Author: Eric Bergman

The Zahedan cluster is named for the city of Zahedan in eastern Iran.
The cluster includes three earthquakes with magnitude 6.0-6.3 from the
Sefidabeh sequence in February 1994. The source region is remote,
inhospitable, lightly populated and sparsely instrumented, especially
prior to about 2009. The large 1994 events have no arrival time data
closer than about 500 km and including them slightly (by ~0.5 km)
degrades the quality of the location calibration, but it was judged to
be preferable to retain the abundant teleseismic arrivals these events
contain. The cluster has a larger geographic extent than is optimal in
order to obtain adequate azimuthal coverage for direct calibration, but
the location calibration still depends primarily on only three stations.
Because of the relatively small number of readings and stations used for
direct calibration and the small number of events in the cluster there
is greater than usual potential for location bias even through the fit
to the data is good. On the other hand, the calibrated locations for the
large 1994 events agree closely with the source model based on InSAR
data that was published in Parsons et al. (2006, GJI), unlike the
locations from an uncalibrated relocation using an early version of mloc
that were included in that paper. All events have depth control from
near-source and local-distance readings or teleseismic depth phases.

The Chagai cluster is based on the code names used for two nuclear tests
carried out by Pakistan on May 28, 1998 (Chagai-I) and May 30, 1998
(Chagai-II). Chagai-I was conducted in the Ras Koh Hills of Chagai
District, Baluchistan. Chagai-I is believed to have consisted of 5
simultaneous shots, but they were fired nearly simultaneously and are
treated as a single event. Chagai-II was located about 115 km southwest,
in the Kharan Desert of Baluchistan. The cluster includes both tests as
well as a dozen well-observed earthquakes in the region and uses the
locations of the nuclear tests for an indirect calibration of the
cluster. The calibration epicenter for Chagai-I is taken from
and
assumes a shot point ~1.3 km above sealevel. The calibration epicenter
for Chagai-II is taken from
> with a shot point ~600 m
below the surface, close to sea-level. The cluster hypocentroid is first
estimated from teleseismic P arrivals between 30 and 90°, and then the
cluster is shifted to best match the calibration hypocenters. The
epicentral shift (15.7 km at an azimuth of 109°) provides a close match
to the calibration epicenters of the two tests, with residuals of ~1.0
km and 1.6 km for Chagai-I and Chagai-II, respectively.

The main uncertainty in calibration concerns the shot times, which have
not been officially announced to sufficient accuracy. Of Chagai-I
Pakistani sources reported at the time of the tests that the firing
switch was activated exactly on the minute of 3:16pm local time (1016
UTC) but that was followed by a lengthy computer-controlled process for
actually firing the emplaced charges that is claimed to have taken 30
seconds or more. An origin time of ~1016:30 for this shot is not
tenable, however as the travel times recorded at regional and
teleseismic distance require the origin time to be close to 1016:15. The
nearest station with any reported arrival times from the cluster (QUE,
Quetta) is ~250 km away, too far to use the arrival times for origin
time constraint. The approach used here to calibrate the origin times is
unique to the Chagai cluster. A version of the cluster that included
earthquakes closer to Quetta was used to make an initial estimate of the
origin times and this was refined by comparing the baseline offset of
teleseismic P arrivals from the shifted Chagai cluster with that of the
nearest calibrated cluster, Ziarat, whose origin times are calibrated
with near-source and local-distance readings at station QUE. The
baseline offset for teleseismic P in the Ziarat cluster is +1.226 s, and
the calibration origin times of the two nuclear tests in the Chagai
cluster have been set (1016:15.45 for Chagai-I, 0654:55.30 for
Chagai-II) such that nearly the same baseline offset is observed for
teleseismic P (+1.257 s) and the relative time of the relocated nuclear
tests agrees within 0.04 s. Other estimates of the shot time seconds
field for these tests range from 15.2-17.8 for Chagai-I and 54.9-57.1
for Chagai-II, but none of these estimates qualifies as calibrated,
being based on uncorrected teleseismic data. Because of the uncertainty
involved in this calibration strategy the uncertainty of the origin
times of calibration events (i.e., the two nuclear tests) are inflated
to 1.0 s. The relative time of Chagai-I is referenced to zero focal
depth, while the true shot point was ~1.3 km above that depth, so the
true origin time of Chagai-I is estimated to have been ~0.2 s earlier
than the calibration time used here. For Chagai-II the shot point is
thought to be very close to sealevel so no correction is needed.

The earthquakes in the Chagai cluster include a 7.2 Mw event on January
18, 2011 and a 6.0 mb event on June 4, 2000. The earthquakes are about
evenly divided between events at crustal depths (10-20 km) and events
associated with the subducted slab from the Makran subduction zone, with
depths of 40-80 km. All earthquakes in this cluster have depth
constraint from teleseismic depth phases.

The Ceiba cluster is named for the town of Ceiba on the northeast coast
of Puerto Rico. It covers the eastern half of Puerto Rico and the
British and US Virgin Islands. The cluster is limited to earthquakes
less than about 60 km deep. The largest event is 5.6 mb on June 26,
1985. About half the events were recorded at teleseismic distances.
Smaller events, recorded only locally or to near-regional distances,
were retained to improve the statistics for location calibration. All
events have depth constraint from near-source and local-distance
readings, or teleseismic depth phases.

The Bordo cluster is named for the town of El Bordo in Cauca Department,
southwest Colombia. The cluster includes some offshore events, including
a 6.8 Mw earthquake on September 10, 2007. All other events have
magnitudes less than 5.3. Many events are small and recorded only to
near regional distances but they are retained to improve the statistics
for location calibration. All events have depth constraint from
near-source or local distance readings or teleseismic depth phases.

The Aysen cluster is named for the Aysén Fjord in southern Chile. The
region has a very low rate of seismicity despite the presence of the
Liquiñe-Ofqui Fault System, a long strike-slip fault that accommodates
some of the relative motion between Nazca and South American plates. In
April 2007 the area experienced two M6 events and a strong aftershock
sequence from which this cluster is assembled. Location calibration is
dependent on arrival time data from two temporary network deployments,
kindly provided by Hans Agurto Detzel and Denis Legrand. Only seven of
the 32 events are recorded at teleseismic distances. All events have
depth constraint from free-depth relocation.

The Nias cluster is named for island of Nias off the southwest coast of
North Sumatra, Indonesia. The selection of events went to ~150 km depth.
The cluster includes the 8.6 MW earthquake on March 28, 2005. It also
includes two events with magnitude >7, a 7.1 MS event on February 4,
1971 and a 7.8 MW event on April 6, 2010, and 20 earthquakes with
magnitudes between 6.0 and 6.9. Arrival time data from station SNSI on
Simeulue Island, GSI on Nias Island and PBSI on Tanahbala Island provide
unusually good azimuthal coverage for a subduction zone cluster. All
events have depth control from near-source and local-distance readings
and many events also have depth control from teleseismic depth phases.

The Umluj cluster is named for the city of Umluj on the Red Sea coast of
Saudi Arabia. The cluster is dominated by small events on shore but
includes a half dozen offshore events in the Red Sea. Most events are
small and observed only to local or near-regional distances but they are
retained in order to provide statistical power for calibrated location
of five events that were recorded teleseismically. The largest
earthquake is a 5.1 Mw event on May 19, 2009. All events have depth
control from near-source and local-distance readings.

The Oriente cluster is named for the Oriente Fault along the southern
coast of Cuba. It is larger in area than usual for GCCEL (~500 km
across) in order to include enough station coverage for stable location
calibration. The cluster includes the 7.7 Mw earthquake on January 28,
2020 and three M6 events: on May 25, 1992 (6.8 Mw), February 4, 2007
(6.2 Mw) and November 10, 2024 (6.8 Mw). In order to stabilize the
relocation many small events were retained. About 1/3 of the earthquakes
were recorded to teleseismic distances. All events have depth constraint
from near-source and local distance readings, as well as teleseismic
depth phases for some of the larger events.

The Sumbing cluster is named for Mount Sumbing, a strato-volcano in
central Java, Indonesia. The cluster includes shallow earthquakes on
land and also earthquakes in the subduction zone, up to ~100 km offshore
and to a depth of ~160 km. It includes three M6 earthquakes: on June 9,
1992 (6.1 Mw), May 26, 2006 (6.3 Mw) and January 25, 2014 (6.3 Mw). Most
events are recorded to teleseismic distances but a few smaller,
locally-recorded earthquakes are retained for improved statistics for
direct calibration. All events have depth constraint from near-source
and local distance readings, as well as teleseismic depth phases in many
instances. The distribution of stations for the events on land is very
good and the location calibration is robust, but like most subduction
zone clusters, offshore events tend to be less stable.

The Paria cluster is named for the Gulf of Paria between Venezuela and
Trinidad and Tobago. There are many relatively deep events immediately
to the north, associated with the Lesser Antilles subduction zone, but
when they are included in the cluster the lateral heterogeneity in
velocity structure causes location bias. The cluster is comprised of
shallow events (less than 35 km). There are two M6 events in the
cluster, on April 11, 1983 (6.1 Mw) and on June 11, 1986 (6.3 mw). The
distribution of seismic stations for direct calibration is quite good.
All events have depth control from near-source and local distance
stations and, in many cases, good teleseismic depth phase datasets. Most
events are recorded to teleseismic distances; the pattern of teleseismic
P residuals features a remarkable discontinuity near 65°.