nfsi-canada / obstools Goto Github PK

When I run atacr.DayNoise.QC_daily_spectra on a stream containing 1728001-sample traces, I get the following error:

Traceback (most recent call last):
  File "run_corrections.py", line 97, in <module>
    main()
  File "run_corrections.py", line 42, in main
    out_streams['ATACR w/o pre-rotation'] = run_atacr(stream)
  File "run_corrections.py", line 72, in run_atacr
    daynoise.average_daily_spectra(fig_average=True)
  File "/Users/crawford/_Work/Programming/OBStools/obstools/atacr/classes.py", line 664, in average_daily_spectra
    np.mean(ftZ[self.goodwins, :]*np.conj(ftZ[self.goodwins, :]),
IndexError: boolean index did not match indexed array along dimension 0; dimension is 341 but corresponding boolean dimension is 342

I see that self.goodwins is based on the length of "t" returned by spectrogram() and that ftZ is calculated using
utils.calc_windowed_fft()

This happens with multi-day data, which of course isn't logical for DayNoise(), but I suspect it would happen with daily data with a "high" sampling rate (20+sps). I can get rid of the problem by reducing the Trace length, but it appears in any case that spectrogram() and utils.sliding_window() don't calculate exactly the same windows, so the windows calculated for calc_windowed_fft() do not correspond exactly to those returned by spectrogram().

I leave it to you to decide if this is important or not.

Regards
Wayne Crawford

Dear developers,

First of all, thanks for the package, it is very useful. Now, I have a question.
I note that on line 368 of atacr/utils.py the function calculate_tilt uses a default
frequency band of 0.005 to 0.035. Is there a reason to use this band?

I ask because the class DayNoise does not change this frequency band when
calling calculate_tilt during the function average_daily_spectra.

Thanks in advance,
Eduardo.

To do list for improvements to OBStools:

• Improve documentation, especially for the utils and plot functions of obstools.atacr
• Add more tests to improve code coverage
• Add new modules for OBS data processing (e.g., orientation of horizontal components, etc.)

Hi, developers, thanks for your excellent work and provding such a powerful tools!
Recently I am learning how to request and process obs seismic data using the OBStools, I encountered the same problem in two environments (one is the local server, another is the supercomputer cluster. Both are Linux system) when using query_fdsn_stdb to create the StDb Database.
ImportError: /lib64/libstdc++.so.6: version 'CXXABI_1.3.8' not found
I solved this problem by adding the following content to the ~/.bashrc file, which utilizes the libstdc++.so.6 in the conda environment.
export LD_PRELOAD=/xxx/software/anaconda3/envs/obseis/lib/libstdc++.so.6
I would like to ask if there is any other solution, especially for non-root users.
Hope you can reply.

Shortly the module obstools.atacr will be refactored to simplify the definitions of the classes and how to initialize the objects. In particular:

• Initialize a StaNoise object with a list of DayNoise objects
• Initialize a TFNoise object with either a DayNoise or StaNoise object

Thanks for sharing this great package. I found that to successfully install the package, you should added in the setup.py file:
packages=['obstools','obstools.atacr'].

Otherwise, the scripts under obstools.atacr cannot be found.

This is an easy fix.

Xiaotao Yang