When it finally breaks up at it's deepest penetration (and slowest velocity) that is where the largest distribution of plasma occurs. An absence of fast high altitude winds must have a role to play in these particular traces. Although a fairly common characteristic they are still relatively rare (because they are fairly bright events) so I don't have many examples. I would certainly expect them to be bright visually. Distinguishing Characteristics: Compact traces lasting less than 1 minute, possibly multi-layered with strong evidence of doppler shift. Most persistant part of the trace is at the lowest frequency.

Bright traces which look like 'upside-down pyramids', lasting 10 seconds or so. These are puzzling and I can think of no easy explanation. Perhaps the originating body was of relatively large mass which distintigrated on impact, the component parts continuing with only the heaviest mass particles penetrating deeply. Not common (because they are fairly bright events) and I don't have many recordings of them either. Again, I would certainly expect them to be bright visually. Distinguishing Characteristics: Compact traces lasting less than 1 minute, possibly multi-layered with strong evidence of doppler shift. Most persistant part of the trace is at the highest frequency.

Complex linear traces, bright and lasting 10 seconds or so. Multiple linear and parallel lines, clearly several fast winds within multiple atmospheric layers are responsible, but no curves are evident. Perhaps the layers are sharply defined with highly differentiated wind velocities. Distinguishing Characteristics: Multi-layered discontinuous trace, components indicate differring doppler shifts.

Complex curvi-linear traces, bright and lasting 10 seconds or so. Presumably poorly differentiated atmospheric layers (or low wind speeds) are responsible for these seemingly random and complex curved lines. Distinguishing Characteristics: Complex multi-layered continuous trace.

Very bright complex traces, lasting several minutes. The larger the mass of the impacting object, then the more complex the resulting trace can be. A very bright trace which is a combination of all these features can just look a mess without much structure. Indeed, I have seen several examples of these, some producing traces lasting several minutes. The brightest Geminid I recorded (2003) you can see the vertical line (high doppler shift) of it's initial impact and decelleration, a subsequent explosion which then spread the plasma into multiple layers of atmosphere, and (via a second transmitter reflecting the same event) high altitude winds affecting various parts of the train. These are the spectacular visual 'fireballs' of negative magnitude. Distinguishing Characteristics: Currently no definition of 'bright' (meaning highly radio reflective), maximum saturation level may be achieved (meaning more ionised material cannot result in a stronger signal - it's already at it's maximum). Further, there is a continuous gradient from the very brightest traces back to one of the other (fainter) catagories. These traces exhibit large bandwith (arbitrarily, greater than 20Hz), persistant train (arbritrarily greater than 1 minute), and are generally without a structure that can be allocated to one of the above catagories.

If one accepts these classifications may be genuine intrinsic properties of the meteor trail (rather than entirely a result of line of sight or methodological recording artifacts), then it may be useful to analyse the distribution of such characterisations for each meteor shower. If the descriptions are consistant between repeated observations, and also between different observers, then it may offer a qualitative as well as quantitative method of classifying each meteor shower. Of course, this is already done for visual observations - typical meteors of one particular shower may be described as 'fast' 'bright' 'faint' 'of particular colour' or (notably) 'leaving persistant trains', but these tend to be simple generalisations. Dopplergrams obtained by radio reflection would provide some additional qualitative data to support these descriptions.

If the above classification system appears too arbitrary, and the distinctions between different classes not well defined, there are still some quantitative measures that would be available from the same dataset:

Population (raw counts)
Area under curve (total radio brightness)
Duration (above noise level)
Peak reflectivity (below the saturation level)

This information would serve to define the timecourse of each meteor shower, and provide distribution data on the relative masses of the component bodies in the stream. Whether the extent of the doppler shift could be included (i.e., some measure of the impact velocity) I'm not sure - certainly more than one observer would be needed with the observations being correlated to remove line-of-sight artifacts.

Summary:

I think what is most clear from the information already available is that this is an area of amateur astronomy worthy of much more attention. Simple counts of radio events provide valuable data on the numbers of meteoric strikes, and from this datathe timing and extent of known showers is known. From this data, orbital determinations can be calculated and a subsequent projection may identify the parent body. Analysis of the dopplergrams described above may add more information - population distributions based on classification of meteor plasma trains. Different meteor streams would be expected to yield different distributions, and this in turn may help in understanding their origin and also the physical properties of the parent object.

An even more ambitious project could combine multiple observations of the same object - possibly enabling construction of a 3-D image of the train by interferometry. This would significantly aid any interpretation of individual events. Such data may be generated either using mutiple receivers (at different locations) tuned to a single transmitter illuminating the target, or by mutiple receivers (at the same location) tuned to two or more transmitters both of which illuminate the target. The former is probably the easiest and more controlable route.

Links:

http://www.tvcomm.pwp.blueyonder.co.uk/G7IZU_Radio_Pages/Radio/index.html The Detection of Sporadic E, Aurora and Meteors by using Radio Spectrum Analysis - Andy Smith G7IZU UK

http://www.radiodesignservices.co.uk/meteor.shtml Radio Meteor Observations - Dave Swan UK

http://ap.stmarys.ca/~lonc/meteors.html Meteors by Radio - Ian McCarthy

http://aio.arc.nasa.gov/~leonid/Global-MS-Net/ Global Meteor-Scatter Network (Global-MS-Net) - NASA

http://www.rmob.org/ Radio Meteor Observing Bulletin (RMOB)

http://www.imo.net/ The IMO (International Meteor Organisation)

http://www.qsl.net/dl4yhf/spectra1.html Audio Spectrum Analyzer ("Spectrum Lab") software - DL4YHF

http://sapp.telepac.pt/coaa/r_meteor.htm R_Meteor software (shareware)

http://www.bnv-bamberg.de/home/ba0249/mop/main.html MOP (Meteor Observation Project). software

Back to Telescope Accessories page

	Gross Count 13min bins. This plot does not differentiate between the intensity or duration (i.e., relative brightness of individual meteor reflections), all are counted as '1'. The odd bin size (13min) is due to the width of the screendump. (Data recorded looking NE - Tx in Sweden).
	Gross Count 52min bins. Same data as above, re-plotted into 52 minute bins to smooth the plot.
	Differential Count 13min bins. Same data as above differentiated to show short and long duration reflections: 'Pings' (less than 5sec), '5sec' (between 5 and 20 second train) and '20sec' (train longer than 20 seconds).
	Differential Count 52min bins. Again, same differentiated data but plotted as 52 minute bins.

	A small 'dot' on the display. Likely to be caused by a very small body which did not penetrate very far into the atmosphere. The ionisation trail was of very short duration (a fraction of a second). A very large number of these are seen, indeed - out of about 1000 records (each a 13minute recording) I don't think I have seen any without at least one of these 'pings' being present. Most likely these are radio observable only - no visual counterpart.	Distinguishing Characteristics: Point-like trace, no evidence of doppler shift, no persistant trail.
	A short horizontal line. This is likely to be a somewhat larger mass, but again it did not penetrate very far - likely exploding almost immediately. It's (presumably) somewhat greater mass was able to produce more ionisation however, so it's trail persisted a little longer (a second or so). Given the high number recorded (compared to overall visual estimates) these would likely not be seen visually either.	Distinguishing Characteristics: Single horizontal linear trace (abitrarily less than 30sec), no evidence of doppler shift.
	'Tadpole', a short horizontal line with a bright spot at start. Possibly due to a medium mass body exploding immediately on entry, contains enough material to produce a plasma trail lasting several seconds. Something like a bright 'ping' with a tail.	Distinguishing Characteristics: Single horizontal linear trace with the addition of a brighter 'head'. No evidence of doppler shift.
	A horizontal line, but bright and lasting many seconds (possibly minutes). This must be a larger mass, and one which did not penetrate very deep into the atmosphere but exploded almost immediately. The large amount of plasma generated remained within a high altitude atmospheric layer of limited depth.	Distinguishing Characteristics: Single horizontal linear trace (abitrarily greater than 30sec), no evidence of doppler shift other than the line may be broad (again arbitrarily, 10Hz bandwidth). No evidence of multiple parallel lines.
	Vertical line. This is a rapidly decellerating object (the vertical spread represents a large doppler shift), but it did not explode immediately. The object penetrated some way into the atmosphere before it was completely vapourised. No horizontal spread means that again the plasma trail lasted only a brief moment. An alternative explanation may be that it's a question of angle of approach. A shallow trajectory passing directly overhead would explain the doppler effect (although from directly overhead it might not be expected to reflect the signal to my receiver?). It is also possible the body is made of something more dense/durable (possibly metallic) which facillitated seep penetration. The fainter examples of these I would not expect to be observed visually.	Distinguishing Characteristics: Single vertical linear trace, marked evidence of doppler shift. No evidence of a persistant train.
	A diagonal line. I don't see many of these, but it could represent a low-velocity object, the apparent slow decelleration (doppler shift X time) possibly enhanced by a flight path which is at an obtuse angle to my observation point. Whatever the reason, it was slow enough such that the velocity change took a measurable amount of time from initial entry to final vaporisation.	Distinguishing Characteristics: Single diagonal linear trace.
	Bright, 'L'-shaped traces - brighter at the bottom and lasting several seconds. Most likely this was a penetrating object which then exploded, the vertical part of the trace is the decelleration, the horizontal bit the plasma fireball after it exploded and resulting in the extended trace. These are fairly common and probably represent the faintest of visually observed meteors.	Distinguishing Characteristics: A combination of vertical trace (beginning) followed by a single persistant horizontal train. No evidence of multiple parallel horizontal lines.
	A bright (short) curved line lasting several seconds (10 or more). Probably of larger mass than the 'Pingers' above, and it's first entry would also have been a straight line on the trace. The plasma train is persistant enough to show the doppler effect of high altitude winds. It didn't penetrate very far though, most of the plasma trail remaining within a relatively short depth of atmosphere. These are perhaps of medium visual brightness (mag 4-3) based on their numbers.	Distinguishing Characteristics: A continuous 'C'-shaped curve, or two (maximum) horizontal lines connected by a continuous curve at the beginning of the trace.
	Complex curved and bright line lasting 10's of seconds. These can often look like curved letter 'E' where each horizontal bit represents part of the train being spread out by differing layers of high altitude winds. Similar to the previous example but more deeply penetrating. I would again expect these to be observed visually.	Distinguishing Characteristics: A continuous 'E'-shaped curve, or Three (possibly up to 5) horizontal lines all connected by a continuous curve at the beginning of the trace.
	Bright traces which look like 'pyramids', lasting 10 seconds or so. Possibly the originating body was of relatively large mass which progressively distintigrated (not exploding immediately) on impact, the detached pieces vapourising to form plasma as it does so. When it finally breaks up at it's deepest penetration (and slowest velocity) that is where the largest distribution of plasma occurs. An absence of fast high altitude winds must have a role to play in these particular traces. Although a fairly common characteristic they are still relatively rare (because they are fairly bright events) so I don't have many examples. I would certainly expect them to be bright visually.	Distinguishing Characteristics: Compact traces lasting less than 1 minute, possibly multi-layered with strong evidence of doppler shift. Most persistant part of the trace is at the lowest frequency.
	Bright traces which look like 'upside-down pyramids', lasting 10 seconds or so. These are puzzling and I can think of no easy explanation. Perhaps the originating body was of relatively large mass which distintigrated on impact, the component parts continuing with only the heaviest mass particles penetrating deeply. Not common (because they are fairly bright events) and I don't have many recordings of them either. Again, I would certainly expect them to be bright visually.	Distinguishing Characteristics: Compact traces lasting less than 1 minute, possibly multi-layered with strong evidence of doppler shift. Most persistant part of the trace is at the highest frequency.
	Complex linear traces, bright and lasting 10 seconds or so. Multiple linear and parallel lines, clearly several fast winds within multiple atmospheric layers are responsible, but no curves are evident. Perhaps the layers are sharply defined with highly differentiated wind velocities.	Distinguishing Characteristics: Multi-layered discontinuous trace, components indicate differring doppler shifts.
	Complex curvi-linear traces, bright and lasting 10 seconds or so. Presumably poorly differentiated atmospheric layers (or low wind speeds) are responsible for these seemingly random and complex curved lines.	Distinguishing Characteristics: Complex multi-layered continuous trace.
	Very bright complex traces, lasting several minutes. The larger the mass of the impacting object, then the more complex the resulting trace can be. A very bright trace which is a combination of all these features can just look a mess without much structure. Indeed, I have seen several examples of these, some producing traces lasting several minutes. The brightest Geminid I recorded (2003) you can see the vertical line (high doppler shift) of it's initial impact and decelleration, a subsequent explosion which then spread the plasma into multiple layers of atmosphere, and (via a second transmitter reflecting the same event) high altitude winds affecting various parts of the train. These are the spectacular visual 'fireballs' of negative magnitude.	Distinguishing Characteristics: Currently no definition of 'bright' (meaning highly radio reflective), maximum saturation level may be achieved (meaning more ionised material cannot result in a stronger signal - it's already at it's maximum). Further, there is a continuous gradient from the very brightest traces back to one of the other (fainter) catagories. These traces exhibit large bandwith (arbitrarily, greater than 20Hz), persistant train (arbritrarily greater than 1 minute), and are generally without a structure that can be allocated to one of the above catagories.