Changes in Geopsy 3.5.2.

[fabiolafallas]

Hello,

I am reaching out with several questions regarding the transition from Geopsy 2.9.1. to 3.5.2., as I have encountered difficulties replicating previous results with the new version. I use Geopsy to process acceleration signals from environmental vibrations and conduct spectral analysis. Below are my questions regarding specific issues I have encountered:

1. In the Filter option under the Waveform tab, I previously used a bandpass filter with a cosine taper method, which allowed me to select the width of the cosine function. In version 3.5.2, this option only becomes available if I select a low-pass or high-pass filter. Is there a way to adjust the width of the cosine function when applying a bandpass filter in this new version?

2. When using the spectrum analysis tool, the generated Fourier spectrum graph’s Y-axis previously displayed the unit "Amplitude". In version 3.5.2., the Y-axis shows "(counts^2)/Hz," which is not the required output in my situation. How can I modify the Y-axis to display "Amplitude" as the unit?

3. In version 2.9.1, the Spectrum toolbox allowed me to specify an output directory where two files per station (one with a “. spec” extension and the other with a “.log” extension) were generated and saved after selecting the start button. In the current version, I cannot find options to select an output directory or export these files, with the only option being to save the graph. In my case it is essential to be able to obtain these files since I use them for subsequent analysis in MATLAB. Is there a way to obtain this data in version 3.5.2?

Thank you in advance.

Best regards,
Fabiola

[admin]

Hi Fabiola,

Here are some answers. It they are incomplete feel free to continue the discussion.

1. You have to select "Tukey".
2. You have to set the signal property "AmplitudeUnit" (see menu "View/Set data fields" for a table to show/edit it) to Acceleration. Set also "CountPerVolt" and "VoltPerUnit" to something different from 1. After the signal modification (unlock table edition in menu "Edit" or "Set header"), start spectrum tool. If it is changed after starting the tool, Y axis are not updated.
3. In the menu bar on top of the spectrum results, you have the menu item "Tools/Save results" which exports the two wanted files. Notes that the same processing can be done in command line with geopsy-hv.

Best regards,

Marc

[fabiolafallas]

Hi Marc,

Thank you for your help. Your response helped resolve the issues related to questions 1 and 3. However, I am still encountering difficulties with the Y-axis configuration of the spectrum graph. While I have set the properties you mentioned, I am uncertain about the appropriate values to assign to "CountPerVolt" and "VoltPerUnit." In the previous version, I was multiplying the signals by a conversion coefficient of 0.000123 gal. I would greatly appreciate your guidance on how to determine the correct different from 1 values to assign to these properties.

Additionally, when generating the graph, the title of the Y-axis currently shows "Power spectral density [(counts^2)^2/Hz]," whereas the required output for my analysis is for the title to read "Amplitude" (as in version 2.9.1). Furthermore, the amplitude values on the Y-axis are not consistent with those obtained in the previous version.

For your reference, I have attached images that illustrate the settings of the signal properties in the table (with arbitrary values for "CountPerVolt" and "VoltPerUnit") and the spectra generated for the same signal in both version 2.9.1 and version 3.5.2. I hope these images help clarify the issues I am facing.

Thank you in advance

Best regards,
Fabiola

[admin]

Can you post the log files generated by each version? Standard deviations and individual curves look quite different between the two versions. Were they calculated with the same parameters? There might be minor differences even though the parameters are at the same values because computation evolved (frequency sampling, log averaging,...). Just comparing visually the mean, they are nearly at the same level. One way to quickly compare them is to copy the layers (see context menu on the graph contents) from version 2.9 on top of the graph generated with version 3.5. You can hide (null opacity), remove or re-arrange the layers of a graph in the graph contents properties.

[fabiolafallas]

Hello Marc,

Sure, here are the .log files for each version. As far as possible I tried to configure the parameters in the same way, however it may be that I made a mistake because I am not completely familiar with the new interface. Thank you for all your help!

Best regards,
Fabiola

[admin]

Hi Fabiola,

I noticed in the log files that you selected time windows between 30 and 3600 s. This option is there a bit for historical reasons. I suggest you do not use it. Prefer "Exactly". This option was there at a time when FFT was not achieved with FFTW and restricted to power of 2 number of samples. The smoothing is also not the same in your two log files: a 20% (set to 40%) relative window in 3.5.2 is equivalent to B=40 in 2.9 (see "What's this" - help menu for details).