Author: Eric Bergman

The Mandalay cluster is named for the city of Mandalay in central
Myanmar. The cluster was motivated by the destructive Mw 7.7 earthquake
on March 28, 2025, but it had to be handled in a fashion different from
most calibrated clusters, due to the scarcity of seismograph stations in
the region. By considering earthquakes in a much larger area than usual
(~700 km in latitude, ~400 km in longitude), it was possible to obtain
enough arrival time data at near-source and local distances to carry out
a direct calibration. A larger area than usual would have been required
in any case if the cluster was intended to represent the aftershock
sequence of the March 28, 2025 earthquake, as the event is estimated to
have ruptured a much longer, as much as 500 km, section of the Sagaing
Fault than expected for an earthquake of this magnitude. The cluster
includes a previous large event (7.5 Ms on September 12, 1946) on the
Sagaing Fault, to the north of the latest sequence, and several events
with magnitude >6 in the years between the 1946 and 2025 events. Because
of the great areal extent of the cluster and unmodeled variations in
crustal structure that perturb the travel times of local phases, the
location calibration is less precise than usual. This is reflected in
larger confidence ellipses than typical. Depth control for this cluster
is weaker than usual as well, since the depths of many events could only
be constrained by small numbers of reported teleseismic depth phases.
Eleven events, mostly prior to 1964 or aftershocks of the March 28
event, were set at a default value of 20 km, based on the mean of focal
depths for events with depth constraint. Because of the importance of
the seismicity in this region for earthquake hazard and seismotectonic
applications and the poor prospects for being able to ever achieve
higher accuracy in location calibration here it was decided to make
these results available in GCCEL. They should be utilized with due
appreciation for the larger uncertainties, as compared to most
calibrated earthquake clusters in GCCEL.

The Tomini cluster is named for the Gulf of Tomini, between the northern
and eastern peninsulas of the island of Sulawesi, Indonesia. Most of the
earthquakes are on or near the several portions of Sulawesi that define
the Gulf. The cluster contains the 7.5 Mw tsunamigenic earthquake of
September 28, 2018 that heavily damaged the port city of Palu, and five
other events with magnitude 6.0 or greater. The cluster is larger than
most in area, ~400 km across, in order to be able to include a number of
events whose hypocenters could not otherwise have been calibrated. All
events are well-recorded at teleseismic distances and all events have
depth control, from near-source and local distance readings, teleseismic
depth phases and waveform modeling.

The Lourdes cluster is named for the city of Lourdes, on the northern
flank of the Pyrenes Mountains in southern France. It includes several
earthquakes across the border in northern Spain. The largest event has
magnitude 5.3 mb on August 13, 1967. The cluster is limited to events
that were recorded to teleseismic distances but the amount of
teleseismic data is modest. The local network is extremely dense. All
events have depth control from near-source or local distance readings.

The Algiers cluster is named for the capital city of Algeria, on the
southern coast of the Mediterranean Sea. It includes two M6 earthquakes,
both a short distance off-shore, Mw 6.0 on October 29, 1989 and Mw 6.7
on May 21, 2003. All events are observed to at least 6° epicentral
distance and most are observed teleseismically. The seismograph network
is very dense and well-placed to calibrate this cluster, but it reports
virtually no S arrivals. The P-arrivals are very good, however, and
provide a strong location calibration. All events have depth control,
mainly from near-source and local-distance readings, but many also have
constraint from teleseismic depth phases. The region is quite
seismically active and many more events could have been included in the
cluster. The ones chosen are generally the largest and best-recorded,
and provide a robust and representative sample of seismic observations
(except Sg and near-regional Sn) from the source region.

The Haouz cluster is named for Al Haouz Province in central Morocco. The
cluster is based on the destructive 6.8 Mw earthquake on September 8,
2023 and consists mainly of the mainshock and larger aftershocks. The
aftershock sequence was very active but no events with magnitude greater
than 4.9 were reported. As a result the cluster contains only a modest
number of teleseismic observations. All events were recorded to
far-regional distances, however. The station distribution for location
calibration is fairly good, but it has an azimuthal gap to the NNW that
causes the confidence ellipse for the hypocentroid to be elongated in
that direction. All events have depth control, mainly from near-source
and local-distance readings, but the mainshock and two of the larger
aftershocks have depth control from teleseismic depth phases as well.
Both the local distance phase arrivals and the depth phases place the
mainshock focus unusually deep, around 25 km, consistent with the
identification of the steeply north-dipping nodal plane of the focal
mechanism as the fault plane, and with the depth range of faulting in
the finite fault model published by the NEIC.

The Tiran cluster is named for the Strait of Tiran and Tiran Island at
the northwestern end of the Red Sea. It includes earthquakes from the
northern Red Sea and the southern portions of the Gulf of Aqaba and the
Gulf of Suez. Local station coverage was poor until about 2004 but is
quite good since then. To achieve adequate location calibration it was
necessary to retain many small events that were observed only to
near-regional distances, but the cluster contains 19 earthquakes, up to
magnitude 5.1, that were recorded teleseismically. Unfortunately it was
not possible to include the Ms 6.6 earthquake of March 31, 1969 due to
insufficient connectivity with the modern events.

The Azua cluster is named for the city of Azua de Compostela on the
south coast of the Dominican Republic. The cluster contains several
earthquakes with magnitude greater than 5.0 but nothing larger than
M5.5. The local network was sparse prior to 2013 but is quite good since
then. The older events have depth constraint from teleseismic depth
phases, the recent ones have depth constraint from near-source and
local-distance readings. All events are recorded to at least
far-regional distances, and most are recorded teleseismically.

The Sosua cluster is named for the beach town of Sosua (Puerto Plata
Province) on the northern coast of the Dominican Republic. The cluster
includes a 6.4 Mw earthquake on September 22, 2003 and several other
events with magnitude greater than 5.0. Prior to 2017 there were few
seismograph stations in the area but events since then are well recorded
at close distances and with good azimuthal coverage. All events have
depth control. The oldest events have depth control from teleseismic
depth phases or waveform modeling; the more recent ones use near-source
and local-distance readings. Most of the earthquakes are recorded to
teleseismic distances but a few that were recorded only to near-regional
distances are retained to improve the location calibration.

An updated version of the Mentone, Texas cluster (mentone5) was uploaded, adding more recent events.

This version of the Mentone cluster replaces an earlier result
(mentone3) posted in GCCEL. This version is quite different in the
included events, which are required to have been observed to at least 10
degrees epicentral distance. The Mentone cluster is named after the town
of Mentone in west Texas, U.S.A. The cluster was originally motivated by
the occurrence of a 5.0 Mw earthquake on March 26, 2020 and the sequence
includes several other eevents near magnitude 5: 5.3 Mw on November 16,
2022, 5.2 Mw event on November 8, 2023 and 4.9 Mw as recently as
February 15, 2025. The region is a major producer of gas and oil and the
bulk of the seismicity is likely induced by excessive waste water
injection, as in Oklahoma and elsewhere. The earliest events in the
sequence are not included, due to lesser quality datasets. The
distribution of seismograph stations in the source region is now quite
dense; this version of the cluster was limited to events since January
2021. Azimuthal coverage and near-source data are exceptionally good for
all events. The entire cluster has been relocated with free depth but
the focal depths of a few events needed manual adjustment to better fit
the near-source data, so the final runs were done with fixed depth (at
the free-depth solution in most cases).

The Jamaica cluster is named for the island nation of Jamaica in the
Caribbean Sea. The arrival time data required a crustal model with quite
shallow Moho, about 20 km, and a single layer crust fits the data well.
However, the overlap between direct crustal phases and Moho-refracted
phases makes phase identification (and depth control) more challenging
than usual. The cluster contains two moderate-sized events, a 5.6 Ms
event on November 12, 1988 and a 5.5 mb event on January 13, 1993. Over
half the events are recorded teleseismically, but smaller events have
been retained to improve the statistical power for location calibration.
All events have depth control, mainly from near-source and
local-distance readings, but several are constrained by teleseismic
depth phases.