TheBitterEnd
Well-known member
Interesting paper on mapping spaces from acoustic echos:-
http://www.pnas.org/content/early/2013/06/12/1221464110.full.pdf
http://www.pnas.org/content/early/2013/06/12/1221464110.full.pdf
Echo location mapping
- Thread starter TheBitterEnd
- Start date
Certainly an interesting idea but it's hard to see how it could be adapted for cave surveying, though I have pondered on whether sonar could be used in sump surveying. The paper's mathematics is beyond me by several orders of magnitude, but I do have a grasp of practical sonar principles, hence my scepticism. The authors drop a clanger too, in their final comments, by using the term "radar" in their suggested applications: it cannot possibly be radar by definition!
A building's interior is primarily of plane surfaces of usually high and fairly even acoustic reflectivity, even given the complexities implied by the photograph of a church, usually a highly reverberant building. Even then the method uses arrays of microphones &/or loudspeakers, with the attendant multiplicity of amplifiers etc. I'm not sure if the technique described works well in high reverberance: reverberation is a chaotic mass of overlapping echoes of echoes of echoes of...
Cave walls and roofs are irregular, and their acoustic properties must vary widely but generally disadvantageously across even a fairly small area. Floors are variable but mud or gravel especially must be of very low reflectivity, down to anechoic (in fact most caves are remarkably anechoic anyway) and of very high scattering and absorbtion characters. A still pool may be a fairly good sound reflector, acting as a plane mirror, but a large stream emits wide-band noise greatly deleterious to acoustic space-measuring, especially at lower frequencies. Most caves are of low reverberance, or even practically anechoic at least in the human audio range (<20kHz).
Bats navigate by echo-location using trains of short pulses of very high-frequency (typically >50kHz) tones, and the higher the frequency the lower the feasible beam-angle, but also the lower the range; and in fact although their calls often exceed 90 or even 100dB re 26?Pa (the reference sound pressure level for air acoustics, and = 0dB, the minimum detectable by human hearing) the animals' power is very low. So their effective range is no more than a few metres, thanks to both low power output and to most of the sound being lost by absorption and scattering.
Whilst theoretically possible to build a sound-ranging set for measuring caves in the way the paper suggests, I propose that it would give little more than rough estimates of volume and wall location. It might give you the approximate shape of a chamber... but a tape-measure and compass would be simpler, cheaper, more accurate and more efficient. In a few reverberant passages or chambers, it may be possible to assess (space) volume by resonance, but the practical difficulties alone make this highly unlikely.
Intriguingly the paper suggests using microphone arrays to detect the whereabouts of a phone user: would radio d/f be more effective?
Given the developments of optical range-finding in cave surveying, LIDAR plotting etc., I think light wins hansomely over sound!
As for sonar for surveying sumps, though perhaps theoretically possible, using high-frequency hence narrow-beam projectors, I doubt it will be part of the cave-diver's kit for a long time yet if only due to the difficulties in building suitable, practical and (relatively!) affordable equipment for the purpose I stand to be corrected if a CDG member says, actually we do...
A building's interior is primarily of plane surfaces of usually high and fairly even acoustic reflectivity, even given the complexities implied by the photograph of a church, usually a highly reverberant building. Even then the method uses arrays of microphones &/or loudspeakers, with the attendant multiplicity of amplifiers etc. I'm not sure if the technique described works well in high reverberance: reverberation is a chaotic mass of overlapping echoes of echoes of echoes of...
Cave walls and roofs are irregular, and their acoustic properties must vary widely but generally disadvantageously across even a fairly small area. Floors are variable but mud or gravel especially must be of very low reflectivity, down to anechoic (in fact most caves are remarkably anechoic anyway) and of very high scattering and absorbtion characters. A still pool may be a fairly good sound reflector, acting as a plane mirror, but a large stream emits wide-band noise greatly deleterious to acoustic space-measuring, especially at lower frequencies. Most caves are of low reverberance, or even practically anechoic at least in the human audio range (<20kHz).
Bats navigate by echo-location using trains of short pulses of very high-frequency (typically >50kHz) tones, and the higher the frequency the lower the feasible beam-angle, but also the lower the range; and in fact although their calls often exceed 90 or even 100dB re 26?Pa (the reference sound pressure level for air acoustics, and = 0dB, the minimum detectable by human hearing) the animals' power is very low. So their effective range is no more than a few metres, thanks to both low power output and to most of the sound being lost by absorption and scattering.
Whilst theoretically possible to build a sound-ranging set for measuring caves in the way the paper suggests, I propose that it would give little more than rough estimates of volume and wall location. It might give you the approximate shape of a chamber... but a tape-measure and compass would be simpler, cheaper, more accurate and more efficient. In a few reverberant passages or chambers, it may be possible to assess (space) volume by resonance, but the practical difficulties alone make this highly unlikely.
Intriguingly the paper suggests using microphone arrays to detect the whereabouts of a phone user: would radio d/f be more effective?
Given the developments of optical range-finding in cave surveying, LIDAR plotting etc., I think light wins hansomely over sound!
As for sonar for surveying sumps, though perhaps theoretically possible, using high-frequency hence narrow-beam projectors, I doubt it will be part of the cave-diver's kit for a long time yet if only due to the difficulties in building suitable, practical and (relatively!) affordable equipment for the purpose I stand to be corrected if a CDG member says, actually we do...
graham
New member
I'd be as sceptical as NigelG. The problem with using sound is that there is no equivalent to a laser beam in light. Thus you cannot work on direct distance measurement in the way that a disto does - a direct echo if you like - & you cannot identify point targets. This technique would require a lot of detectors (microphones) & even more complex post-processing that a lidar scan. Even then, given that the wavelength of sound at its shortest is no shorter than that of light, by several orders of magnitude, you can never theoretically better the measurements that are possible using lasers.
TheBitterEnd
Well-known member
I quite agree that surface roughness would be an issue with finding a solution to the Euclidean Distance Matrix but may be not impossible. Still, it's an interesting approach and something that could be investigated for not much outlay. Obviously time-of-flight laser scanners are "better" and I suppose one day the price of those may come down...
Pitlamp
Well-known member
Just for the record, sonar has been used very effectively by CDG members on several occasions. The first occasion I'm aware of was in Pridhamsleigh Cavern to get a simple 3D visualization of the big lake. Nick Bennett made an underwater 3D mapper as a college project in the 80s which he used in Keld Head. Various people have used devices based on sonar since then including Pete Mulholland, Scoff, DMP (I think) - and more recently Christine (on this forum, who kindly gave me some useful information about them only a few weeks ago). The Pozo Azul surveying has been done with a device which I think works on sonar - and very effectively.
Sonar has certain advantages in underwater cave surveying. It speeds the job up compared with a human surveyor, which becomes increasingly important in deeper sumps (which have proportionately higher decompression issues). It's extremely valuable in British sumps where you can't see roof, walls or floor - not just for surveying but for passage detecting (when width measurements suddenly give a big reading).
Don't ask me what a "Euclidian Distance Matrix" is though . . . . .
Sonar has certain advantages in underwater cave surveying. It speeds the job up compared with a human surveyor, which becomes increasingly important in deeper sumps (which have proportionately higher decompression issues). It's extremely valuable in British sumps where you can't see roof, walls or floor - not just for surveying but for passage detecting (when width measurements suddenly give a big reading).
Don't ask me what a "Euclidian Distance Matrix" is though . . . . .
Pitlamp
Well-known member
That's right Graham - the point is it may not be entirely precise but when logistics dictate that there isn't time physically to measure all the distances from centre line to roof, floor and walls (as cave diving is a time limited activity) it's a whole lot better than guessing!
Duncan Price
Active member
graham said:
I've used a hand-held SONAR wand to survey a number of sumps, including Gough's Sump 1 from the downstream limit beyond Dire Straits through Lloyd Hall and to Bishop's Palace. I have also done Wookey Hole 3 to 22 and Pwll-y-Cwm to Daren Cilau. The latter had measurements done at 5 m increments so that's 130 sets of LRUD measurements + extras at splays. The device I used can be read to the neareast 0.1' since it of US origin and the reproducibility of measurements is about two to three times this. Not as accurate as using a tape measure to do the centreline in a sump but commensurate with estimating distances from line tags (i.e. better than guessing).
The technique has potential to identify side passages in sumps though none of the anomalies that I've measured have turned out to go anywhere. It is pretty good at detecting airbells above you as the refraction of the sound at the air-water interface gives rise to very erratic readings (I normally take multiple shots and compare them as in big passage/poor visibility you are shooting "blind"). The "time limitation" that John Cordingley referred was mitigated by employing a rebreather in Pwll-y-Cwm and also Wookey Hole. For the former the entrie sump was done in 4 dives of 90 minute duration covering 150-200 m at a time.
The Pol Azul survey uses completle different technology (John Volanthen's "Lazy Boy" mapper) - it is essentially a 3 axis compass and pressure cell coupled to a propeller driven distance meter. The data are logged at 1 second intervals and used to construct a centreline. The diver simply has to swim or scooter through the cave with the device. An early version was tested at Wookey Hole just using the diver's swimming speed as a measure, it was also used to survey Sump 3 (and 2 for which we already had conventional data) in Gough's Cave. A transit in both direcetions improves the accuracy.
I have a contact with access to a military spec submersible inertial navigation system who I was trying to get to borrow it for some underwater surveying. This type of device coupled with side-scan sonar was used to produce a 3D survey of Wakulla Springs in 1998. Errors in the internal mapper were corrected by reference to dropped radiolocation beacons. Bill Stone has gone on to develop an autonomous submersible mapper for cave exploration (DEPTHX) and it has also been used in the Anarctic (ENDURANCE). The intent is to depoly one of these on Europa to look for signs of life there.
A far better and simpler tool are the underwater cave-mapping sensors developed at the Southwest Research Institute.
First: Thankyou for the information from the divers - a very interesting field!
Secondly, a CORRECTION - of my own clanger! I'm suprised no-one else spotted it. In the paragraph about bats I gave the reference SPL for air acoustics as 0dB re 26?Pa. it should of course be 20?Pa. I'd recalled and used wrongly the difference in decibels between the 20?Pa for in-air measurements, and the 1?Pa used in underwater work. Sorry about that!
If you think about the numbers, and given that you need 100 000 micro-Pascals just to make 1Bar (standard atmospheric pressure at sea-level), then you can easily work out the pressure in the faintest whisper you can hear (assuming fully healthy hearing) - it shows how incredibly sensitive our ears are.
Acoustics, both in air and in water, is a fascinating and extremely complex subject - I am not a mathematician & I wouldn't know a matrix if it fell on me, Euclidean or not - and my interest arose from memories of previous work in a factory making sea-bed mapping side-scan sonars being aroused by pondering on the mental imaging of bats. Do they "see" in their minds something like a side-scan image (looks a bit like a TV image), albeit probably at very low resolution?
It's rather awe-inspiring to realise that though a bat's detection range is only a few metres, the speed, complexity and precision of its signal-processing and consequent body control in such a small volume and weight are not replicable artficially in any present-day engineering.
Whether air-borne acoustics can be used for practical, reasonably accurate dry cave measuring remains to be seen, though it does look from the cave-divers' responses here that sonar for sump measuring is sufficiently well-developed already to be at least a useful adjunct to other techniques.
The accuracy is enhanced by raising the frequency, which has the advantage of reducing the physical size (and mass) of the transducer, but at some cost of reducing the range. The higher the frequency of the projected sound, the more rapidly it is attenuated by the water through which it travels, though over the likely range necessary the loss is probably within sensible bounds. The other main advantages of high frequencies is that the sound can be sent as short pulses, allowing clear reflections to be received and discriminated; and the relative ease of emitting narrow sound beams for greater accuracy. These are exploited in echo-locating by animals such as bats and dolphins homing in on prey.
As an aside, I recall experiencing what may have been natural acoustic resonance in a cave on the remote Elgfjell plateau in South Nordland, central Norway. Breathing Cave, as Alan Marshall and I called it, is a short, dry passage with a single oxbow, descending steeply to a complete sand choke. Having finished the surveying, we waited in the entrance for a rain squall to pass. While there we noticed the cave was draughting alternately inwards and outwards, with a cycle from memory of at least 30s, more like a minute. Since we had seen no other openings in the cave walls, we can only guess the air in the cave was resonating with the wind across its entrance, like that induced by blowing across a bottle top.
Secondly, a CORRECTION - of my own clanger! I'm suprised no-one else spotted it. In the paragraph about bats I gave the reference SPL for air acoustics as 0dB re 26?Pa. it should of course be 20?Pa. I'd recalled and used wrongly the difference in decibels between the 20?Pa for in-air measurements, and the 1?Pa used in underwater work. Sorry about that!
If you think about the numbers, and given that you need 100 000 micro-Pascals just to make 1Bar (standard atmospheric pressure at sea-level), then you can easily work out the pressure in the faintest whisper you can hear (assuming fully healthy hearing) - it shows how incredibly sensitive our ears are.
Acoustics, both in air and in water, is a fascinating and extremely complex subject - I am not a mathematician & I wouldn't know a matrix if it fell on me, Euclidean or not - and my interest arose from memories of previous work in a factory making sea-bed mapping side-scan sonars being aroused by pondering on the mental imaging of bats. Do they "see" in their minds something like a side-scan image (looks a bit like a TV image), albeit probably at very low resolution?
It's rather awe-inspiring to realise that though a bat's detection range is only a few metres, the speed, complexity and precision of its signal-processing and consequent body control in such a small volume and weight are not replicable artficially in any present-day engineering.
Whether air-borne acoustics can be used for practical, reasonably accurate dry cave measuring remains to be seen, though it does look from the cave-divers' responses here that sonar for sump measuring is sufficiently well-developed already to be at least a useful adjunct to other techniques.
The accuracy is enhanced by raising the frequency, which has the advantage of reducing the physical size (and mass) of the transducer, but at some cost of reducing the range. The higher the frequency of the projected sound, the more rapidly it is attenuated by the water through which it travels, though over the likely range necessary the loss is probably within sensible bounds. The other main advantages of high frequencies is that the sound can be sent as short pulses, allowing clear reflections to be received and discriminated; and the relative ease of emitting narrow sound beams for greater accuracy. These are exploited in echo-locating by animals such as bats and dolphins homing in on prey.
As an aside, I recall experiencing what may have been natural acoustic resonance in a cave on the remote Elgfjell plateau in South Nordland, central Norway. Breathing Cave, as Alan Marshall and I called it, is a short, dry passage with a single oxbow, descending steeply to a complete sand choke. Having finished the surveying, we waited in the entrance for a rain squall to pass. While there we noticed the cave was draughting alternately inwards and outwards, with a cycle from memory of at least 30s, more like a minute. Since we had seen no other openings in the cave walls, we can only guess the air in the cave was resonating with the wind across its entrance, like that induced by blowing across a bottle top.
Duncan Price
Active member
NigelG said:
A cave resonance effect driven by cave divers' exhaust bubbles is proposed as the cause of an underwater landslide which caused the death of one of them: PDF (637 kB)
Chocolate fireguard
Active member
NigelG said:
Actually I did spot it but thought things might have changed in the 30 years since I worked in acoustics.
And as it only makes about 2dB difference I decided it didn`t really alter your subsequent statement about the SPL of bat calls.
However, I am not sure about the 2nd paragraph copied above. 1 Bar is 100,000 Pa, near enough, and as there are 1,000,000 micro Pa in one Pa I think you need a few more noughts on the number you give.