Originally Posted by asc12
Can anyone enlighten us about the resolution of sidescan sonars? The BEA images show that the Remus AUV was using 120kHz. What sort of spatial resolution could one expect at this frequency? And how is the data displayed?
The speed of sound in water is approximately 1500 m/s (it varies in important ways with temperature, salinity and pressure, but that is not relevant to this question). The wavelength at 120kHz is thus: 1500m/s / 120kHz = 12.5mm. How many wavelengths across does an object need to be to detect it? That's hard to say, because it depends so much on the reflective and resonant properties of the target. I'm not an expert in this area (same math as radar, so maybe others will contribute), but I would say that 10-100 wavelengths would likely be detectable; so that is roughly 10cm to 1m. That's for detection: having a strong enough echo to detect above the noise and clutter..
Resolution is a different question. Along-track resolution is determined by the beam width of the transducer and the motion of the vehicle between pings. The second is easy: at 600m range, for conventional side-scan, the sound has to travel 1200m to leave the vehicle, reflect from a target and return. With some dead time between pings, that makes about 1 ping/second. In that time the vehicle has moved 2 m, so that will be the best possible along-track resolution (unless the sonar is chirped or uses some other way of getting more sound in the water). I don't know the beam width of the system they are using, but it is likely about 0.5degree; that gives about a 5m wide beam at 600m, so that is another limit. Remember that this is about resolution, not detectability.
The other resolution limit, not usually the controlling limit, is the across-track resolution; that is, in the direction between the target and the vehicle. This is limited by some transducer and wavelength considerations, and also by the number of samples, in time, made of the returning echos. Typically that is between 250 and 2000 (what the eye can see), and so this is easy to make better than the along-track resolution.
The data are commonly displayed as an image built up of scans where each scan is the echo intensity (normalized for range and absorption) as a function of range, or more accurately: time scaled and corrected to represent range. Each scan line is perpendicular to the motion of the vehicle (hence the term "side-scan", the sonar is looking to the port and starboard sides using two separate transducers and receivers).
In the released image, you can see that the vehicle proceeded in an "up" or "down" direction with respect to the image (nearly north or south, note the dark bands under the vehicle where there are no immediate reflections), and that the targets are smeared more in the along-track direction than in the across-track direction. I think that picture is a mosaic of views from both sides of the debris (there is more than one dark "nadir" under the vehicle).
In other words, what are we really looking at
in this image? Are the redder parts colored to represent things that are physically located higher, or have a stronger return, or something else? What causes objects to have a stronger return with sonar? At 120kHz, how small an object could one expect to pick out?
The only thing the side-scan can measure is the time and amplitude (and phase, for some systems) of the returning echos. The colors chosen by the operator to represent the amplitude could be anything. In this case the darker colors represent weaker signals, and the brighter colors represent stronger signals.
Return strength from a target is a very complicated topic. It depends on the material, the internal cavities, the shape, the size the orientation toward the receiver, and many other factors. For this long range phase of the search, they were just looking for targets that did not look like the background everywhere else. While the resolution at 120kHz may not be great, higher frequencies are absorbed by the water and thus the maximum range gets shorter. A 55 gallon oil drum would be a reasonable target to detect using 120kHz at 600m range; it would not be possible to resolve the shape and thus identify this target at that frequency and range, using the side-scan presently installed on these vehicles.
After acquiring the debris field they will have surveyed it with higher ferquency side-scan, at intentionally shorter ranges (better along-track resolution). Of course, by now they will also have about 100,000 photos, most of which will show mud with no A/C parts, but the rest will be interesting.