Meteors and passive radar

[RJN]

Hi Everyone,

I decided to continue this discussion on the NSL bulletin board so that important parts can be archived and more easily accessed in the future if and when needed. It is part of creating "institutional knowledge" on which the project is dependant.

Noah, in your latest email, you said:

I guess what you mean by passive radar is just over-the-horizon propagation of signal from distant stations that are reflected by meteor trails. This is what radio hams call DX contacts. Unfortunately, these detections that can be achieved with about 1k$ worth of equipment do not teach one enough about the meteor. This is because the only information one has is that a meteor happened at a certain instance and that the reflection was at a certain wavelength.

In order to do some physics we must have information at least about the exact location of the meteor (longitude, latitude, altitude). The way to do this passively is to use at least three receivers, well separated, and to correlate their received signals. The technique is called, if I am not mistaken, "TDOA=time difference of arrival" and is fairly common in electronic warfare. Each pair of receivers provides a possible locus that is a point on a hyperbola (if I am not mistaken) and the reflection is at the intersection of the three hyperbolas. One needs at least three receivers because one deals with 3D localization since the meteor plasma can be at any altitude.

I am probably stating the obvious, but I was thinking that each passive radar stations would be somehow physically coupled with the optical concam stations. If they can fit inside the CONCAM box, great, but I would guess a long antenna would not. Anyway, the idea would be to get not only radar information but coincident optical trails as well. My hope is that radar information and optical trail information complement each other.

In the ideal case, three optical concams and their three corresponding simultaneous radar detections would occur for a given bright meteor. Given a multi-pixel meteor trail, that should give significant orbital information. Do we still need an optical chopper in this case? I am not sure!

The the single station case, a single optical streak would be paired with a single radar echo. I don't think we get an orbit from that. I am not sure about the two station case. It would be fun to work that out, if it has not been worked out already.

- Bob

[nbrosch]

Bob, my guess is that there is no way to include a passive radar system in a CONCAM box. At most, one could fit a simple VHF receiver; these exist in really small sizes with only the antenna having a sizable dimension.

However, once we start taking TDOF, interferometric systems, etc., both the receiving elements and the eelctronics get complicated. These would not be the table-top devices that CONCAM-1s are but more like refrigerator-size devices.

Noah

[mgmeyer]

Hi Noah;

Nice to hear from you.
Actually, the passive radar I have been working on is not over-the-horizon. It's a line-of-sight VHF system -- which is currently implemented as a bistatic device, with the receivers separated by a few hundred km. The illumination is commercial FM radio stations, with one receiver dedicated to obtaining the reference illumination, and the other for collecting scattered signals. TDOA-type measurements are possible; I have used multiple FM transmitters to detect point targets (airplanes, and also a meteor once) and used that to estimate the target location. The scatter can be received in a single location so long as the transmitters are spaced enough apart (sort of the opposite of what you described with listening to the scatter from a single station in several different locations).

I've also done a good amount of work with interferometry and imaging. I think that would be the way to go for finding angles of arrival for meteor scatter. It would require multiple antennas (3 or 5) (or else we have to work hard at deconvolving the mutual coupling of the antennas to put them close together).


The HF frequencies that the hams use for OTH DXing *are* more useful for detecting meteors. The radar meteor folks (and radio amateurs interested in meteor bounces) split meteor events into two categories -- those with underdense and those with overdense trails. The threshold here is I believe about 10^14 free electrons generated per meter of trail (I suppose the overdense trails come from bigger meteors). At HF you can see the underdense trails, but as the frequency climbs, of course the lower density trails become invisible. (The HF "SuperDARN" radar chain at the north and south poles sees tons of meteors, and has to work pretty hard to see through the clutter they cause! While my system, at VHF, only sees an occasional blip.)

All of this is in reference to "specular-like" reflections from the plasma of meteor trails, but there is another scattering mechanism that can allow radars to detect meteor ablation, and that is Bragg scatter from density irregularities in the trail plasma. These density waves/turbulence result from instabilities like the Kelvin-Helmholtz and density gradients on the edges of the plasma tube. At VHF frequencies, I suspect the majority of our meteor detections are due to this effect. But we'll see specular reflections from overdense meteor trails that are oriented so that their long dimension is perpendicular to the radar line of sight. (Only the really powerful radars, like arecibo, altair, etc can detect "head echoes" - scatter from the meteor itself.)

A complication is that while HF is better for detecting meteors (any system serious about monitoring meteor traffic should probably be HF), there is a lack of good "free" illumination at HF to use in a passive radar system. Plus, atmospheric refraction effects come into play and will have to be dealt with.

The analog TV channels are around 60-70 MHz... still VHF, but lower at least. I'm not sure how useful they'd be for meteor trails. But they don't make great radar waveforms, because of a horizontal scan sync pulse, which introduces a range ambiguity of (I believe) 8 km.

Satellites are a possible illumination source, although their signal strength tends to be low and their frequencies high.
I'm not familiar with the GLONASS/Galileo satellites, but GPS is at an even higher frequency, which will sail right through the vast majority of meteor trails, and the Bragg scattering cross section for the plasma density irregularities also decreases with frequency.
There are many people who use differential GPS measurements to do ionospheric tomography, determining integrated electron density (from GPS altitude (20000km) to ground). I'm not sure if the GPS signals would be useful for meteor trail detection, but I suspect not.

Given that we could find some useful illumination: I had not been thinking of a TDOA-like approach, but that could be useful, either as a primary detection scheme or as validation. Currently my passive radar setup is bistatic - requiring two spatially-separated receivers to implement the full radar system - so adding a third and doing TDOA for point targets would be feasible. I have been experimenting with distributed multistatic passive radar as a tool for studying space weather-related effects, and I'm very interested in additional uses for sprinkling these receivers around the globe. The more science we can get out of these instruments, the better.

Not sure if you've seen this stuff, but here are a couple references on deducing orbits of meteors ((what is the difference between a meteor and a meteoroid?)) from radar AOA information:
Baggaley et al., 1993 (http://adsabs.harvard.edu/abs/1993mtpb.conf..245B)
Morton and Jones, 1982 (http://adsabs.harvard.edu/abs/1982MNRAS.198..737M)

Thanks again for involving me!
Cheers,
Melissa

Noah Wrote: I guess what you mean by passive radar is just over-the-horizon propagation of signal from distant stations that are reflected by meteor trails. This is what radio hams call DX contacts. Unfortunately, these detections that can be achieved with about 1k$ worth of equipment do not teach one enough about the meteor. This is because the only information one has is that a meteor happened at a certain instance and that the reflection was at a certain wavelength.

In order to do some physics we must have information at least about the exact location of the meteor (longitude, latitude, altitude). The way to do this passively is to use at least three receivers, well separated, and to correlate their received signals. The technique is called, if I am not mistaken, "TDOA=time difference of arrival" and is fairly common in electronic warfare. Each pair of receivers provides a possible locus that is a point on a hyperbola (if I am not mistaken) and the reflection is at the intersection of the three hyperbolas. One needs at least three receivers because one deals with 3D localization since the meteor plasma can be at any altitude.

Ideally, one would want to have systems for different wavelengths, since the character of the plasma, which is one of the measurables, should be studied at different lambdas.
A more ambitious project would use as the illuminator signals from the GPS/GLONASS/Galileo constellations. At any time there are a number of satellites overhead and their signals propagate on straight lines, but a small fraction reflects from the meteor trail. The problem is orders of magnitude more complicated than ground-based TDA since the transmitters move continuously on their orbits, there are signal propagation problems at low altitudes, etc.

[nbrosch]

Melissa:

[quote="mgmeyer"]
Actually, the passive radar I have been working on is not over-the-horizon. It's a line-of-sight VHF system -- which is currently implemented as a bistatic device, with the receivers separated by a few hundred km. The illumination is commercial FM radio stations, with one receiver dedicated to obtaining the reference illumination, and the other for collecting scattered signals. TDOA-type measurements are possible; I have used multiple FM transmitters to detect point targets (airplanes, and also a meteor once) and used that to estimate the target location. The scatter can be received in a single location so long as the transmitters are spaced enough apart (sort of the opposite of what you described with listening to the scatter from a single station in several different locations).

I've also done a good amount of work with interferometry and imaging. I think that would be the way to go for finding angles of arrival for meteor scatter. It would require multiple antennas (3 or 5) (or else we have to work hard at deconvolving the mutual coupling of the antennas to put them close together).

[b]I guess that I agree with you. We would need interferometry and keeping track of the phases to find exactly where a metaor happened.[/b]


The HF frequencies that the hams use for OTH DXing *are* more useful for detecting meteors. The radar meteor folks (and radio amateurs interested in meteor bounces) split meteor events into two categories -- those with underdense and those with overdense trails. The threshold here is I believe about 10^14 free electrons generated per meter of trail (I suppose the overdense trails come from bigger meteors). At HF you can see the underdense trails, but as the frequency climbs, of course the lower density trails become invisible. (The HF "SuperDARN" radar chain at the north and south poles sees tons of meteors, and has to work pretty hard to see through the clutter they cause! While my system, at VHF, only sees an occasional blip.)

[b]I think that the line density is two orders of magnitude lower, if I remember correctly.[/b]

All of this is in reference to "specular-like" reflections from the plasma of meteor trails, but there is another scattering mechanism that can allow radars to detect meteor ablation, and that is Bragg scatter from density irregularities in the trail plasma. These density waves/turbulence result from instabilities like the Kelvin-Helmholtz and density gradients on the edges of the plasma tube. At VHF frequencies, I suspect the majority of our meteor detections are due to this effect. But we'll see specular reflections from overdense meteor trails that are oriented so that their long dimension is perpendicular to the radar line of sight. (Only the really powerful radars, like arecibo, altair, etc can detect "head echoes" - scatter from the meteor itself.)

[b]One thing we found with a large L-band phased array was that during meteor showers we detect echoes from 200km+ altitudes. This has been partly confirmed by unpublished EISCAT observations. Our interpretation is that there we see plasma produced by sputtering, since the atmospheric density is too low to support classical ablation.[/b]

A complication is that while HF is better for detecting meteors (any system serious about monitoring meteor traffic should probably be HF), there is a lack of good "free" illumination at HF to use in a passive radar system. Plus, atmospheric refraction effects come into play and will have to be dealt with.

The analog TV channels are around 60-70 MHz... still VHF, but lower at least. I'm not sure how useful they'd be for meteor trails. But they don't make great radar waveforms, because of a horizontal scan sync pulse, which introduces a range ambiguity of (I believe) 8 km.

[b]There are, in principle, two research avenues one could follow. The first is to concentrate on "classical" meteors, those between 80 and 130 km, and measure them both with optical means (modified CONCAMs) and with a passive radar that would provide accurate plasma parameters such as density and time development. The science would come from correlating the optical with the plasma diagnostics and these would then be input into an ablation model that would have to have shock wave heating, different materials, rotation of an irregular body, etc. The second would be to concentrate on the high altitude exotic meteor component, detect and measure the plasma parameters, and make a serious attempt to increase the number and quality of the optical detections at these altitudes.[/b]

Satellites are a possible illumination source, although their signal strength tends to be low and their frequencies high.
I'm not familiar with the GLONASS/Galileo satellites, but GPS is at an even higher frequency, which will sail right through the vast majority of meteor trails, and the Bragg scattering cross section for the plasma density irregularities also decreases with frequency.
There are many people who use differential GPS measurements to do ionospheric tomography, determining integrated electron density (from GPS altitude (20000km) to ground). I'm not sure if the GPS signals would be useful for meteor trail detection, but I suspect not.

Given that we could find some useful illumination: I had not been thinking of a TDOA-like approach, but that could be useful, either as a primary detection scheme or as validation. Currently my passive radar setup is bistatic - requiring two spatially-separated receivers to implement the full radar system - so adding a third and doing TDOA for point targets would be feasible. I have been experimenting with distributed multistatic passive radar as a tool for studying space weather-related effects, and I'm very interested in additional uses for sprinkling these receivers around the globe. The more science we can get out of these instruments, the better.

Not sure if you've seen this stuff, but here are a couple references on deducing orbits of meteors ((what is the difference between a meteor and a meteoroid?)) from radar AOA information:
Baggaley et al., 1993 ([url]http://adsabs.harvard.edu/abs/1993mtpb.conf..245B[/url])
Morton and Jones, 1982 ([url]http://adsabs.harvard.edu/abs/1982MNRAS.198..737M[/url])

[b]I have met Jack Baggaley and I know Peter Brown and Margaret Campbell-Brown, so I am familiar with their papers. Thanks for the refs, though.

Noah[/b]

[RJN]

I find myself again wondering if (optical) CONCAMs can really contribute to meteor science. Given an impressive enough radar array, won't the meteor orbit be completely specified without the help of a CONCAM?

I guess the niche I see for optical CONCAMs is therefore essentially financial. For perhaps one meteor a night (if that!), optical CONCAMs can get information that it would otherwise be expensive for radar arrays to get. Since CONCAMs will typically see, say, one meteor a night, it seems that whole radar arrays capable of seeing many more meteors would not benefit from the optical CONCAM images. That is why I saw a good fit between passive FM radar and CONCAM. But I don't really know the numbers: could passive FM radar see one meteor a night? Would this meteor be identifiable as the bright meteor streak that the optical CONCAM(s) saw?

I have six CONCAM1s that can be deployed relatively cheaply, for almost nothing if we don't include a chopper. If crude passive FM radar setups are also very cheap, we may be able to deploy a meteor array in very short order for very little money. Would this generate useful science? Or will the useful science only be uncovered if we shoot for the big time, creating powerful CONCAM4s coupled with and arrays of HF radar? At what minimum budget does the meteor science get out?

- RJN

[nbrosch]

Bob, I do not think that with a meteor per night one could do much science. The passive radar, if it performs as well as regular radars, should yield hundreds of meteors per night but it would be somewhere within a few 100-km from the station. There would be no way of correlating optical with radar since the timing of he optical meteor would be the cadence of the camera.

The least one would need would be a meteor radar coupled with a number of optical cameras equipped with choppers, then go for an optical-radar correlation to evaluate the optical efficiency as well as the ablation parameters.

[quote="RJN"]I find myself again wondering if (optical) CONCAMs can really contribute to meteor science. Given an impressive enough radar array, won't the meteor orbit be completely specified without the help of a CONCAM?

I guess the niche I see for optical CONCAMs is therefore essentially financial. For perhaps one meteor a night (if that!), optical CONCAMs can get information that it would otherwise be expensive for radar arrays to get. Since CONCAMs will typically see, say, one meteor a night, it seems that whole radar arrays capable of seeing many more meteors would not benefit from the optical CONCAM images. That is why I saw a good fit between passive FM radar and CONCAM. But I don't really know the numbers: could passive FM radar see one meteor a night? Would this meteor be identifiable as the bright meteor streak that the optical CONCAM(s) saw?

I have six CONCAM1s that can be deployed relatively cheaply, for almost nothing if we don't include a chopper. If crude passive FM radar setups are also very cheap, we may be able to deploy a meteor array in very short order for very little money. Would this generate useful science? Or will the useful science only be uncovered if we shoot for the big time, creating powerful CONCAM4s coupled with and arrays of HF radar? At what minimum budget does the meteor science get out?

- RJN[/quote]

[RJN]

Noah,

Forgive me for being argumentative this post. I have gotten this far in wondering if optical CONCAMs can do valuable meteor science partly because there were positive indications from Peter and you. Why would you be developing optical CONCAM4s for use in Antarctica if they would not help generate useful meteor science? Wouldn't those CONCAM4s likely see only a few meteors a night? Why not just deploy a more impressive radar array?

Next, Peter Jenniskens replied in a recent email (copied to you) that

You asked for a scientific rationale on why one would want to derive pre-atmospheric orbits of bright meteors in light of the many thousands of orbits measured by radar by Jack Baggaley (and others). The answer is that Jack is looking mostly at sporadic meteors part of the "meteoroids" peak in the mass influx curve. Only 1-2% of his meteors are part of meteor showers and those he has to fish out of a strong background. In contrast, if you measure orbits of bright meteors (+1 to -12 magnitude), you will sample exclusively meteor showers (with a small component of meteorite dropping fireballs).

Most are ecliptic streams that are widely dispersed. Precise orbits are needed to discriminate those streams. The detection and discovery of streams has become important now more and more NEO are found to be dormant comets and parents of our meteor showers. Each shower provides a historic record on when the comet broke. The more precise an orbit can be measured, the further back in time one can integrate and the better the stream stands out from other streams.

The implication is that optical CONCAMs can make a significant contribute to the orbits of the meteors they detect and hence to meteor science. Is this correct? Is this unique enough or good enough justification to deploy optical CONCAMs primarily for meteor detection?

- RJN

[RJN]

Some more thoughts about this. It seems it is all a question of the magnitude of the errors. I don't know much about radar detections of meteors, but it seems to me that radar is non-imaging. One would need several simultaneous radar measurements from separated locations to determine the 3D location of the meteor and its velocity vector. I don't know how accurate these would be. If they are more accurate than the optical CONCAM positions, then I just don't see how optical CONCAMs can be useful.

However, it seems to me an implication from both Noah and Peter that an optical trail would somehow give more accurate positional and directional information than radar, for the few meteors that are caught optically. I guess it all boils down to that: how true is this? What are the positional and orbital magnitudes of the radar errors versus the optical errors? If it is true, does the decrease in 3D location errors created by optical CONCAMs allow us to cross any important science thresholds?

The Night Sky Live project has built up significant infrastructure that can be used to track meteors in the optical. Is leveraging this infrastructure useful for meteor science? I know I am sounding repetitive, but that is the key question for me. If yes, I will try to leverage this infrastructure to bring out that science. If no, than this is just a waste of time for everyone, and we would all do better forgetting trying to use optical CONCAMs for meteor science and spending our time during something more scientifically productive.

This leads to the question: what are the positional errors with a CONCAM? A CONCAM pixel is about 10 arcminutes. Assume a meteor distance of 100 km. Then each pixel gives a meteor position accurate to about 0.3 kilometers. Let's say a typical meteor trail runs 10 pixels. The trajectory error would then be known to about 1/10 radians or about 6 degrees. How does that compare with radar accuracy? Can that accuracy resolve streams internal to known meteor showers?

- RJN