A Pure-Python Robust Frequency Band Automatic Phase Picker for Seismic Monitoring
Abstract
We modify the FPPICK algorithm of Lomax et al. (2012) and implement an automatic phase picking algorithm implemented in Python. The algorithm takes advantage of existing seismological Python libraries, Obspy. The algorithm is designed to work on a variety of instrumentation and automatically adapts to different sampling rates. The time series signals are band-pass filtered for each band, octave, considered within the picker algorithm. The energy of the signal is calculated over an averaging window and multiplied by the instantaneous energy of the signal. This energy time-series is the statistic we can then examine for each frequency band considered. The summary statistic, which allows the identification of a trigger, is simply the maximum value of any frequency bands energy statistic at each sample time. A trigger is identified by using a control chart type statistic to identify when our statistics summary is changing rapidly and exceeds a specified number of standard deviations from the mean of the summary energy statistic. This has the advantage that the picker parameters don't necessarily need to be modified when processing data from a wide variety of instrumentation with different response characteristics. The algorithm also contains a method to determine the first motion direction associated with a pick as well as an uncertainty for the pick. As with any automatic phase identification system false picks can and do occur. A few simple algorithms are implemented to avoid false-picks, the picker can be configured not to include these checks. These algorithms remove picks that occur very close in time, and picks for which a phase has a smaller RMS than the previous time interval. The algorithm uses many techniques within Numpy to improve computation times. The algorithm effectively picks both P- and S-phase from local and regional earthquakes with only small amounts of picker parameter modifications. The picker can pick both P and S phases on local and regional earthquakes from a wide variety of instrumentation with no modifications by instrument type or sample rate. We demonstrate the robustness and effectiveness of this picker by comparing manual picked earthquake phase arrivals with those obtained from this picker. In addition, because the picker picks both P- and S-phase arrivals, pick association algorithms can be enhanced by the additional phase arrival picks. We demonstrate the effectiveness of a local earthquake phase associator algorithm written in Python.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFMIN51B1542C
- Keywords:
-
- 1906 INFORMATICS Computational models;
- algorithms;
- 1932 INFORMATICS High-performance computing;
- 1930 INFORMATICS Data and information governance