453 Posts in 222 Topics by 67 members
If this is your first visit, you will need to register before you can post. However, you can browse all messages below.
|Page: 1||Go to End|
|Author||Topic: Reverberation/Bubble after Butterworth filter||861 Views|
14 June 2018 at 12:33am
Hi, I'm processing some vintage MCS data (160 channel-25m receiver distance) and when I perform an initial Butterworth filter just to remove swell noise and avoid aliasing, I get some kind of low-frequency noise (like a reverberation or bubble) that messes up and worsens the data, instead of cleaning it. I attach the original shot and the shot and the same shot after 2 different Butterworth filters: 1-3-100-120 and 5-7-100-120
I also tried to make my own Butterworth filter with more points using the wavelet utility, but the result is similar
I've never had this bad result before... Some ideas?
14 June 2018 at 2:27pm
What does the frequency spectrum of the data look like?
You could test a range of different Butterworth filters to see what result they have on the data - for example try making the lower bound of the filters a bit wider i.e. maybe 1-7 and 5-12 instead of 1-3 and 5-7, you could also try increasing the pass band - eg 5-15-150-160 etc.
If you are trying to remove swell noise you could try the DUSWELL module in GLOBE Claritas and see if this returns better results for the data.
There is an example jobflow in our 2D marine tutorial that looks at testing different swell noise attenuation methods and includes the BUTTERFILT (high pass) and DUSWELL modules - 02_swell.job.
After testing, if the results still aren't looking any good you could send your jobflow and a sample of the data (say a select shot or two) to firstname.lastname@example.org and we can try and take a look for you.
14 June 2018 at 8:09pm
I already tested smoother slopes for the filter (I attach some examples including the ones you suggested) but I still get bubble-kind noise.
I also tried the duswell module, precisely taking the 02_swell.job as model, but even if I tested different values for the Duswell's parameters, the results were as noisy as the raw data (image attached).
I think I'll follow your advice and send the jobflow and a couple of shots to your support service to see if you can find the problem.
|Go to Top|