Questions about suitable data format for MASW analysis.

[shandou]

Dear all,

I am a newbie on geopsy, so this is a fairly low-tech question. I got the package successfully installed on my Linux workstation yesterday, but today I keep struggling with my own improper input data format. I used a group of SAC files as inputs for linear f-k analysis. I can visualize waveforms in a graphic window, but whenever I tried to do a linear f-k analysis, an error dialog box would pomp up telling me that I basically did not select any stations. What are the correct steps from inputs to dispersion curve outputs? I tried to get hints from the documentation pages, but things are still foggy for me at the moment.

If someone can show me an example of steps I should take, or a sample input file for me to know the proper input format, I will be more than thrilled!

Thank you very much!

[admin]

The file format is certainly OK. Once you managed to view your signals in a graphic, that's fine. Active FK processing requires several meta data to be set for each of your signals. With SAC format I'm pretty sure that receiver coordinates are not available right after importing signals. You can set them manually by unlocking the table edition (menu Edit), using Set Header or Set Receivers (from the same menu). But it can be something else, keep posting here, if it is still not solved or even if it is solved.

You can find an example (including files to download) at http://www.geopsy.org/wiki/index.php/Ac ... %28MASW%29

[riccardo]

good morning
is it possible to have the final model of the "compressed file" exemple_masw.seg2 at the page
http://www.geopsy.org/wiki/index.php/Ac ... %28MASW%29

[admin]

Do you mean a snapshot of the obtained dispersion? Effectively, this would improve the page's content. I've just modified the wiki in that way, does it give a good answer to your question?

[riccardo]

Wonderful!
but if you post also the Dinver inversion could be PERFECT
do you think it's a good pick?

Frequency (Hz) Slowness (s/m) Stddev (s/m) Weight
6.31357552581262 0.00213942307692308 0 1
7.78202676864245 0.00365503080082135 0 1
8.33269598470363 0.00392070484581498 0 1
9.34225621414914 0.0037006237006237 0 1
10.2600382409178 0.0041108545034642 0 1
10.9024856596558 0.004 0 1
11.453154875717 0.00413953488372093 0 1
11.9120458891013 0.00451776649746193 0 1
13.3804971319312 0.00485013623978202 0 1
13.6558317399618 0.00537764350453172 0 1
14.9407265774379 0.00585526315789473 0 1
18.0611854684512 0.00671698113207547 0 1
19.8049713193117 0.00679389312977099 0 1
21.4569789675003 0.00703557312252696 0 1
22.4665391969407 0.00747899159663865 0 1
24.5774378585086 0.00729508196721311 0 1
28.0650095602294 0.00798206278026905 0 1
34.3977055449331 0.00787610619469026 0 1
38.8030592734226 0.00820276497695852 0 1
41.9235181644359 0.00798206278026905 0 1

[admin]

I would not interpret too much below 10 Hz. Though the curve features between 6 and 10 Hz look quite stable upon variations of processing parameters, a comparison with ambient vibration measurements at lower frequency is certainly required here. The identification of mode is not really straight forward here, the little through between 8 and 10 Hz may be the signature of a mode jumping. Again to be confirmed with ambient vibrations. I cannot remember where these data were taken from.

You can run "Adjust" from menu "Actions" in tab "Curves" from toolbox to fix exactly the FK dispersion curve.

Thanks for the Dinver suggestion. I'll fill this later.

[riccardo]

Wow!
Is what I want !!!

I'll study your comment to my picked curve, and tomorrow I'll ask to you many question !

[shandou]

geopsy_SACwaveforms_demo.jpg

(372.75 KiB) Not downloaded yet

Thank you very much for the help! Sorry for getting back to this so late. Indeed, it is because the receiver coordinates needed to be corrected. After manually doing the editing in the table, now I can get dispersion curves.

However, I have one more questions. My SAC format inputs are a group of synthetic seismograms that are supposed to have very clean onset. I did not spot any problems when I check the waveforms in MATLAB or in SAC interface, but there are big spikes present that are quite close to the starting time of the time series. Why do these spikes show up? Are there any pre-processings that I should apply to the time series to avoid having spikes?

I attached a screen shot of the graphic window. We can see the spike that I mentioned above showing up at around 0.017s after the beginning of the time series.

To summarize, I have two questions at the moment:
(1) Is there any other way to change the receiver location information ahead of time for SAC files? Always need to edit the associated information in table editing does not seem to be the most efficient way when the size of data set grows bigger. Or data format like SEGY, SEGY2 works better than SAC for geopsy to recognize source and receiver locations?

My SAC files have headers that already include source and receiver location and epicentral distances, but the header information does not seem to be all recognizable for geopsy.

(2) Why did I get the spike right after the beginning point of the time series? Are there any additional pre-processing steps that I should use to avoid these spikes?

Thank you very much!

The file format is certainly OK. Once you managed to view your signals in a graphic, that's fine. Active FK processing requires several meta data to be set for each of your signals. With SAC format I'm pretty sure that receiver coordinates are not available right after importing signals. You can set them manually by unlocking the table edition (menu Edit), using Set Header or Set Receivers (from the same menu). But it can be something else, keep posting here, if it is still not solved or even if it is solved.

You can find an example (including files to download) at http://www.geopsy.org/wiki/index.php/Ac ... %28MASW%29

[admin]

There are three way of changing receiver and source coordinates:

• Manual edition as you did
• Set header from menu Edit, which lets you enter a script to be executed for all signals. Intesting release it is rather strict, further version leaves more flexibility.
• Set receiver/set source to read coordinates from a text file

Receiver can be store in SAC header: http://www.geopsy.org/wiki/index.php/Ge ... C_.28gw.29

SEG2 or SEGY have a more official way of storing receiver and source coordinates.

Can you re-send the screen snapshot with a lower display gain? Play with mouse wheel button while pressing SHIFT to decrease amplitudes. I'd like to see the exact shape of these spikes.

[shandou]

Thank you very much for providing detailed instructions on different ways of including coordinates information -- Tremendously helpful!

Sure. I attached another screen-shot in this post.

DisplayComparison_geopsy_vs_SAC.jpg

(257.67 KiB) Not downloaded yet

This jpg image only contains seismogram from one receiver so as to get a clear view of the spike. For comparison, there are two panels present in the displaying window -- The upper one is geopsy graphic display and the lower panel is a display window from SAC software. I drag them together for comparing them roughly side-by-side.

The geopsy image is not scaled to its real maximum amplitude. As you can see, the signal is normalized by the maximum amplitude of individual traces, and the current scale on the vertical axis is around 1e-19, which is consistent with SAC header information "DEPMIN = -1.275487e-19, DEPMAX = 1.256976e-19". However, the geopsy seismogram should have been normalized by the maximum amplitude of the entire trace, which should then set the vertical scales to be around 1. Could it be possible that my original seismograms have quite small amplitudes that somehow challenged the numerical stability in geopsy displaying tools?

I will experiment with multiplying factors to the original synthetics to see whether I can then get rid of those spikes. If you have any insights or suggestions, please let me know!

Many thanks!

P.S. The reason I did not show the full-scale spike is because I would need to scroll mouse wheel button for quite a while to change the scale from 1.0e-19 to 1.0, and somehow the sliding bar at the bottom of the graph window for amplitude control is not a quick fix either. This brings in another question: Is there a direct editing way to specify the vertical scale of the plot that one wants to have, something like xlim(), ylim() functions in matlab? Dragging sliding bars or scrolling mouse wheel button is convenient and flexible, but sometimes it might be faster to type in numbers to make quick change of the scales. Thanks a lot!

There are three way of changing receiver and source coordinates:

• Manual edition as you did
• Set header from menu Edit, which lets you enter a script to be executed for all signals. Intesting release it is rather strict, further version leaves more flexibility.
• Set receiver/set source to read coordinates from a text file

Receiver can be store in SAC header: http://www.geopsy.org/wiki/index.php/Ge ... C_.28gw.29

SEG2 or SEGY have a more official way of storing receiver and source coordinates.

Can you re-send the screen snapshot with a lower display gain? Play with mouse wheel button while pressing SHIFT to decrease amplitudes. I'd like to see the exact shape of these spikes.

[shandou]

After multiplying the SAC files by a factor of 1.0e+19 (Does not have to be this number, but I tried to get the resulting amplitudes have min/max close to -1.0/1.0), the spike is no longer there. However, I ran into another problem here -- Somehow signals displayed in my geopsy graphic window are no longer normalized by maximum amplitude of each individual trace. If I only plot one seismogram, I can still change the vertical axis gain by pressing shift key whiling scrolling my mouse wheel button. When I dragged and dropped multiple traces into the displaying window, I cannot effectively change the vertical axes scales and each trace does not seem to have been normalized correctly either . I might have set certain values way off the scale accidentally, and the property seem to have been locked to be what it is until I can find it out and make changes. Would you please let me know what might have caused the malfunction of multiple traces' display in this case?

Thank you very much and happy holidays!