West Coast/Alaska Tsunami Warning
Center
Operations Manual
NOAA/NWS/WCATWC
Palmer,
http://wcatwc.arh.noaa.gov
Section
4.1.13.1 – p-picker
Last
Updated: 5/2007
The P-pick algorithm in use at the WC/ATWC was developed by Veith (1978),
and the initial implementation was described by Sokolowski, et al. (1983). The attached diagram is a flowchart for the P-picker
routine. In general, the signal must pass through three phases before a pick is
declared.
The first stage is a simple test which looks for higher than normal
signal amplitude. A running average of the accumulated amplitude differences
between the samples (MDF) is updated every LTASeconds. The average MDF over
LTASeconds is averaged with the previous average MDF to obtain the new average
MDF. If the present MDF is greater than half the specified signal-to-noise
ratio for the station times the average MDF (the trigger threshold), phase 1 is
passed. The time of this sample is saved as the P-time in case all phases pass.
The three tests of phase 2 are all conducted simultaneously. Test 1
requires energetic signal by testing for multiple passing of phase 1. That is,
the trigger threshold in phase 1 must be exceeded lNumConsec times before
3*lNumConsec samples pass (lNumConsec = Sample Rate / (2 * Minimum Frequency of
interest)). Test 2 verifies that the signal strength exceeds the set
signal-to-noise ratio times the average signal level sometime during phase 2.
Test 3 checks if the trigger threshold was not exceeded more than half of the
time at any point during phase 2. If test 3 is true, all variables are reset
and phase 1 must be passed again
Phase 3 requires the signal to be oscillatory. This phase must be
completed before 12 * lNumConsec samples pass. The signal's MDF must exceed the
trigger threshold six times in opposite directions (which requires three full
cycles of signal). If the MDF is below the trigger threshold more than half the
time it takes to pass the oscillation test and is at least four seconds under
the threshold, phase 3 fails. If phase 3 passes, a pick is declared and the
time of phase 1 passage is declared the P-time.
Veith's (1978) algorithm also has a procedure for terminating the event
based on signal strength. In pick_wcatwc, this is neglected since the signal is
processed for Mb, Ml and Mwp magnitudes for a set time after the P onset. The
pick is terminated and variables reinitialized when all magnitude information
has been gathered for that P-pick.
A few processes have been added to the original P-pick algorithm of
Veith. Real-time magnitude information is processed as soon as phase 1 is
passed. The maximum peak-to-trough amplitude (MDF) in the first MbCycles
half-cycles of signal after the P onset is used to determine Mb. The number of
samples is doubled to compute period. The signal for a total of LGSeconds,
after the first MbCycles half-cycles, is processed to determine the maximum
amplitude for Ml. One hundred fifty seconds allows Ml computations for up to approximately
nine degrees epicentral distance. Automatic Mwp computations (Tsuboi, et al.,
1995; Tsuboi, et al., 1999; Whitmore, et al., 2002) are performed MwpSeconds
after the P arrival on broadband seismic data. One other function added to the
original algorithm is automatic station calibration for those WC/ATWC network
short period stations equipped with a sine-wave calibrator. The 60s, 1 Hz sine
wave calibrations are automatically triggered twice a day. The P-picker will
pick these as a P wave. If the first six full cycles of signal have the
characteristics of a calibration pulse (that is, 1 Hz frequency and constant
peak MDF) the pick is declared a calibration and the amplitude can be used to
set the station's magnification.
After a pick is made, the pick information is logged and reported to the
PICK_RING in the PICKTWC and in the PICK_GLOBAL Earthworm message formats.
Magnitude information is added as it is computed. The pick can be altered
interactively using modules develo or hypo_display. Pick information (time and
magnitude data) are re-reported to the PICK_RING as pick_wcatwc updates the
magnitude information or it is interactively changed in another module.
Typically, the larger the earthquake, the more accurate the P-picks.
Picks are normally determined within a few samples for distinct arrivals;
whether from local earthquakes or teleseisms. Very emergent P waves can produce
erroneous automatic P-picks. For this reason an easy method of adjusting picks,
as provided by modules develo and hypo_display, is necessary. The P-pick
algorithm works best on short period data or short period filtered, broadband
data. It was tested unsuccessfully on raw, broadband data.
References
Sokolowski, T.J., G.W. Fuller, M.E. Blackford, and W.J. Jorgensen (1983).
The Alaska Tsunami Warning Center's automatic earthquake processing system, in Proceedings, 1983 Tsunami Symposium, Hamburg, FRG,
August, 1983, 131-147.
Tsuboi, S., K. Abe, K. Takano, and Y. Yamanaka (1995). Rapid
determination of Mw
from broadband P waveforms, Bull. Seism.
Soc. Am., 85,
606-613.
Tsuboi, S., P.M. Whitmore, T.J. Sokolowski (1999). Application of Mwp to deep and teleseismic
earthquakes, Bull. Seism. Soc. Am.,
89, 1345-1351.
Veith, K.F. (1978). Seismic signal detection algorithm, Teledyne Geotech
Technical Note 1/78, 10 pp.
Whitmore, P.M., S. Tsuboi, B. Hirshorn, and T.J. Sokolowski (2002).
Magnitude-dependent correction for Mwp,
Science of Tsunami Hazards Journal,
20, 187-192.