more on auto option using Spac

[luigiV]

Hi Marc,
I submit this new problem that always concerns the auto option with linear array and Spac.
The array is made up of 18 stations spaced by 5m, acquisition time 25 minutes (station 16 was cut because of bad quality).
Entering the coordinates in the table with increasing x (y = 0) and applying Spac does not activate the auto option (see x_coord.jpg).
If on the contrary the coordinates are for increasing y (x = 0) the auto option works correctly (see y_coord.jpg).
Is there an explanation for this?
Finally, is a setting with a 5m rings thickness correct compared to the automatic one (0m)(rings_setting.jpg) ? The target file with 5m rings thickness seem to me more defined than 0m thickness one (comparison.jpg) Do you agree?
Thanks and best regards

luigi

[admin]

Hi Luigi,

I checked the "auto rings" along X or Y on a sample data file: it works equally in both directions with the latest release (3.5.0-preview). The problem you encounter is probably linked to a bug on the current release. Thanks for your vigilance.

About the thickness of rings, I have no particular opinion. The zero thickness is closer to Aki's original SPAC. Having a look at your plot, I could see very little difference. If you prefer the 5 m in this case, maybe that for another case the 0 m will be better.

Best regards,

Marc

[luigiV]

Hi Marc,
I would be really grateful if you could give me your opinion on the reason of the different spectral response of the "Spac" data and ReMi data refer to the same array (same number of geophones, same spacing, different acquisition time - 30 'for "Spac", 15 records 20s long for ReMi, different sampling frequency - 0.008 for "Spac", 0.002 for ReMi). What could be the reason? insufficient sampling frequency for "Spac" or other ...?

Thanks

Luigi

[admin]

Can you swap acquisitions and processing methods?
SPAC can be easily applied on the 15 twenty seconds records. For the opposite, I do not know.
Your SPAC picking could have been slightly higher in terms of velocity, resulting in a curve higher than REMI at higher frequency. You are probably within the range of uncertainties for such kind of sensor layout. At low frequency, SPAC looks like definitively higher than REMI.
Source distribution might be part of the problem. The only way to avoid such questions is to do 2D arrays (at least L-shape array).

Best regards,

Marc

[luigiV]

Hi Marc,
I tried, if I understood your answer correctly, to apply Spac to a 20s ReMi record. I hope I got the parameters wrong, but I didn't get any results. From the message window it is clear that there are not enough time windows to have 34 non overlapping blocks.
The opposite is not possible because the ReMi software does not admit files longer than 20s.

In another site I compared the Spac result of an L array (2m spacing) with the ReMi result of the long side of the L array. The Spac seems to provide a dispersion curve with too high velocities based on the frequency of the H / V peak and the depth of the bedrock (greater than 15m, drill hole data).
In my little experience I have often verified how the dispersion curves obtained by FK and Spac are characterized by higher phase velocities than those sampled on ReMi spectra (when these are reliable, of course).

Best Regards

Luigi

[luigiV]

Hi Marc,
I found a way to get a target file from the ReMi data by setting a frequency dependent window = 2-3T. The result is not encouraging (remi_spac1_allrings.jpg) and the superposition of the ReMi1 dispersion curve makes it evident that no dispersion is present. Reducing the number of rings and using only 4 rings (second, third, fourth and fifth) it seems to me that the target file is less confused and, with good will, we can guess a faint dispersive trend (ReMi1_spac_4rings.jpg).
Do you share or is it just my imagination?
On the contrary, the use of only the first rings in the Spac files seems to me to be decidedly more significant as it seems to me that the dispersive trend appears much more focused than the one that results using all the rings. I also considered only 10 geophones and 4 rings and it seems to me that the dispersive trend is closer to that of the ReMi spectrum. This would mean that with an array of 10 geophones I can have the same result (in terms of depth of investigation) as that obtainable with an array of double length. If this were true, the maximum wavelength formula for a linear array would become, in this case, lambda = 2pi/0.5k.
Does this seem correct to you or is it a statement devoid of consistency?
I attach three target plots images (freq.dep. Window = 30T) and an image of R1_10geos and R1_allgeos where the four dispersion curves calculated by the target files and the ReMi spectrum are superimposed.
I would be very interested in having your opinion.

Thanks and best regards