Infrasound Detection of Meteors

[sheeny]

Attached are my calculations for an Infrasound detection of a meteors yesterday with my Raspberry Shake and Boom.


Pity there aren't more Raspberry Booms close by to calculate trajectories...


The one at 04:07:46UTC is a medium speed class I + meteor. The + indicates a positive phase i.e. the first part of the wave is a positive pressure spike. My interpretation of this (I haven't yet found a reference that describes this) is that the meteor is generally approaching the observer so the positive pressure wave arrives first.


The one at 15:37:41UTC is a fast class I- meteor. In this case the phase is negative so I interpret this as it traveling away from the observer so the negative pressure wave arrives first.


Al.

[sheeny]

Another meteor detection last night 2/3/22 UTC.

This one appears to have been quite close due to the shape of the wave.

The waveform shows some remnants of the near field wave form shape i.e the positive phase has higher amplitude and lower period than the trailing negative phase. (refer to the second picture taken from "Numeric Prediction of Meteoric Infrasound Signatures" by Marian Nemec, Michael J.Aftmodis, and Peter G. Brown).


Al.

[Shiraz]

really interesting reading Al. Thanks for posting the info and reference.

For interest, did the seismo side of your system have enough sensitivity to see the recent 3.7 Adelaide earthquake? https://earthquakes.ga.gov.au/

cheers Ray

[sheeny]

G'Day Ray,


The Shake probably was capable of detecting it, but as the installation is not ideal (it's in the house on a floating floor as I failed to get it to connect to the network in the observatory) I didn't actually detect it.


Through ShakeNet all the other Raspberry Shakes are accessible, so I downloaded the signals from a Shake at Mt Elliot near Mt Barker. I also checked Shakes at Broken Hill and Horsham, but they did not detect it either, so not much chance of me getting it here. M3.8 is pretty small.


Al.

[sheeny]

Here's today's effort. I found a couple of meteors in yesterday's infrasound trace.


The one at 07:09:36 UTC I believe entered the atmosphere past me (in the direction of travel) hence the negative phase.


The one at 04:05:57 UTC is very slow. This is because 04:07 UTC is 15:07 AEDT so at this time of the afternoon both the motion of the earth around the sun and the rotation of the earth on it's axis are working to slow the relative speed of the meteor. Meteors are rarely detected this time of day visually so it appears slower than "normal".


Slow meteors don't punch such a big hole in the atmosphere so the dominant signal period is less.


Also of interest in this wave form is the apparent change of phase after the first cycle. Purely speculation on my part, but I wonder if this indicates the point of closest approach to the observer - i.e. towards until that point and then further away after.


The sub classification "a" means there is a decaying oscillation following the primary signal.


Al.

[sheeny]

I managed to detect a slow Class IIa- meteor yesterday. Calcs attached FYI.


Al.

[sheeny]

The meteor identification and classification method I have been using on my Raspberry Shake and Boom is based largely on this paper: https://arxiv.org/ftp/arxiv/papers/1407/1407.6331.pdf


The three main diagrams/tables from this paper that I use are attached FYI.


Obviously the best way to identify meteors is to be able to confirm them visually, but part of the attraction of the RBoom (Raspberry Boom) is to be able to identify meteors when they can't be seen e.g. daylight, cloudy, or too lazy :P.


So for now I have been identifying potential meteors by:
1. looking for the expected wave shape(s) in the infrasound recording;
2. eliminating any that coincide with a seismic signal in the RShake;
3. checking the Dominant Signal Period is in the right range; and
4. the Frequency Spectrogram is strong below 10Hz and relatively weak above.


Strong Infrasound signals can create a linked seismic signal (and vice versa) so by ignoring signals with coincident seismic signals I am probably losing some strong meteor signatures, but I am also eliminating any seismic sources that may just produce a similar signature in infrasound as a meteor would. If I had a confirming visual photographic observation of the meteor that matched the timing of the signal (allowing for the speed of sound) then the signature would be confirmed and the seismic signal could be identified as being linked and produced by the infrasound.


I hope this helps anyone interested in infrasound detection of meteors get a start in doing so.


:)


Al.

[sheeny]

Found and interesting potential meteor this morning.


A Slow Class Ia- with a complex tail.


A bit of analysis with low and band pass filters suggests it is possibly 2 bodies: one a Class IIa- and the other a class IIa+ or * (* is indeterminate).


The Dominant Signal Period (and hence the Dominant Signal Frequency) is a function of the particle size, not just it's speed. The DSP is determined by the time taken to close the vacuum behind the super sonic particle, so both speed a size have an effect. Given that 2 particles travelling together will have the same speed, their DSPs will reflect their relative sizes.


In this case the larger particle (Class IIa-) had a DSP of 0.414s while the smaller particle had a DSP of 0.221s.


Al

[sheeny]

I haven't detected many meteors lately, partly due to slackness on my part but when I have looked I haven't seen many likely signals.


I got a good one this morning though. A complex medium sized Class III c or d +.


Further analysis split this into a pair of meteors traveling together: a medium size Class III + and a small Class II +.


There was a small seismic signal around 18 to 22Hz detected on the Raspberry Shake, but as the start of this coincided with the end of the first half wave of the infrasound, I interpreted this as the seismograph being excited by the infrasound rather than the other way around.


Al.

[gary]

Hi Al,

This is really fantastic! :)

Best Regards

Gary

[sheeny]

Thanks Gary. I've just learned a lesson last night. It's hard to detect meteors with infrasound when it rains! Rain makes a lot of infrasound.


Al.

[sheeny]

The RS&B detected a large meteor this morning at 6:13am. It's a large Class I -.


As the first phase of the signal is negative I interpret this as the meaning that at the time the meteor entered the atmosphere it was heading away from the observing station. That's the only scenario I can come up with that allows the negative pressure from the vacuum behind the particle to reach the observer first.


Also note that the second half of the N wave is lower in amplitude and longer period than the first half so there are still remnants of the near field pressure profile evident - so it must have been relatively close. The signal has not had sufficient atmosphere to attenuate to a balanced N wave.


Al.

[sheeny]

Latest infrasound meteor detection from 4:46pm today.


a small class I b +.


Because the waveform is messy, this suggests more than one particle in the "meteor".


The spectrogram shows two clouds, so this gives a clue to suitable frequencies to resolve the components of the meteor using band pass filters.


This one resolved to a small class II+ followed by a small class IV + about 0.12 seconds behind. They took about 0.78 s and 0.73 s respectively to ablate or slow to below the speed of sound.


Al.

[AstroViking]

I never would have thought of using sound to detect meteors.

What a great idea - and kudos to you for sifting through all that data to find the good bits!

[sheeny]

Another complex meteor detected this morning. I was able to split it into two particles, but suspect the smaller of the "two" is actually a pair of similar sized particles itself, but I'm not able to resolve it with bandpass filters.


Al.

[sheeny]

Got a simple one today. Still ran it through a band pass filter and the N wave came up nice and clean and balanced without the noise.


Al.

[sheeny]

So far I've been concentrating on finding meteor signals and then analysing them, but this is a very slow process.


I've cooked up a process to speed up detection, which I have just started using.


First I tile the EHZ (seismic) and HDF (infrasound) helicorders from the RS&B side by side in SWARM software.


Then I set up a 0.5 Hz to 10Hz bandpass filter on the infrasound channel to eliminate as much noise as possible that doesn't relate to meteors. In the past I've been scanning the raw signal for possible meteors, so I think this should make more meteors detectible.


Then I set the zoom time span to about 15 minutes so I can scan each 30 minute line on the helicorder in two goes, and I look for small sharp spikes which are equally balanced above and below zero, and note the time to the nearest minute. I come back to that later. The first screen shot below shows what this looks like.


After scanning as much of the helicorder as I want to, I then come back to each of the recorded times and zoom in to confirm whether or not it is a potential meteor or not (i.e. N wave form, first cycle biggest, etc and no seismic trigger). The second screenshot below shows what that looks like for a successful detection. Note that there is a small seismic signal that coincides with the N wave in infrasound, but as this signal is small, and there are other seismic signals of similar size and frequency that do not trigger an infrasound signal, there's no reason to believe this one has triggered this N wave.


As we had rain last night, conditions were far from ideal, as rain produces a lot of infrasound. Despite this in 9 hours of data I was able to detect 24 possible candidates and then eliminate 4 that were definitely not meteors, and 14 that are very likely meteors. The remaining 5 might also be meteors as well but the signals were so weak they were hard to separate from the noise.


The sporadic meteor rate is usually 5 to 10 per hour detected visually in ideal conditions. In not ideal conditions with rain falling, the RS&B achieved 1.5 meteors per hour (14 in 9 hours) or about 15 to 30% of ideal visual rate.


This process needs more testing, but I'm encouraged that my RS&B doesn't seem as deaf to meteors as first thought.

[sheeny]

I found an interesting signal today which at first glance doesn't fit the classic N wave model for a meteor signature, however, it's nature is such that I could see it was made up of N waves and so is more likely to be a meteor than anything else.


One of the things that has been a puzzle about the meteor classifications is what constitutes a meteor of indeterminate phase (* in the classification)? Over the last few weeks I've been looking at and processing quite a few meteor signals and slowly developing an idea of what an indeterminate phase signal is. Processing this signal has definitely helped my understanding. I won't repeat myself here, please read the notes in the screenshots.


Al.

[sheeny]

Two more meteor detections and analyses found in amongst last nights storm.


Al.

[AdamJL]

Fascinating thread!!

[sheeny]

Here is a summary of how I interpret the infrasound signals of meteors. Consider it like any hypothesis - it's the best I have till it's disproved.


Al.

[sheeny]

Stumbled on this example of a two particle negative phase (entry to atmosphere past the observer) meteor while browse the helicorder looking for interesting signals.


Al.

[graham.hobart]

Hi Sheeny, this is an interesting thread and one close to my heart having some raspberry pi seismometers and a home built meteor cam based on the GMN ones. Would be great to chat via email for more info.
My gmail is
grashie69@gmail.com
cheers graham

[sheeny]

Back from a week away, I decided to do a little infrasound survey on last night's data to see if i could detect a change in the rate of meteors after eta Aqu rose. Looks to have detected the change, though obviously not all meteors that would have been visible were detected and identified by infrasound.


To do the survey, I set up a 0.5 to 10 Hz bandpass filter and scanned the helicorder about 15 to 20 minutes at a time looking for sharp spikes that are balanced above and below zero. I noted the minute these occurred then came back later to zoom in to the wave form and check that it is an N wave (or multiple), check it's not a seismic triggered event and to classify the meteor type. No analysis of the meteor signal was done to split component particles if any were evident.


I need to set the camera up for some star trails and capture some meteor images to correlate to the infrasound.


Al.

[sheeny]

With the eta Aquarid shower in progress I decided to get up early this morning and shoot some meteors in a timelapse and use the time from each photo identify the corresponding infrasound signal.


Sounds simple in a statement like that.


It turned out more difficult than expected, and partly due to some rookie mistakes.


When I first hatched the idea of using timelapse images to time the meteors, I envisaged the camera pointing to the zenith. Intuitively I knew that was the best place for it because that's where the meteors will be closest.


However, it's the eta Aquarids... so I pointed the camera at the radiant and didn't think about it enough.:rolleyes:


Next time... at the zenith!


Not capturing at the zenith increases the distance to the captured meteor. This does a couple of things:


it decreases the signal strength;
it increases the sound travel time, and so increases the timing errors.

The session was successful in that I learned what to do next time. It was not very successful for correlating the meteor images to the infrasound however. The signals were too faint, so mired in noise if detectable at all, and the compound errors from estimating expected heights of meteors and angle of elevation meant it was not possible to clearly identify a single signal that corresponded to the image.


It was also the first time I shot a timelapse with this camera and used the in camera timelapse function rather than my old interval timer dongle. Not being familiar meant the interval between shots was longer than intended. I thought I set 1s between shots but it ended up much longer, so with 15s exposures about 50% of time was missed.:rolleyes: I only realised how much by comparing the gap in a satellite trail between shots. Noise reduction was turned off so that wasn't the culprit.



Attached are my calcs and notes. I didn't get as many shots as I intended as the clouds rolled in, but it was better than nothing.


Al.

[sheeny]

Too cloudy to try to correlate photos to infrasound again this morning, so had a look through the data and found this one that looked interesting.


Resolved it to 3 particles but I suspect there's possible 4 in it, but the final particle signal was pretty weak, so maybe it was real or maybe not.


Al.

[sheeny]

The SpaceX Crew-1 Dragon Trunk, jettisoned just before the de-orbit burn in May 2021 re-entered the atmosphere over southern NSW and Victoria yesterday.


EarthquakesGA received multiple "felt" reports for an earthquake from Cooma, Albury and Wagga Wagga around 7:10am, but no earthquake was detected.


This was the first I'd heard about it, so if it wasn't an earthquake people were feeling, it must be infrasound. I checked the RSnB and confirmed no earthquake then looked at the helicorder of the infrasound trace for anything unusual or obviously not local.


The detected signal (below) was received at 7:23am. 13 minutes travel time at the speed of sound correlates to 270kms distance which is right in the middle of a triangle formed by Albury, Wagga and Cooma.


I posted results on the Australian Meteor Reports Facebook Page and from reports there found it was the SpaceX Crew-1 Trunk re-entry at 7:05am heading SE from near Balranald, past Euroa and Bendigo.


Given the event time of 7:05am and detection at 7:23am the detection distance was 370km.


The multiple reports on Australian Meteor Reports FB page are worth reading to get a picture of what people heard, felt and saw.


Interestingly, from a technical point of view, the Dominant Signal Period (DSP) was 0.5s, with only one period obviously larger at 0.6s which may be due to tumbling.

Al.

[sheeny]

After playing about with this for five months now, I think it's worth adding some comments based on the experience so far.


Detection of meteors via infrasound is possible and practical, but the rates of detection are far lower than visual observation or photographic methods. The problem is that small meteors (say the size of grain of sand) produce only very small infrasound signals which are often attenuated before reaching the detector (Raspberry Boom or Shake and Boom).


The best chance of detection is for meteors at zenith, where the distance to the detector is minimal (50 to 100kms).


Obviously large and fast meteors are easier to detect as they make louder sonic booms on entry to the atmosphere.


The advantage of infrasound detection is it can still take place when visibility is limited, such as during daylight especially and to a lesser extent when cloudy. The reason infrasound is less effective when cloudy is the small water droplets that make up the clouds absorb the sound of the meteor in the same way as the water sprays used beneath a rocket absorb dangerous sound waves rocket. Even humidity increases sound attenuation.


Another advantage of infrasound detection is that for large bolides, and re-entering space junk, detection is possible at large distances - perhaps well beyond what's possible visually or photographically. For example, the detection of the SpaceX Dragon Trunk Re-entry in the previous post.


Unfortunately, rain and wind also produce a lot of infrasound so again the ability to detect meteors in these conditions is reduced as well.


Al.