Dear community!
I have RESP file and mSeed recorded on broadband seismometer (0.2 Hz to 50 Hz). How I can convert data from counts to m/s, m/s^2 or m?
Which preprocessing I have to apply?
Is it sufficiently use only “remove_response” function with filter or I have to do anything more?
pre_filt = (0.1, 0.2, 50.0, 55.0) # Is it correct???
ch1.remove_response(inventory=RESP_E, output=‘DISP’, pre_filt=pre_filt, plot=True)
Hi @microseism, what you are doing is correct, and is the same as what I would do. Although make sure you choose the the upper corners of your pre_filt to be below the Nyquist frequency (depends on your sample rate).
[shicks-seismo], thanks! I understand with upper range. And what about LF corners? Is it depends on gain-frequency characteristic of seismometer only?
And i have to detrend and demean signal before or after instrument correction?
In general all that to get maximum displacement/velocity/acceleration values and for further analysis of the H/V spectra.
Yes, if you see https://docs.obspy.org/packages/autogen/obspy.core.trace.Trace.remove_response.html, zero_mean is set to True by default, and a small amount of tapering is applied too.
In terms of the LF corners, I typically adjust the pre_filt parameters to adapt to the signal band I am interested in. It should be stable to filter beyond the flat response period, assuming your response poles and zeros are correct of course!
Also, it’s always a good idea to do all preprocessing (like instrument correction) on a larger time window and then cut to the desired time window right before processing (easiest/safest way to make sure you have no artifacts at start/end).