I am applying Obspy’s PPSD workflow to a ‘Guralp’ ‘CMG-6T’ ‘1s - 100Hz’ ‘2400’ instrument, for which I created the StationXML from the NRL. It turns out that the PPSD amplitudes are well over estimated. I have to change the instrument stage gain from 2400 to 2400000000 (a factor of 10^6 increase) for the PPSD to fall within the global noise models. Clearly, something is off in the way the intrument response is deconvolved based on the StationXML. Has anyone else encountered similar issues? Could it be a missing unit conversion? I am suspicious of this because the expected stage gain is ~10^6 higher than what is used in the NRL.
I have omitted using the data logger response in my stationXML, and my mseed time series data appear to be in the expected count range, i.e. -6000 < x(t) < 6000
I don’t have this issue when I deconvolve the instrument response in SAC using sac_pzs for the same instrument gain.
Thanks very much!
Research Fellow | Earthquake Seismology
School of Earth Sciences | McCoy Bldg. 200
University of Melbourne | Parkville 3010 VIC