FK passive linear

[luigiV]

Hi Marc,
I found in an Eucenter's work on the 2012 Emilia earthquake (GEOTECHNICAL ASPECTS OF MAY 20, 2012 M5.9 EMILIA EARTHQUAKE, ITALY Version 1.0July 13, 2012) four attractive examples of FK elaborations made with Geopsy on linear ReMi arrays with small geophones spacing and 30 minutes of recording.
I remember that in geopsy versions prior to 3.0 version there was an icon for the passive linear F-K which no longer exists in later versions. I tried anyway with version 3.4 to process a file with 24 channels, 2 m spacing and 30 minutes of recording by accepting the "passive conventional" option, the setting for the grid step (0.0341 rad / m) and modifying the grid size from 6.4rad / m to 0.9 rad / m. The result was incomprehensible and I believe devoid of meaning and usefulness.
Is it no longer possible with Geopsy to do the F-K of linear passive arrays? Do I need to retrieve an old version for this operation?
Thank you


Interested in the problem, I downloaded the version 2.5 and processed the data obtaining a result that is very similar to that of the ReMi and calculated with the same corner frequency (20Hz, see the pic). The FK spectrum also defines the average curve of the PDF maxima which, in the FK of 2D arrays, approximates the dispersion curve. In the case that I submit to you, accepting this picking would be contrary to the theoretical principle underlying the ReMi tests which requires sampling the energetically lower limit of the fundamental mode because the azimuth of the sources and their nature is not known.
Can I ask your opinion on this? Could the application of the FK technique to linear passive seismic be considered of little use and interest?


Comparing my F-K spectrum with those of Eucenter it is evident that the processing used is not that of the linear passive FK available in the versions of Geopsy downloadable from the repository.
Can you please tell me what it is?

Thanks again

Luigi

[luigiV]

Hi Marc,
why the FK tool for linear passive (ReMi) is no longer available in Geopsy versions after 2.9?
Using the tool for passive conventional FK of versions 3 and higher, the result obtained is completely incomprehensible.
Is there a reason for the exclusion of this investigative approach from the passive methods considered by Geopsy?
Thanks
Luigi

[admin]

Hi Luigi,

If I'm correct (I haven't checked), you can use the classical passive FK tool. Is the .max file that you qualify as "incomprehensible"? The .max file can be analysed by gphistogram or gpviewmax.

Note that you can also use the SPAC tool with linear passive arrays.It might be better.
Sorry for the many old unanswered messages, I've not so much time these days to cover all of them.

Best regards,

Marc

[luigiV]

Hi Marc, no problem.
But I propose again part of my previous post related to the same problem and I am attaching the max file obtained with Geopsy 3.4.1 with the "passive conventional" option making clear the term "incomprehensible" that I used.
I am also attaching the output obtained with the same data file (24 geophones, 2m spacing, 30 minutes) using the FK linear tool of version 2.5. In the same image I also included the ReMi spectrum which shows that the two spectra are comparable. The difference is that the dispersion curve proposed by FK joins the energy maxima wich have much higher Vphase values ​​than those that I would have sampled using the ReMi spectrum (energy minima).
(The log window, when I active the Fk linear, shows, after loading all 24 traces, "Error no root found searching kmax")

Finally, I also attach the spectra obtained with Geopsy by Eucenter in a work related to the 2012 earthquake in Emilia (GEOTECHNICAL ASPECTS OF MAY 20, 2012 M5.9 EMILIA EARTHQUAKE, ITALY Version 1.0 July 13, 2012) that had inspired my applicative interest. The result obtained is decidedly different and I cannot understand the reason (the starting files have the same geometry and the same acquisition time of mine). Hence my post and the request for your authoritative opinion to try to understand how to achieve the same result with Geopsy.

I will follow your suggestion using SPAC for linear passive even if I don't understand how it works.

Many thanks and sorry for my insistence on this topic
Luigi

[admin]

Hi Luigi,

In lin_pass_25, the velocity range is around 500 m/s while in the 3.4.1, everything is around 50 m/s.
Can you see something strange in the parameters for 3.4.1? If not, can you send me an example to check here what's going wrong (eventually by email if there are confidentiality issues)?

Best regards,

Marc

[luigiV]

Hi Marc
I did not notice anything strange in the parameters of 3.4.1. However I attach 4 pics related to parameter settings and the log window.
By example, do you mean the gpy file I used? It's a bit heavy (144Mb) ... maybe reducing it to about ten minutes could be better and I would send it to your email with wetransfer
Let me know
Thank you
Luigi

[admin]

If you can send just the half an hour record shown in par1.jpg that's fine. I noticed that a few receivers have very weak signal. It is probably better to remove them from the processing.

• Either add gaps, with the context menu, over the periods of weak signals. With the option "Select minimum station duration" get rid of the receivers not available over, for instance, 80% of the time.
• Or move to a table, select only the "good" receivers and move to a new graphic.

[luigiV]

Hi Marc,
I am sending you the clean gpy file of the 3 tracks where the signal was not good. I checked in the table that the x coordinates were correct and took into account the deleted tracks.
Hope you can find the reason why FK passive conventional doesn't work right.
One last question, before you say me to go to hell: how did Eucenter get those spectra I sent you with an old Geopsy version?

Thanks for your patience
Luigi

[admin]

Hi Luigi,

I would need also access to the signal files: FK_l_09/FKLineare_AMT09_01.txt to 24.txt
The .gpy contains only metadata and not the plain signals.

In the old geopsy 2.*, there was a specific tool for computing linear passive arrays (REMI). In the newest, all FK processings have been unified in a single core. In principle you should achieve the same results.

Cheers

Marc

[luigiV]

Hi Marc,
I was also surprised that such a small file contained all the txt data. But I had tried moving it to a directory that didn't contain them and FK worked fine .... so I thought it might work for you too.
However, since the file is too large to send from the forum, I will send it to you by e-mail.
Greetings and thanks
Luigi

p.s. I tried to send but I got a return message saying that the file is too big (552 5.5.2 Channel size limit exceeded (20000 Kb)) even deleting the three files with bad data.
So what can I do to get them to you?

[luigiV]

Hi Marc,
I was surprised that I still haven't had any reply to the mail with linear passive FK data , but I'm sure I'll read to you shortly.

With regard
Luigi

[admin]

Hi Luigi,

I downloaded your data and I checked the FK processing. I obtained the same erroneous results as you. I still have to enquiry for the reasons of this puzzling result. Be patient, I'll get in touch with you whenever I get news.

Marc

[luigiV]

Will be patient, Marc, but don't forget me, please.

Cheers
Luigi

[admin]

Hi Luigi,

I finally considered your test case again. The current FK engine do not fully support linear arrays for passive experiments like in the past. Re-introducing it can be done under 3.5.0 but not for the next 3.4.2 release. Note also that SPAC technique can be used on such arrays. It is relatively straight forward. The automatic definition of rings works pretty well. You just have to adjust the frequency range and the output file name. I could not compare with your figure because the X scale was not available. The SPAC catches a nice dispersion between 6 and 10 Hz but I'm unsure if it is compatible with the context.

Best regards,

Marc

[luigiV]

Many thanks, Marc.
Indeed, a very evident dispersion emerges.
I am attaching the ReMi spectrum of the same array where I have indicated the frequency limits of the Spac dispersion.
I would say that there is agreement on the trend but not on the Vphase values: Spac dispersion corresponds to the maximum modal energy, while ReMi picking follows (should follow) the minimum modal energy level in order not to overestimate the Vphase, not knowing the position of the sources of the noise. Spac's Vphase at 10Hz is 235m/s, that of ReMi picking 185m/s. At 6Hz the Vphase of Spac is 480m/s, that of ReMi 285m/s. Therefore, following this principle, I should define on the PDF Spac a dispersion curve that corresponds to the minimum value of the trend of maximum probability. But on a much more uncertain basis than that of the ReMI spectrum (when this is reliable, of course).
Returning to the FK spectra of Lai et al.(Eucenter pics) I don't know how they worked, but the representation of the spectral data is very different from the FK passive linear one of the releases greater than 2.5 and the picking of the dispersion correctly corresponds to the maximum energy. Is it possible to download a previous version, that of 2012-2013 for example to see if there is this difference?

Best regards
Luigi