I would like to know a bit more about the theory the AR picker is based on
Is there any paper related to this picker or any further explaination
by the developers?
I found the documentation pages
but I couldn"t find background information.
(For the other picker pk_baer Baer 1987 is cited, something like that would be
From scholar &
Autoregressive (AR) methods (Sleeman and van Eck 1999; Leonard 2000) are used to determine a pick time after a phase onset has already been detected. AR methods work on the principle that a window of the time-series containing signal from a seismic event has different statistical properties than a window containing noise. Using one or more models for the form of the signal, these methods analyse different subwindows of the time-series to find an optimal point that separates a subwindow with statistical properties of noise from one of noise plus signal. This optimal point determines the pick time.
thanks for pointing out that the reference is missing! The implemented
picker is based on a (grey lit) paper by Takashi Akazawa which is
published with the title: A technique for automatic detection of onset
time of P- and S-Phases in strong motion records, "13th World Conference
of Earthquake Engineering, Vancouver, B.C., Canada, August 1-6, 2004,
paper no. 786. If you do not find it - let me know.
This citation and some more information is also being added to the
official documentation: https://github.com/obspy/obspy/pull/1559