Bottom Check Distance

Hi,

I'm looking at the bottom check distance from a VectrinoII and comparing it to velocities at different depths. The bottom distance is measured as 5.7cm but the first velocity bin is at about 5.4cm. Is this within the tolerance of the bottom check? I've attached a short binary as an example.

Thanks,

Tim

And apologies for the massive text block there. I forgot to hardcode the paragraph breaks in HTML.

Thank you for attaching the data file, it actually helps a lot to find the problems. The bottom distance measurement is independent of the velocity measurements. We haven't tested the near field limit on detection, but I think it's probably much closer than 5.7 cm. There is a problem with your bottom check data however. The bottom echo is quite loud and saturating the receiver. In the Bottom Distance configuration tab, there is a setting called "Gain Reduction" which takes a value in decibels. Set this so the bottom peak is below -10 dB in the center beam amplitude profile (available as one of the tabs during data collection). For the file you attached, I'd start out with a value of about 10-20 dB and see what effect this has. The bottom peak is quite broad in this file and what it should look like is a fairly sharp Gaussian bump. If you look in the software manual and search for "gain reduction" there's a bit more information on this setting. Also, in the Data.Profiles_Range variable your first velocity cell is listed as being 5 cm away. Where did you come up with the 5.4 cm? P.J.

Sorry, I think i used the wrong terminology. I was expecting my first velocity above the bed to be located at 5.9 cm, strictly based on the the graphs in the velocity tab it looked like the first measurement above the bed was at 5.4. Perhaps I'm just misunderstanding reference points and where things are being measured to/from. Any distance, regardless if it is a bottom check or distance to sample volume is measured from the transmitter? When I configure to profile from 50 to 60 mm, 55mm is the center of the sample volume? - pretty sure I can figure this out. I'd rather focus below...

I've adjusted the gain reduction, if I'm aiming to get the bottom peak (green line on the center beam tab) to be less than -10db, I'm not even close. I'm barely less than -1 with a gain reduction of 60. Should I keep going?

Am I in the right spot or do you want another file?

The yellow line represents the raw amplitude return. This is the one you want to be below -10 dB. The green line is the bottom peak location estimate and should only vary in y location.

 

Everything is referenced to the central transmitter for range. Be aware you won't be able to measure right above the bed. We expect you can measure from 5-10 mm away from the boundary depending on specific experimental conditions such as the boundary type. Configuring to profile from 50 to 60 mm means you will have velocity cells centered at 50, 51, 52, …, 60 mm range from the central transducer. 

Thanks P.J. I'll slowly decrease the gain reduction until my bottom peak is below -10dB. Is there a practical lower limit? Also, just as a theorhetical excercise, what's the downside of setting a large gain reduction and forgetting about it, I'm guessing there is a trade off?

Tim 

Previously Tim Calappi wrote:

Thanks P.J. I'll slowly decrease the gain reduction until my bottom peak is below -10dB. Is there a practical lower limit? Also, just as a theorhetical excercise, what's the downside of setting a large gain reduction and forgetting about it, I'm guessing there is a trade off?

Tim 

hopefully this does not muddy the waters but I suppose I should have asked if there is a practical UPPER limit, since Gain Reduction implies a negative.

This will have to be one of those things checked at the start of data collection to make sure it's correctly set.

I think the noise floor for this is about -80 dB, so you want the peak to be larger than this. I'm really just guessing as I didn't write the bottom detection algorithm, but I'd think you'll want a peak of at least 10-20 dB above this value.

Setting the gain reduction to be too large will mean the bottom peak is lost in the background noise and won't be as reliably detected. Because different bottom surfaces will provide different echo intensities, there's range attenuation of the signal, etc. I don't think you'll be able to get away with setting a large value and just forgetting about it and still get good results. The receiver is also only linear within a certain range, so aiming for the middle of the receivers range should yield the best results.

Thanks P.J. I'll try your suggestion today.

The SNR improvements are even more impressive when you realize the sample volume is about 1/7 the size of the original Vectrino's! The engineers who designed the VII did a really great job on the hardware.

Generally speaking and with still some learning on our end, SNR above 20 should be good. You'll want to be a little more careful at the ends of the profile, especially the far limit, because the geometry of the head causes a decrease in SNR there. I've actually never been a huge fan of relying on SNR filtering as a good outlier detection scheme, so I would recommend maximizing SNR for your experiment and then tossing really low values (say below 10 or so depending on the dataset). A specific outlier filter can then be used on each to bin to remove statistical outliers from the dataset. There are many options on how to do this, but I'm a fan of a simple Gaussian outlier detection since it's simple to code myself and fairly robust.

How's this for a center beam check? I'm concerned about the asymmetry in the yellow line. The VII is level and the bed is flat. I'm using a gain reduction of 40 over a sand bed. 8cm total depth of water 90cm wide flume.

 

Thanks for all the quick responses, much appreciated.

It looks fine, I wouldn't be too worried about the asymmetry, this could be affected by many different variables and is probably a little due to the shallow water depth.  What cell size are you using for the bottom check?

Bottom check cell size is 1mm. I've attached a short sample I just collected if you want to look at other parameters.

 

 

No problem. It just looked a little coarse, but I'll admit it's hard to tell from the screenshot. I think we've got most of this figured out, but if you have any further questions feel free to post them.