1MHz Aquadopp Profiler signal

Hi, 

I remember this was addressed a lot a few years back, but since then the boards have moved and I can not find all the old posts, and could not get in on my old user name, or search the answers to my old posts.  I don't know even then if the answers were there.

Using the 1 MHz Aquadopp profiler, to look at ice crystals in the water we got some strange results, particularly after accounting for the transmission loss according to the very informative  Nortek Techincal note 3 on sediments.  Could you clear up a few things for me to reference.

Firstly the conversion to decibel ranged from .40-.47, and values were given in posts on the forum for specific instruments, is there a more specific value for the 1MHz aquadopp?  I thought I remembered 0.43 being used but see no specific reference to it, and the specifications were 0.45dB resolution in signal.

Secondly, the technical note refers to the signal output as both intensity and amplitude, is this like a unit of power intensity, W m^-2, not a voltage amplitude unit, as from what I understand the backscatter cross-section is proportional to power and power intensity are proportional to the voltage^2. 

Thirdly the signal output is described as approximately linear over a range of 70dB, is this from the lower end, close to instrument noise with the non linear part at extremely high signals?  Or is the non linear part where the signal is weak.

The issue I had was with the instrument looking down, in very clear Antarctic water under the sea ice.  Using only the geometric beam spread term of transmission loss, I would generate a uniform scatter profile.  Then adding the water absorption term I would end up with an increase of scatter cross section with depth, with no real reason to expect more scatters with depth (500m+ water depth).  When there were ice crystals in the water column, the signal was much greater and the background signal would also rise, and the scatterer profile would then become linear after applying the water absorption term, but no scatterer absorption.

A possibly related issue was that the vertical velocities were corrupted, show a systematic increase with bin number of several cm/s for each m, for all but the highest signals.  We are not even confident of one bin of vertical velocity although the signal is sufficient for 8-10 bins of horizontal velocities. However with the Doppler shift being in radial direction,  I thought vertical velocities would be the most accurate  (downward orientation all 3 beams offset 25o from vertical).  There was nothing in the deployment software to indicate the accuracy of the vertical velocity would be less, or that it could result in such an extreme reported velocity increase.

The explanation I am considering is some sort of high reverberation signal from side-lobes scattered by the ice-water interface above.  We were careful when deploying to place the head of the ADCP far enough below the ice-water interface, watching the signal in real time as we deployed.  We also ignored the first bin which was abnormally noisy.  There were guidelines for avoiding side lobe interference for horizontal deployments, but no information on side-lobes extending beyond 90o, behind the instrument.  Can you provide any information on how strong these rear facing side-lobes could be, and is it possible they could elevate or confuse the signal in further bins? 

I can of course provide profiles or files showing this data, but it is really self explanatory, and I work with my own custom matlab software on the data set now and not the original files.

Regards

Craig Purdie

Dear Jonas,

Thank you for those emails, I am impressed you remembered from that far back, since I changed emails!  It was a good refresher and contained a lot of information there.

Unanswered questions:

I never saw any references confirming that the correct value for the aquadopp profiler 1MHz model was 0.43 dB/count and not 0.45 dB/count as given in our specifications.  Both values are used for different instruments.  Is there a single best value for that particular model or is it just an average in that range.

Dear Craig,

I have a really hard time finding the broken link. I will continue searching for it.

The dB/counts number is specific for each mainboard. I have been looking at the last 200 produced and they vary from 0.42 to 0.52 with an average of 0.47. This is not the hardware platform you've got, but the principle is the same. The 0.47 is if you calculate between a very high signal and a very weak signal.

Attached is the average for the last 200 produced. The dB/count numbers vary quite some bit dependent where you are on this curve. 0.43 between -120dB and -100dB and 0.57 between -90dB and -60 dB.

Please tell me if there are uncertainties.

Best regards

Jonas Røstad

Dear Jonas,

I was looking at this plot again and wondering if the absolute value of the injected signal has any significance at all, in relation to the maximum output of the transducer?  Or is this just a test of the instrument circuit, maybe not even using the transducer?

Most of the work I have done trying to analyse the signal is trying to determine if our 30 count background signal represents a return closer to -120dB or -80dB.  This has been very frustrating when most of the other acoustic instruments which have published work on this have at least had a rough 1 point calibration of signal levels to within 10dB.

Also were you making these measurements back for our instrument in 2002?  The serial number was AQD 0736.

Thanks again for your response and the plot to show how the value dB/counts varies.

Craig

Dear Craig,

As you are suggesting - this is an electronics test. 

I am not quite sure if I understand, but from the graph 31 counts is -120dB. We have always done these measurements, but need to dig deep to find the data of the 9 year old main board. 

I do not know what other manufacturer are promising in accuracy and use of the signal strength. We have done quite some job in finding out if the amplitude signal could have any scientific value, some of the early work are commented/described at http://www.nortek-as.com/lib/technical-notes/seditments and http://www.nortek-as.com/en/knowledge-center/forum/current-profilers-and-current-meters/579860282#542836693

These type of instruments are designed for measuring velocities in the water column. The returned signal strength is more of a quality control parameter than a scientific value. With that said quite some customers have used it to get  valuable information.

Best regards

Jonas

How does the -120dB relate to the instrument output power?  Does the value of -120dB at 30 counts for your circuit test not require a reference level defined such as the output power of the instrument.  Then from EL = SL - TL + BSv,  the -120dB would relate to (EL-SL), although the characteristics of the transducer would need to be taken into account.  For other instruments we have information on both SL, the source signal level, or the echo level scale is related to SL.  Even if the -120dB is not related to instrument max power input, if the power scale it was referenced to was known ...

The technical note on Sediments provides all the information on the relative measure of sediments, even saying that the 70db linear measurement range corresponds to sediment concentrations of 1-10000mg/l, (which must assume some size distribution as an increase in numeric concentration would only have produced a 50db range?).  There are significant size ranges of sediment in different ocean waters and rivers etc.

Relating the signal return to the instrument power output at the minimum instrument level of 30 counts, (which you seem to do) is the last unknown in our calculations.   This is a specification of the instrument, a rough measure only depends on the electronics, and basic transponder parameters.  The technical note states this varies only about 5-10% between instruments, but with the range of sediment sizes affecting the signal return by 60log(a) it is only required within an accuracy of about 20db.  So this -120db value in the plot is very important if it is related to the maximum instrument output power, or to a known power that can be related back to the instrument output.

Our work is looking at characterising different populations of scatters, of greatly differing sizes, rather than changing sediment concentrations, where the change in signal can be related to a change in concentration, which is why we have been trying to get information on instrument parameters, and unfortunately the instrument is no longer available.

Anyhow, thankyou for all the information, I hope that -120dB can be the parameter we are after, or lead to its calculation.

Cheers

Craig

Hi Craig,

First I want to repeat that your task is a hard one especially since the instrument is no longer available.

An estimate of the Spectral Strength (dB ref μPa@1m) of an 1MHz Aquadopp Profiler is 186dB when highest power level and largest cell size is chosen. 

I don't think this will help you a lot since you don't have any reference value of what this will be reported as in the AquaPro. The dB term used above is actually dBm and have no absolute value other than it is a 1mW@50ohm injected signal. The only thing you can use this to is for calculating the slope dB/count.

As the links above are describing you need to have a reference measurement. This can done by using a calibrated target (for example a steel ball) and refer all data to the measured target strength. An optical backscatter sensor could also give a reference point. 

Without a reference measurement all measurements will be relative - but that might be good enough.

 From the front of our webpage it says "Nortek develops and manufactures Acoustic Doppler Current Profilers (ADCP) and current meters for use in the ocean, rivers, lakes and  laboratories." Although we are instrument makers, we try to help our customers in solving their scientific missions. I am sorry that I cannot give you the missing part in your formula.

Best regards

Jonas Røstad