Towards Assessing the Quality of Surface Wave Associations in the Reviewed Event Bulletin

. ABSTRACT
When monitoring for possible underground nuclear tests,
identifying shallow earthquakes from explosive sources can
be achieved using the ratio of the body-wave magnitude to
the surface-wave magnitude (mb:Ms criterion), with explosive
sources producing less energetic surface wave excitation.
Current methods for automated surface-wave detection at the
International Data Centre (IDC) rely on a dispersion test - a
global group-velocity model is used to predict a time window
based on event origins in the IDC Reviewed Event Bulletin
(REB). The data in the predicted time window are narrowband
filtered into eight frequency bands - if the time of the maximum
energy of at least 6/8 of the bands sits within a specified error
of the expected dispersion curves, a surface wave is said to be
detected. Stevens et al. (2001) added phase match filtering to
the process to improve the signal-to-noise ratio, and this was
implemented into provisional operations at the IDC in 2010,
under the name Maxpmf.

A number of issues can potentially arise with this automatic
detection technique, leading to false detections and mis-associations, these include:
• local noise passing the dispersion test and being erroneously associated;
• surface waves detected at close-to-regional distances
experience little dispersion and hence impulsive signals
can pass the dispersion test;
• since automatic detection is only attempted for REB
events, some surface waves may be missed entirely, as
they lack an origin from which to calculate an arrival-time
window.

Assuming random noise and that the signals are independent,
Stevens (2007) defined parameters that determine the false
alarm rate, determined empirically from the network as it was
in 2007. Stevens (2007) recommended that these parameters be
continually reviewed. Since automated surface wave processing
at the IDC was implemented, the number of International
Monitoring System (IMS) seismic stations with at least one
surface-wave detection in the REB has significantly increased
(from around 50 stations in 2002, to around 145 in 2020)
without review of the false alarm rate parameters.
We have designed interactive software to manually review
stages of the IDC automatic surface-wave detection algorithm.
We will use this to investigate the false-detection rate and
how it has changed over time, and interrogate whether the
independence and random noise assumptions this prediction is
predicated on are still valid for a larger network.

REFERENCES
Stevens, J. L., 2007. Automatic surface wave processing support
and documentation, Tech. rep., CTBTO Vienna International
Centre.
Stevens, J. L., Adams, D. A., & Baker, G. E., 2001. Improved
surface wave detection and measurement using phasematched filtering with a global one-degree dispersion model,
Tech. rep., Science Applications International Corp San
Diego CA.