You are on page 1of 21

Geomatics, Natural Hazards and Risk

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tgnh20

High-rate GNSS data in seismic moment


tensor inversion: application to anthropogenic
earthquakes

Iwona Kudłacik, Anna Tymińska, Grzegorz Lizurek, Jan Kapłon & Jacek
Paziewski

To cite this article: Iwona Kudłacik, Anna Tymińska, Grzegorz Lizurek, Jan Kapłon & Jacek
Paziewski (2023) High-rate GNSS data in seismic moment tensor inversion: application to
anthropogenic earthquakes, Geomatics, Natural Hazards and Risk, 14:1, 2232084, DOI:
10.1080/19475705.2023.2232084

To link to this article: https://doi.org/10.1080/19475705.2023.2232084

© 2023 The Author(s). Published by Informa


UK Limited, trading as Taylor & Francis
Group.

View supplementary material

Published online: 10 Jul 2023.

Submit your article to this journal

Article views: 474

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tgnh20
GEOMATICS, NATURAL HAZARDS AND RISK
2023, VOL. 14, NO. 1, 2232084
https://doi.org/10.1080/19475705.2023.2232084

High-rate GNSS data in seismic moment tensor inversion:


application to anthropogenic earthquakes
skab
Iwona Kudłacika , Anna Tymin , Grzegorz Lizurekb , Jan Kapłona
c
and Jacek Paziewski
a
Institute of Geodesy and Geoinformatics, Wrocław University of Environmental and Life Sciences,
Wrocław, Poland; bInstitute of Geophysics Polish Academy of Sciences (IG PAS), Warsaw, Poland;
c
Department of Geodesy, University of Warmia and Mazury in Olsztyn, Olsztyn, Poland

ABSTRACT ARTICLE HISTORY


Earthquakes are traditionally monitored by seismic networks. Received 14 November 2022
However, the progress in high-rate Global Navigation Satellite Accepted 27 June 2023
Systems (GNSS) observations caused them to be included as a
KEYWORDS
standard supplementary tool for strong natural earthquake moni-
High-rate GNSS; moment
toring. Here, we demonstrate that the displacement time series tensor inversion; mining
obtained with high-rate GNSS data can be included as a supple- tremors; anthropogenic
mentary tool for the characterization of low-magnitude anthropo- earthquakes; seismogeodesy
genic earthquakes to monitor the dangerous impact that such
tremors may have on infrastructure. We analyzed two mining
tremors with magnitudes of 3.7 and 4.0, respectively, and utilized
the spectral amplitudes of seismic and high-rate GNSS observa-
tions in the seismic moment tensor calculation. Our study reveals
that the high-rate GNSS time series can be successfully included
in moment tensor calculations and might also be crucial if there
is a lack of seismic data.

1. Introduction
Mining tremors induce clearly visible effects of underground mining activities. In
Poland, we can distinguish three areas where felt anthropogenic earthquakes occur:
the Upper Silesian Coal District and Legnica-Głog ow Copper District (LGCD) in
Southern Poland, and the Bełchatow Lignite Mine in Central Poland. The intensities
of Polish mining tremors in underground mining areas are similar to those of small
natural earthquakes. High-energy mining tremors are felt on the Earth’s surface, espe-
cially in areas with concentrated buildings, and they are also a potential source of
damage. The problem of determining mining tremors’ impact on structures and the
Earth’s surface is very complex due to the fact that the rock mass is often diverse;

CONTACT Iwona Kudłacik iwona.kudlacik@upwr.edu.pl


Supplemental data for this article can be online at https://doi.org/10.1080/19475705.2023.2232084
ß 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://
creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any
medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the
Accepted Manuscript in a repository by the author(s) or with their consent.
2 I. KUDŁACIK ET AL.

this impact also depends on the nature of the seismic event. The maximum local mag-
nitudes of mining-induced seismicity in the mining regions in Germany and the
United Kingdom rarely exceed 3.5 (Bischoff et al. 2010; Wilson et al. 2015). One of the
strongest tremors in Germany reached a local magnitude of 5.6 (Cesca et al. 2013). In
Poland, tremors with local magnitudes in the range of 3.5–4.5 are more frequent. The
intensities of mining-induced events are comparable to those of small natural earth-
quakes. In the LGCD mining area, reverse and normal faulting are observed; no strike-
slip faults are expected there. The use of the amplitude and polarity of P-wave moment
tensor (MT) inversion is universal enough to solve all types of faulting, as well as col-
lapses, pillar bursts, or any other typical mining-related focal mechanisms. The mining
tremor mechanisms depend strongly on the geology and mining conditions (Gibowicz
1990). Stress and strain changes determine each particular event and its mechanism. A
convenient and common method for focal mechanism determination is the MT inver-
sion. The most common decomposition of the MT for both natural and anthropogenic
earthquakes divides the mechanism into double-couple (DC), compensated linear vec-
tor dipole (CLVD), and isotropic parts (Jost and Herrmann 1989). This MT decompos-
ition allows the rapid determination of the event’s character and indicates if the
shearing or non-DC mechanism is overweighted. For natural events, non-DC compo-
nents are considered to be the effect of rock mass anisotropy or due to the influence of
noise. For anthropogenic events, where isotropic changes or uniaxial tension and com-
pression are present because of human activity, decomposition is used to determine the
prevailing mechanism and its complexity. Mining tremors have a bimodal distribution,
which distinguishes ‘mining’-induced events associated with mining operations and
located near or within excavation areas, and the triggered events related to discontinu-
ities and residual tectonic stress (Gibowicz 1990; Whidden and Pankow 2012;
Rudzi nski et al. 2016; Lizurek 2017; Sılenỳ and Milev 2017). The triggered events are
characterized by high DC components with nodal planes oriented similarly to existing
tectonic discontinuities. Induced events have small shearing components and nodal
planes can be associated with the mining front orientation or excavated geometry (Sen
et al. 2013; Lizurek 2017).
The LGCD area is located in the south-west part of the Fore-Sudetic Monocline,
which is strongly disturbed by numerous tectonic dislocations. The copper ore is sur-
rounded by hard rocks, which accumulate the strain energy and, in combination with
tectonic conditions, cause seismic events (Orlecka-Sikora et al. 2014) and in-ground
subsidence (Ilieva et al. 2020). Copper-ore exploitation in the LGCD is performed at
depths of 800 to over 1200 m beneath the Earth’s surface. Despite the regular seismic
events and various depths of extraction, the frequency of tremors does not increase
with the depth of mining (Bartlett et al. 2013). The mining tremors are often felt on
the surface, and the high-energy events result in damage to buildings. From January
2014 to the end of February 2021, in the LGCD area, there were over 450 earthquakes
with magnitudes (Mw) in the range 2.5–4.0 registered in the database available from
the EPISODES platform; 12 of these earthquakes had magnitudes of over 3.5
(EPISODES Platform 2017). The characteristics of the events registered in the LGCD
area indicate that most events have high non-DC components; this was observed by,
e.g. Rudzi nski et al. (2016), Lizurek (2017), and Ilieva et al. (2020).
GEOMATICS, NATURAL HAZARDS AND RISK 3

Generally, the mining areas are monitored with standard methods, such as seismic
networks used to detect and characterize earthquakes (Kijko 1977; Swanson et al.
2009; Lizurek et al. 2015; Yang et al. 2018; Johnson et al. 2021), and periodic geomet-
ric levelling is carried out to detect long-term deformations (Blachowski et al. 2009;
Furst et al. 2021). Today, remote sensing techniques, such as Interferometric
Synthetic Aperture Radar (InSAR), are also being used to describe long-term defor-
mations (Ilieva et al. 2019; 2020; Yang et al. 2020; Antonielli et al. 2021; Pawluszek-
Filipiak and Borkowski 2021). In addition, it is possible to record ground vibrations
with high-rate GNSS (HR-GNSS) data. This method is a standard technique for nat-
ural earthquakes; in particular, it is used to supplement seismic data for earthquake
characterization and early warning systems (e.g. Benedetti et al. 2014; Psimoulis et al.
2018; Li et al. 2019). This application of an HR-GNSS usually concerns vibrations
with multi-centimetre amplitudes. Nevertheless, the progress in GNSS processing
techniques (Bock et al. 2011; Paziewski et al. 2018), as well as the advances in receiver
signal acquisition algorithms (Hussain et al. 2021; Iakovidis et al. 2022), may make it
possible to record the mining tremors of smaller amplitudes with an HR-GNSS, as
well as provide complementary sensors that can be used alongside seismic instru-
ments (Saunders et al. 2016; Paziewski et al. 2019; 2020; Kudłacik et al. 2021). The
possibility of using seismogeodetic data for natural earthquakes with magnitudes of
4.0–4.2 at short epicentral distances was confirmed by Saunders et al. (2016). The
integration of seismogeodetic results, in particular by using the Kalman filter to
improve the compatibility of seismic-GNSS stations, has been demonstrated in the lit-
erature (Bock et al. 2011; Hohensinn and Geiger 2018; Shu et al. 2018); however, this
remains beyond the scope of this study because we aimed to analyze the impact of
replacing SM stations with GNSS stations to determine the mining tremor
mechanism.
Although the recording frequency and sensitivity of HR-GNSS stations are lower
than those of seismic sensors, HR-GNSS stations can capture the waveforms prop-
erly (Genrich and Bock 2006; Smalley 2009; Michel et al. 2017; Ruhl et al. 2019).
Generally, the smaller amplitudes generate seismic waves of higher frequencies, but
on the other hand, the mining tremors are characterized by lower-frequency con-
tent across the entire waveform (Swanson et al. 2009). Moreover, the HR-GNSS
can record not only short-term ground vibrations but also the long-term co-seis-
mic displacement (Li et al. 2014). The reliability with which an HR-GNSS can
retrieve the waveforms was confirmed with MT inversions for large-magnitude
natural earthquakes, including the Mw 7.2 El Mayor-Cucapah and Mw 8.3
Tokachi-oki earthquakes, which resulted in very high deformations (Melgar et al.
2012; O’Toole et al. 2013; K€aufl et al. 2014). On the other hand, mining tremors
usually result in small amplitudes, which only slightly exceed the noise level of
the HR-GNSS coordinate time series, and as a result, the HR-GNSS was never
considered for MT inversion for such small events. In this paper, taking advantage
of the HR-GNSS data recorded during mining tremors in the LGCD area, we
applied the HR-GNSS results to MT inversion to determine the effectiveness of
combining two independent datasets to resolve small anthropogenic earthquake
mechanisms.
4 I. KUDŁACIK ET AL.

2. Materials and methods


Seismic activity is continuously monitored by traditional seismic sensors that are
located on exploration levels and the surface. The local seismological surface network,
called LUMINEOS, comprises 10 accelerometers and 18 seismometers. Additionally,
since 2018, in the LGCD area, there has been a continuously operating high-rate
multi-GNSS station (LES1), co-located with the seismic station KOMR. This station
aims to monitor the long-term subsidence and detect ground vibrations caused by
the mining industry. In 2019, two other HR-GNSS stations were also operating in
this area; they were also located near other seismic stations (TRZB, TARN). Two of
the few tremors recorded by the HR-GNSS technique were selected for this analysis.
First, one of the strongest mining tremors in the LGCD in recent years was ana-
lyzed—this Mw 3.7 event was recorded on 29 January 2019 at 12:53:45 UTC. This
tremor was located at 800 m depth and resulted in the collapse of 600 m of the mine
corridor. The second analyzed tremor was that with the highest magnitude value in
recent years—it had a magnitude of Mw 4.0 and an unusual depth of 2255 m, which
is almost 1 km below the mining level. This event was recorded on 8 July 2020 at
05:18:59 UTC.
Two types of sensors, seismic instruments (SMs) and HR-GNSS stations recorded
both analyzed tremors, as presented in Figure 1. During the 2019 event, 15 seismom-
eters and two accelerometers were active, and in 2020, 11 seismometers and three
accelerometers were active. The HR-GNSS observations of the 2019 event were per-
formed by three stations co-located with strong-motion sensors. One GNSS station
(LES1) collects data with a frequency of 10 Hz, while two others (TARN and TRZB)
collect data with a frequency of 50 Hz. The details concerning the sensors are

Figure 1. Map of earthquake epicentres and measurement sensor locations, including GNSS sta-
tions and selected stations of the seismic network that were active during the events.
GEOMATICS, NATURAL HAZARDS AND RISK 5

Table 1. Characteristics of co-located GNSS-SM sensor pairs used in this research.


SM sensor GNSS SM-GNSS
sensor
Sample Sample separation
Name Type rate (Hz) Name Antenna/radome Receiver rate (Hz) (km)
KOMR GeoSIG AC-63 250 LES1 LEIAR20 LEIM LEICA GR30 10 0.002
TRBC GeoSIG AC-63 250 TRZB TRM57971.00 NONE TRIMBLE NETR9 10 0.045
TRN2 GeoSIG VE-53/BB 100 TARN TRM57971.00 NONE TRIMBLE NETR9 50 0.316

Table 2. Epicentral distance, positioning accuracy (for filtered HR-GNSS displacements), and peak
ground displacement (PGD) value of co-located sensors.
Accuracy of displacements (mm)
Epicentral distance PGD vertical
Date Sensor Name (km) E N U PGD (mm) (mm)
2019-01-29 GNSS LES1 2.5 0.4 1.1 1.6 9.0 6.6
SM KOMR 0.014 0.003 0.004 9.0 5.0
GNSS TRZB 1.0 1.0 2.2 2.7 14.9 10.6
SM TRBC 0.013 0.009 0.007 14.5 6.6
GNSS TARN 3.3 1.2 1.9 2.3 8.6 8.6
SM TRN2 0.003 0.004 0.001 2.2 1.6
2020-07-08 GNSS LES1 1.4 0.5 0.9 1.7 17.3 10.6
SM KOMR – – – – –
Positioning accuracy was calculated as a standard deviation of displacements in the 10-s moving window within
2 min before an event.

provided in Table 1. However, considering the frequency content of the ground


vibrations during the mining tremors, the 50 Hz observations were down-sampled to
10 Hz. On the other hand, the 2020 event was observed with only one HR-GNSS sta-
tion; this station is co-located with the strong-motion sensor, which unfortunately
was not working at that time. The locations of the analyzed tremors, HR-GNSS sta-
tions, and seismic stations are presented on the map in Figure 1. As listed in Table 2,
the co-located stations have inter-sensor distances of hundreds of metres (i.e. the
maximum distance is from TARN to TRN2).
The HR-GNSS observations provided an excellent opportunity to analyze the
impact of the using HR-GNSS observations for an MT calculation with real (non-
simulated) data.

2.1. HR-GNSS data processing


Since the common GNSS systems for all three analyzed HR-GNSS stations were
GPS þ Galileo, and the high-rate clock corrections for Galileo were unavailable during
the calculations, we decided to use only the GPS observations to obtain consistent
results. The GPS observations were processed to obtain the displacement time series
using the RTKlib v2.4.3 software (Takasu 2009) with the kinematic Precise Point
Positioning (PPP) approach. The processing parameters included final orbits and 5-s
clock corrections provided by the Center for Orbit Determination in Europe (CODE)
(Dach et al. 2020). The ionosphere-free combination and Global Mapping Function
were used to reduce the atmospheric errors. The PPP results were validated by com-
paring them to solutions from the online CSRS-PPP application (The Canadian
Geodetic Survey of Natural Resources Canada 2021). They showed an excellent
6 I. KUDŁACIK ET AL.

Table 3. Pearson’s correlation coefficients between filtered GPS displacements obtained with two
pieces of software, RTKlib and CSRS-PPP, calculated for each event and station.
Earthquake GNSS station name E N U
2019-01-29 LES1 1.00 0.99 0.98
2019-01-29 TRZB 1.00 0.98 0.99
2019-01-29 TARN 0.99 0.99 0.98
2020-07-08 LES1 0.99 0.98 0.98
Median correlation: 1.00 0.99 0.98

Figure 2. Filtered GPS-displacement (left) and GPS-velocity (right) time series for station LES1 on
2020-07-08. The sampling frequency is 10 Hz. Note that in the right panel (velocity), the vertical
scales for the east and up components are identical, whereas, for the North component, the scale
is different due to higher velocity values. The vertical dashed line denotes the earthquake time
according to the seismic catalogue.

agreement, with a Pearson’s correlation coefficient of 0.99 during the tremors


(Table 3). The position changes calculated with PPP were then converted into dis-
placements (denoted here by the prefix ‘GPS’). The final GPS-displacement time ser-
ies was obtained after applying the 2nd-order Butterworth band-pass filter to remove
the high- and low-frequency noise. The analysis of the GPS-displacements and the
detailed GPS-derived displacement time-series processing methodology and a com-
parison with the seismological data were presented by Kudłacik et al. (2021). The fil-
tration of the HR-GNSS time series for the GPS-displacements of both analyzed
earthquakes was performed in the same frequency window (0.15–2.00 Hz), based on
the dominant frequency analysis of strong-motion displacements (denoted here by
the prefix ‘SM’) of the 2019 event. In the rest of this paper, only the filtered GPS-
displacement time series are considered. The relative accuracy of the filtered displace-
ments is presented in Table 2. The visualization of the filtered GPS position changes
for the 2020 event is shown in Figure 2, and it is shown for the 2019 event in
Supplementary Figure 1. The filtering procedure significantly decreases the noise level
(Kudłacik et al. 2021); however, the residual high-rate noise remains. Nevertheless,
the correlation between the GPS- and SM-displacement time series reaches over 0.8
for the horizontal components and over 0.6 for the vertical components (Kudłacik
et al. 2021). The correlation level might be a consequence of the GNSS noise and the
fact that the HR-GNSS stations are located on rooftops and the SM stations are
GEOMATICS, NATURAL HAZARDS AND RISK 7

located in basements. Additionally, note that TRZB and TARN have non-zero separa-
tions of 45 and 316 m, respectively.

2.2. Spectral amplitude determination


The double integration of accelerometric observations and the integration of seismo-
metric recordings resulted in the SM-displacement time series. To overcome the pos-
sible baseline errors, the integration was followed by high-pass filtering with a cut-off
frequency of 0.015 Hz. These time series, together with the GPS-displacement time
series, were used to determine the first arrival of P-wave displacement in the vertical
direction for a set of stations. It is generally known in the geodetic community that
the vertical component has greater noise than the horizontal component. Here, the
filtered horizontal displacements are characterized by a standard deviation of about
1 mm and the filtered vertical displacements have a standard deviation of about
2 mm, as shown in Table 2, and for the analyzed events, the peak ground displace-
ments reach about 9 mm. Since the amplitudes of mining tremor vibrations are only
slightly greater than the filtered GPS-displacement noise, the first arrival of a seismic
wave is difficult to detect directly from the displacement time series. Therefore, the
spectral levels were obtained and included in the MT inversion. The vertical GPS-
and SM-displacement time series were used for both events. The single-sided ampli-
tude spectra of the vertical components that are included in further MT calculations
are presented in Figure 3, which shows the correspondence between the amplitude
spectra derived from the HR-GNSS and SM data for the 2019 event. The median cor-
ner frequency of the vertical displacements for the 2019 event is 0.56, and for the
2020 event, it is 0.59. Thus, considering the Nyquist frequency, the HR-GNSS data
sampling of 10 Hz is enough to capture the displacements adequately. Figure 4
presents the spatial distribution of stations and the spectral levels of the displace-
ments. Moreover, one can notice that for the 2019 event, the strong motion stations’
spatial distribution is better than it is for the 2020 event.

2.3. Moment tensor inversion


The seismic MT is a nine-element matrix describing the body forces acting in an
earthquake source. The displacement registered at seismic stations is connected to the
MT as follows (Shearer 2009):

@Gij ðx, t; x0 , t0 Þ
ui ðx, t Þ ¼ Mjk ðx0 , t0 Þ,
@ ðx0 Þk

where u denotes the measured first peak displacement, Mjk is the moment tensor
matrix with six independent components, and Gij is the Green’s function describing
the rock mass response to waves propagating from the source.
The MT is calculated by resolving a set of linear equations. The minimum number
of stations that can be used to invert the MT is six since there are six independent
MT components. The software that we used, focimt (Kwiatek et al. 2016), has a min-
imum number of eight stations, set by the authors of this software. The optimal MT
8 I. KUDŁACIK ET AL.

Figure 3. Single-sided amplitude spectra of vertical components are included in the MT calcula-
tions. The left panel shows the amplitude spectra of vertical GPS and SM displacements for the
2019 event, and the right panel shows these spectra for the 2020 event. For the 2020 event, the
co-located SM data were unavailable.

solution gives the minimum root-mean-square (RMS) error, which is defined as the
absolute difference between the synthetic and measured displacement amplitudes. The
most common method of MT calculation, especially for small earthquakes, is to use
the amplitudes of seismic body waves in a time-domain inversion. For example, the
first P-wave ground-displacement pulse inversion can also be used with a spectral
amplitude (Kwiatek et al. 2016). The most popular decomposition of the MT matrix
provides a DC component for pure shear, the CLVD for uniaxial compression or ten-
sion, and an isotropic (ISO) component for volume changes (Jost and Herrmann
1989). The share of individual components is given as a percentage, with the ‘’ sign
indicating uniaxial compression or isotropic implosion and the ‘þ’ sign indicating
uniaxial tension or isotropic explosion. A comparison of nodal planes and tension/-
compression axis directions for the ISO, CLVD, and DC solutions makes it possible
to determine the stability of the entire solution. Using this kind of decomposition
will enable us to assess an event’s character. High DC values indicate shearing on a
fault, while high CLVD and ISO values are characteristic of, in essence, induced or
volcanic events (Gibowicz 1990). Based on the seismic moment tensor, conclusions
about the directions and strength of the forces in the foci and about the character of
the event are drawn.
In this study, the spectral amplitudes of 17 seismic stations were provided for the
2019 event. For the 2020 event, spectral amplitudes were supplied from 15 seismic
stations. The polarity of the first P-wave peak was applied with spectral amplitudes
for the MT inversion. The one-dimensional velocity model used in the calculations is
available on the EPISODES Platform (2017). We performed the Jacknife single station
rejection test for both events to find the stations that were crucial for a stable and
reliable MT inversion. During each test, bootstrap data from one of the considered
Figure 4. Amplitude spectra of vertical time series used for the MT calculation. The left panel shows the 2019 event and the right panel shows the 2020
event.
GEOMATICS, NATURAL HAZARDS AND RISK
9
10 I. KUDŁACIK ET AL.

stations were excluded from the calculations. Then, stable results (where nodal planes
for full, isotropic, and DC solutions have the same direction) with physically correct
MT solutions were further considered to determine which stations should be taken
into account for the final calculation. Stations with seismometric data that were fre-
quently excluded from accurate solutions in the Jacknife test were also excluded in
the final MT calculation. To decrease the RMS error, stations located near each other
on the focal sphere were also removed because of amplitude differences in close loca-
tions. We calculated the rotation angles between the SM and mixed solutions to
evaluate the influence of the HR-GNSS data. The rotation angle is the smallest angle
between two DC mechanisms. We performed the MT inversion for just SM-displace-
ment spectral amplitudes and different seismic and HR-GNSS data combinations, as
described in subsection 3.1.

3. Results and discussion


3.1. Finding the MT solution
We calculated the focal mechanism for the 2019 event based on the displacement
amplitudes from 12 seismic stations and three GNSS stations for various combina-
tions: A  12 SMs, no GPS (reference solution); B  11 SMs, sequentially removing
SM data from co-located sensors; C  11 SM stations þ 1 GNSS station, sequentially
including the corresponding GNSS sensor instead of the removed SM sensor; and
D  9 SM stations þ 3 GNSS stations, which replaced three co-located SMs. These
combinations are presented in Figure 5 (limited to the full MT inversion only) and in
Figure 6, where the deviatoric and DC solutions and additional combinations of 10
SM stations þ 2 GNSS stations are shown. The basic MT solution with the use of
only seismic data (combination A, Figure 5) was stable, with the prevalence of the
CLVD component, which is in accordance with the roof collapse that followed the
event. There are three significant stations for this study in Komorniki (KOMR),
Trzebcz (TRBC), and Tarn owek (TRN2), where the strong-motion station is co-
located with the GNSS station. To determine if GNSS data can be used together with
seismic data for the MT inversion, primarily the influence of the lack of data from
each of the co-located stations was analyzed (combination B, Figure 5). Second, the
seismic data from the co-located stations were replaced by GNSS data (combinations
C and D, Figure 5). We present the details of the full, deviatoric, and DC solutions
for this event in Figure 6. The RMS error of the MT solution is low or very low for
almost all cases, and all solutions are stable, indicating the solutions’ reliability. The
RMS error reaches 0.17 for a solution with 12 SM stations and 0.20 for a solution
with 9 SM þ 3 GNSS stations. For mixed combinations (e.g. 11 SM stations þ 1
GNSS station or 10 SM stations þ 2 GNSS stations), the RMS error value is between
0.14 and 0.29. For calculations with HR-GNSS data in place of seismic records, the
MT solutions are consistent with other results. The rotation angle between the nodal
plane solutions with only seismic data and those using the HR-GNSS data is below
23 . In this case, the Jacknife station rejection test showed the instability of solutions
without the TRBC station (combination B3, Figure 5). This means that it is crucial to
include the TRBC station, which is closest to the epicentre, in the MT inversion.
GEOMATICS, NATURAL HAZARDS AND RISK 11

The ability to apply HR-GNSS data stabilizes the solution in the case of 11 SM
stations þ TRZB (combination C3, Figure 5). With HR-GNSS data from the TRZB
station, the MT decomposition is close to the solutions calculated using seismic data;
additionally, the RMS error is low, and the rotation angle is negligible, which indi-
cates the good quality of the solution. For this event, the LES1/KOMR station was

Figure 5. Full MT solutions with different sets of seismic and GNSS data for the 2019 event: (A)
the solution with 12 SM stations; (B1–B3) the impact of removing particular SM stations; (C1–C3)
the solutions for which particular SM stations were replaced with GPS-PPP results; (D) the solution
calculated with data from nine SM stations and three GPS stations.
12 I. KUDŁACIK ET AL.

Figure 6. Detailed MT solutions for the 2019 event with different sets of seismic and HR-GNSS
data.

not essential for the solution because of event characteristics and station deployment
(combination B1 vs. combination C1, Figure 5). However, it is worth noting that
applying the HR-GNSS data from TRN2 in place of seismic data makes the solutions
highly DC (combination B2 vs. combination C2, Figures 5 and 7(A)), which is
improper for roof collapse. The non-DC component is unstable in the solutions; this
is shown in the source-type plot (Figure 7(A)). This instability is not strictly related
to using three HR-GNSS stations in the solution; instead, it is related to the general
features of the MT inversion for such a shallow event recorded by the LUMINEOS
GEOMATICS, NATURAL HAZARDS AND RISK 13

Figure 7. Source-type plots with solutions for various datasets of events: (A) 2019 and (B) 2020.
The horizontal axis presents CLVD participation in the solution, and the vertical axis presents the
isotropic component.

Figure 8. Detailed MT solutions for the 2020 event with two different sets of seismic and HR-GNSS
data.

surface network, as shown by Tymi nska and Lizurek (2022). The non-DC compo-
nents from such an inversion are very sensitive to both noise and the focal coverage,
which is also the case in our work. On the other hand, the general geometry of the
focal mechanism is well resolved in all cases compared to the MT inversion solution
obtained with regional broadband stations published by Ilieva et al. (2020). The full,
deviatoric, and DC solutions are similar, which shows that GNSS data may be added
to the MT inversion in place of data from seismometers if necessary, without a sig-
nificant loss of the solution quality. However, this is valid only for events with magni-
tudes over 3.5 for which GNSS stations are close enough to the seismic source in the
mining areas. In general, this suggests that if we would like to monitor the dynamic
surface effects and focal mechanisms of high-energy mining seismic events, GNSS sta-
tions located near the source area may provide significant additional information that
can improve the quality of the MT inversion.
14 I. KUDŁACIK ET AL.

One GNSS station, LES1, registered the second event, which occurred in 2020 and
had a local magnitude of 4.0. Unfortunately, the nearby seismic station KOMR was
disabled, and all the utilized data are from stations located west and south-west of
the station, resulting in inadequate focal coverage. Therefore, MT solutions were cal-
culated with 13 seismic stations and with the additional LES1 station based on spec-
tral amplitudes. Both solutions have the same RMS error value of 0.28, which is
higher than that of the previous event. This may have been caused by unfavourable
focal coverage. The mechanisms are stable, and the nodal plane orientations are simi-
lar in both the seismic and combined (SM þ GNSS) moment tensor calculations, as
the calculated rotation angle value is low. The mechanism according to the HR-GNSS
data involves more shearing than the solution from the seismic network only.
Considering the location of the LES1 station on the focal sphere, we can assume that
additional information from the GNSS station improves and stabilizes the MT solu-
tion. In this case, including the LES1 station amplitude and polarity in the MT inver-
sion improves the focal coverage and adds significant high-amplitude data from a
close distance, which reduces the RMS error in comparison to solutions without this
station and provides a more realistic magnitude estimation compared with the cata-
logue value. The full, deviatoric, and DC MT solutions of this earthquake for the two
variants are presented in Figures 7(B) and 8. The calculated local magnitudes are
overestimated, which might be misleading. However, this kind of magnitude overesti-
mation can happen during MT calculations, especially for shallow, weak events for
which the velocity model is complex. Therefore, the magnitudes calculated with dif-
ferent sets of stations and velocity models might differ, e.g. the magnitude registered
for this event by EMSC was mb ¼ 4.9, while in the local seismic catalogue presented
in the EPISODES Platform (2017), the magnitude is Mw ¼ 4.0.
As shown by Lizurek (2017), better focal coverage from the sensors is essential to
the MT inversion quality. Although the number of seismometers is significantly larger
than the number of HR-GNSS stations, the HR-GNSS data might be crucial when
there is a lack of seismic data; this occurs for the presented mining tremors, especially
for the 2019 event and the TRBC/TRZB station, which stabilizes the solution.
Jacknife tests and the solution presented in Figure 5(B3) show that the TRBC/TRZB
station is crucial to the MT inversion solution. A lack of data from this station makes
the solution incorrect. After applying the HR-GNSS data from this station in place of
seismometric data, the solution is in accordance with the pure seismometric MT
inversion (the values of components in the second and last rows in Figure 6 and the
very low value of the rotation angle in the same figure). This shows that it is possible
to apply HR-GNSS data to improve the MT solution when there is a lack of seismic
data for stations that are significant for the calculations.

3.2. P-wave first motion amplitude approximation


The maximum P-wave amplitudes are possible for the MT inversion in the studied
area were calculated for synthetic events with an assumed mechanism, depth, and
magnitude with the focimt software. We generated the synthetic data for magnitudes
ranging from 2.5 to 4.5 and depths of 600, 800 m, 1, 1.2, 1.4, 1.6, 1.8, 2, and 2.2 km.
GEOMATICS, NATURAL HAZARDS AND RISK

Figure 9. Approximated maximum P-wave amplitude for different magnitudes and event depths according to the applied velocity model: (A) real velocity
model for the LGCD area (real rock medium), (B) velocity model for an isotropic medium, and (C) velocity model for a gradually-increasing-velocity medium.
15
16 I. KUDŁACIK ET AL.

The mechanism applied for synthetic event generation was a reverse fault (strike/di-
p/rake: 165/50/90 ). The range of selected depths covers the localization of tremors in
the LGCD. As expected, the highest amplitudes are obtained for shallow events with
high magnitudes (Figure 9). In an isotropic or gradually-increasing-velocity medium,
the magnitude of vibrations on the Earth’s surface decreases with increasing source
depth, as shown in Figures 9(B,C). The seismic wave propagation in the high-velocity
layer (for depths of 800–1000 m) of the rock medium in the LGCD differs from the
propagation in the isotropic or gradually increasing-velocity medium. The elongation
of the seismic wave path caused by the layer below the high-velocity layer causes
stronger damping and lower amplitudes to be registered on the surface. Figure 9(A)
shows that in the case of the LGCD area, for the source located at depths of around
800–1000 m, the P-wave amplitude on the Earth’s surface will be smaller than it
would be for a source depth of ca. 1200 m. The influence of the geological model on
the LGCD area was described in more detail by Tymi nska and Lizurek (2022). The
HR-GNSS noise level was determined at 1.5 mm, so only the shallowest tremors’ first
P-wave onset could be determined at the station directly above the hypocentre.

4. Conclusions
Polish mining regions are some of the most seismically active mining regions in the
world. The mining industry in this area causes ground subsidence and seismic events.
Therefore, the observation of the ground motion with a combination of different
techniques is essential. The accurate determination of the seismological parameters of
mining-related seismic events is particularly important not only for scientific research
but also to prevent people and buildings from being harmed, as well as for future
insurance proceedings. It is also an opportunity to test novel sensors that could be
applied to seismic monitoring and also to other applications in which vibrations exist,
such as structural health monitoring. Therefore, the densification and diversification
of existing ground motion sensors are suggested.
In this paper, with a network of HR-GNSS stations located close to strong-motion
stations, we found that the HR-GNSS-derived position changes, together with seismic
data, can be included in the seismic MT calculation for low-magnitude anthropogenic
earthquakes.
Many factors, such as good focal coverage and the signal-to-noise ratio, which are
crucial for reliable solutions, can influence the MT calculations. In the first event,
which occurred in 2019, the TRBC station is essential for a stable solution, so high-
rate GNSS data fully compensated for the insufficient seismic data. For the second
event, which occurred in 2020, the focal coverage was unfavourable and additional
HR-GNSS data improved the MT solutions. The stations were mainly located south
and west of the epicentre; this is visible in Figure 4 and the focal spheres in Figure 5.
Considering both events, we conclude that high-rate GNSS position changes can be
successfully combined with seismic data to obtain reliable MT calculations, especially
when there is a lack of seismic observations. Therefore, it is possible to use HR-GNSS
data to calculate the MT solution as a supplement to displacement or spectral ampli-
tude data in first P-wave amplitude inversions for anthropogenic events with
GEOMATICS, NATURAL HAZARDS AND RISK 17

magnitudes of about 3.5. We also demonstrated that for such seismic mechanisms
(with a large vertical component), the HR-GNSS data described the short-term verti-
cal deformations well. In terms of the prospects of this method, we suppose that the
low-cost GNSS technology could be used to densify the HR-GNSS and SM networks,
complementing the existing monitoring networks.

Funding
This research was conducted within the European Plate Observing System (EPOS), which is co-
financed by the European Union from the funds of the European Regional Development Fund
(POIR.04.02.00-14-A0003/16 and POIR.04.02.00-00-C005/19-00). This research was partially
financed by Polish National Science Centre grant No 2021/41/B/ST10/02618 and National
Statutory Activity of the Ministry of Education and Science of Poland grant No 3841/E-
41/S/2023. Seismological data for this study were provided by the Institute of Geophysics, Polish
Academy of Sciences, and are available at https://episodesplatform.eu/?lang=en#episode:LGCD.
The GNSS data from the TRZB and TARN stations were provided courtesy of KGHM Cuprum
Sp. Z. o. o. and the University of Warmia and Mazury in Olsztyn, which are co-financed by the
National Centre for Research and Development (POIR.04.01.04-00-0056/17). This work was sup-
ported by the Wroclaw Centre of Networking and Supercomputing (http://www.wcss.wroc.pl)
computational grant using MATLAB software license No 101979.

ORCID
Iwona Kudłacik http://orcid.org/0000-0003-2548-7575
Anna Tymi nska http://orcid.org/0000-0001-7802-2566
Grzegorz Lizurek http://orcid.org/0000-0002-4801-4888
Jan Kapłon http://orcid.org/0000-0002-0068-0865
Jacek Paziewski http://orcid.org/0000-0002-6033-2547

Data availability statement


The seismological data are available from the repository of the European Plate Observing
System (EPOS) Thematic Core Service Anthropogenic Hazards (EPISODES platform) at
https://episodesplatform.eu/?lang=en#episode:LGCD (last accessed in April 2023). The non-fil-
tered HR-GNSS coordinate time series used in this study are available at
https://doi.org/10.5281/zenodo.7503137. The HR-GNSS raw data will be shared on reasonable
request with the corresponding author. The GNSS processing was performed using RTKlib
(http://www.rtklib.com/). The figures in this paper were generated using QGIS and MATLAB.

References
Antonielli B, Sciortino A, Scancella S, Bozzano F, Mazzanti P. 2021. Tracking deformation
processes at the Legnica Glogow Copper District (Poland) by satellite InSAR—I: room and
pillar mine district. Land. 10(6):653. doi:10.3390/land10060653.
Bartlett S, Burgess H, Damjanovic B, Gowans R, Lattanzi C. 2013. Technical report on the pro-
duction of copper and silver by KGHM Polish Copper JSC in the Legnica-Głogow. Lubin:
KGHM Polska Miedz SA.
Benedetti E, Branzanti M, Biagi L, Colosimo G, Mazzoni A, Crespi M. 2014. Global navigation
satellite systems seismology for the 2012 Mw 6.1 Emilia earthquake: exploiting the VADASE
algorithm. Seismol Res Lett. 85(3):649–656. doi:10.1785/0220130094.
18 I. KUDŁACIK ET AL.

Bischoff M, Cete A, Fritschen R, Meier T. 2010. Coal mining induced seismicity in the Ruhr
area, Germany. Pure Appl Geophys. 167(1–2):63–75. doi:10.1007/s00024-009-0001-8.
Blachowski J, Caco n S, Milczarek W. 2009. Analysis of post-mining ground deformations
caused by underground coal extraction in complicated geological conditions. Acta Geodyn
Geomater. 6(3):351–357.
Bock Y, Melgar D, Crowell BW. 2011. Real-time strong-motion broadband displacements from
collocated GPS and accelerometers. Bull Seismol Soc Am. 101(6):2904–2925. doi:10.1785/
0120110007.
Cesca S, Rohr A, Dahm T. 2013. Discrimination of induced seismicity by full moment tensor
inversion and decomposition. J Seismol. 17(1):147–163. doi:10.1007/s10950-012-9305-8.
Dach R, Schaer S, Arnold D, Orliac E, Prange L, Susnik A, Villiger A, J€aggi A. 2020. CODE
final product series for the IGS; [accessed 2022 Mar 10]. http://www.aiub.unibe.ch/down-
load/CODE. doi:10.7892/boris.75876.4.
EPISODES Platform. 2017. Episode: LGCD; [accessed 2022 Nov 8]. https://episodesplatform.
eu/?lang=en#episode:LGCD. doi:10.25171/InstGeoph_PAS_ISEPOS-2017-006.
Furst SL, Doucet S, Vernant P, Champollion C, Carme JL. 2021. Monitoring surface deform-
ation of deep salt mining in Vauvert (France), combining InSAR and leveling data for
multi-source inversion. Solid Earth. 12(1):15–34. doi:10.5194/se-12-15-2021.
Genrich JF, Bock Y. 2006. Instantaneous geodetic positioning with 10–50 Hz GPS measure-
ment: noise characteristics and implications for monitoring networks. J Geophys Res.
111(B03403). doi:10.1029/2005JB003617.
Gibowicz SJ. 1990. Seismicity induced by mining. In: Dmowska R, Saltzman B, editors.
Advances in geophysics. Vol. 32. Elsevier; p. 1–74. doi:10.1016/S0065-2687(08)60426-4.
Hussain A, Ahmed A, Magsi H, Soomro JB, Bukhari SSH, Ro JS. 2021. Adaptive data length
method for GPS signal acquisition in weak to strong fading conditions. Electronics. 10(14):
1735. doi:10.3390/electronics10141735.
Hohensinn R, Geiger A. 2018. Stand-alone GNSS sensors as velocity seismometers: real-time
monitoring and earthquake detection. Sensors. 18(11):3712. doi:10.3390/s18113712.
Iakovidis DK, Ooi M, Kuang YC, Demidenko S, Shestakov A, Sinitsin V, Henry M,
Sciacchitano A, Discetti S, Donati S, et al. 2022. Roadmap on signal processing for next gen-
eration measurement systems. Meas Sci Technol. 33(1):012002. doi:10.1088/1361-6501/
ac2dbd.
Ilieva M, Polanin P, Borkowski A, Gruchlik P, Smolak K, Kowalski A, Rohm W. 2019. Mining
deformation life cycle in the light of InSAR and deformation models. Rem Sens. 11(7):745.
doi:10.3390/rs11070745.
Ilieva M, Rudzi nski Ł, Pawłuszek-Filipiak K, Lizurek G, Kudłacik I, Tondas D, Olszewska D.
2020. Combined study of a significant mine collapse based on seismological and geodetic
data-29 January 2019, Rudna Mine, Poland. Rem Sens. 12(10):1570. doi:10.3390/rs12101570.
Johnson SW, Chambers DJA, Boltz MS, Koper KD. 2021. Application of a convolutional
neural network for seismic phase picking of mining-induced seismicity. Geophys J Int.
224(1):230–240. doi:10.1093/gji/ggaa449.
Jost ML, Herrmann RB. 1989. A student’s guide to and review of moment tensors. Seismol
Res Lett. 60(2):37–57. doi:10.1785/gssrl.60.2.37.
K€aufl P, Valentine AP, O’Toole TB, Trampert J. 2014. A framework for fast probabilistic cen-
troid-moment-tensor determination-inversion of regional static displacement measurements.
Geophys J Int. 196(3):1676–1693. doi:10.1093/gji/ggt473.
Kijko A. 1977. An algorithm for the optimum distribution of a regional seismic network-I.
PAGEOPH. 115(4):999–1009. doi:10.1007/BF00881222.
Kudłacik I, Kapłon J, Lizurek G, Crespi M, Kurpi nski G. 2021. High-rate GPS positioning for
tracing anthropogenic seismic activity: the 29 January 2019 mining tremor in Legnica-
Głogow Copper District, Poland. Measurement. 168(September 2020):1–9. doi:10.1016/j.
measurement.2020.108396.
GEOMATICS, NATURAL HAZARDS AND RISK 19

Kwiatek G, Martınez-Garzon P, Bohnhoff M. 2016. HybridMT: a MATLAB/shell environment


package for seismic moment tensor inversion and refinement. Seismol Res Lett. 87(4):964–
976. doi:10.1785/0220150251.
Li X, Guo B, Lu C, Ge M, Wickert J, Schuh H. 2014. Real-time GNSS seismology using a sin-
gle receiver. Geophys J Int. 198(1):72–89. doi:10.1093/gji/ggu113.
Li X, Zheng K, Li X, Liu G, Ge M, Wickert J, Schuh H. 2019. Real-time capturing of seismic
waveforms using high-rate BDS, GPS and GLONASS observations: the 2017 Mw 6.5
Jiuzhaigou earthquake in China. GPS Solut. 23(1):1–12. doi:10.1007/s10291-018-0808-9.
Lizurek G. 2017. Full moment tensor inversion as a practical tool in case of discrimination of
tectonic and anthropogenic seismicity in Poland. Pure Appl Geophys. 174(1):197–212. doi:
10.1007/s00024-016-1378-9.
Lizurek G, Rudzi nski Ł, Plesiewicz B. 2015. Mining induced seismic event on an inactive fault.
Acta Geophys. 63(1):176–200. doi:10.2478/s11600-014-0249-y.
Melgar D, Bock Y, Crowell BW. 2012. Real-time centroid moment tensor determination for
large earthquakes from local and regional displacement records. Geophys J Int. 188(2):703–
718. doi:10.1111/j.1365-246X.2011.05297.x.
Michel C, Kelevitz K, Houlie N, Edwards B, Psimoulis P, Su Z, Clinton J, Giardini D. 2017.
The potential of high-rate GPS for strong ground motion assessment. Bull Seismol Soc Am.
107(4):1849–1859. doi:10.1785/0120160296.
Orlecka-Sikora B, Cesca S, Lasocki S, Lizurek G, Wiejacz P, Rudzi nski Ł. 2014. Seismogenesis
of exceptional ground motion due to a sequence of mining induced tremors from Legnica-
Głogow Copper District in Poland. Geophys J Int. 198(1):40–54. doi:10.1093/gji/ggu109.
O’Toole TB, Valentine AP, Woodhouse JH. 2013. Earthquake source parameters from GPS-
measured static displacements with potential for real-time application. Geophys Res Lett.
40(1):60–65. doi:10.1029/2012GL054209.
Pawluszek-Filipiak K, Borkowski A. 2021. Monitoring mining-induced subsidence by integrat-
ing differential radar interferometry and persistent scatterer techniques. Eur J Rem Sens.
54(sup1):18–30. doi:10.1080/22797254.2020.1759455.
Paziewski J, Kurpinski G, Wielgosz P, Stolecki L, Sieradzki R, Seta M, Oszczak S, Castillo M,
Martin-Porqueras F. 2020. Towards Galileo þ GPS seismology: validation of high-rate
GNSS-based system for seismic events characterisation. Measurement. 166:108236. doi:10.
1016/j.measurement.2020.108236.
Paziewski J, Sieradzki R, Baryla R. 2018. Multi-GNSS high-rate RTK, PPP and novel direct
phase observation processing method: application to precise dynamic displacement detec-
tion. Meas Sci Technol. 29(3):035002. doi:10.1088/1361-6501/aa9ec2.
Paziewski J, Sieradzki R, Baryla R. 2019. Detection of structural vibration with high-rate pre-
cise point positioning: case study results based on 100 Hz multi-GNSS observables shake-
table simulation. Sensors. 19(22):4832. doi:10.3390/s19224832.
Psimoulis PA, Houlie N, Habboub M, Michel C, Rothacher M. 2018. Detection of ground
motions using high-rate GPS time-series. Geophys J Int. 214(2):1237–1251. doi:10.1093/gji/
ggy198.
Rudzi nski Ł, Cesca S, Lizurek G. 2016. Complex rupture process of the 19 March 2013, Rudna
mine (Poland) induced seismic event and collapse in the light of local and regional moment
tensor inversion. Seismol Res Lett. 87(2A):274–284. doi:10.1785/0220150150.
Ruhl CJ, Melgar D, Geng J, Goldberg DE, Crowell BW, Allen RM, Bock Y, Barrientos S,
Riquelme S, Baez JC, et al. 2019. A global database of strong-motion displacement GNSS
recordings and an example application to PGD scaling. Seismol Res Lett. 90(1):271–279.
doi:10.1785/0220180177.
Saunders JK, Goldberg DE, Haase JS, Bock Y, Offield DG, Melgar D, Restrepo J, Fleischman
RB, Nema A, Geng J, et al. 2016. Seismogeodesy using GPS and low-cost MEMS accelerom-
eters: perspectives for earthquake early warning and rapid response. Bull Seismol Soc Am.
106(6):2469–2489. doi:10.1785/0120160062.
20 I. KUDŁACIK ET AL.

Sen AT, Cesca S, Bischoff M, Meier T, Dahm T. 2013. Automated full moment tensor inver-
sion of coal mining-induced seismicity. Geophys J Int. 195(2):1267–1281. doi:10.1093/gji/
ggt300.
Shearer PM. 2009. Introduction to seismology. Cambridge: Cambridge University Press.
Shu Y, Fang R, Geng J, Zhao Q, Liu J. 2018. Broadband velocities and displacements from
integrated GPS and accelerometer data for high-rate seismogeodesy. Geophys Res Lett.
45(17):8939–8948. doi:10.1029/2018GL079425.
Sılenỳ J, Milev A. 2017. Mechanism of mining-associated seismic events recorded at
Driefontein – Sibanye gold mine in South Africa. In: Rock mechanics and engineering. Vol.
5. London: CRC Press; p. 353–378. doi:10.4324/9781315708119.
Smalley R. 2009. High-rate GPS: how high do we need to go? Seismol Res Lett. 80(6):1054–
1061. doi:10.1785/gssrl.80.6.1054.
Swanson P, Koontz W, Abshire J. 2009. Seismic network operations at a deep underground
coal mining district in Western Colorado (USA). Controlling Seismic Hazard and
Sustainable Development of Deep Mines: 7th International Symposium on Rockburst and
Seismicity in Mines (Rasim7); Vol. 1 and 2; p. 1407–1412. https://www.cdc.gov/niosh/min-
ing/works/coversheet1017.html.
Takasu T. 2009. RTKLIB: open source program package for RTK-GPS. Free and Open Source
Software for Geospatial (FOSS4G) Conference 2009; Tokyo, Japan.
The Canadian Geodetic Survey of Natural Resources Canada. 2021. The Canadian spatial refer-
ence system precise point positioning (CSRS-PPP); [accessed 2022 Mar 10]. https://webapp.
geod.nrcan.gc.ca/geod/tools-outils/ppp.php.
Tymi nska A, Lizurek G. 2022. Noise influence on moment tensor inversion with the use of
first P-wave amplitude based on VERIS and LUMINEOS networks. Proceedings of the
Third European Conference on Earthquake Engineering and Seismology; Bucureşt,
Romania. Conspress. ISBN 978-973-100-533-1. https://3ecees.ro/wp-content/uploads/2022/
08/Proceedings_3ECEES_2022.pdf.
Whidden KM, Pankow KL. 2012. A catalog of regional moment tensors in Utah from 1998 to
2011. Seismol Res Lett. 83(5):775–783. doi:10.1785/0220120046.
Wilson MP, Davies RJ, Foulger GR, Julian BR, Styles P, Gluyas JG, Almond S. 2015.
Anthropogenic earthquakes in the UK: a national baseline prior to shale exploitation. Mar
Pet Geol. 68:1–17. doi:10.1016/j.marpetgeo.2015.08.023.
Yang H, Chu R, Sheng M. 2018. Source characterization of the 2015 gypsum mining earth-
quake in Pingyi, Shandong. Prog Geophys. 33(1):125–132.
Yang Z, Li Z, Zhu J, Wang Y, Wu L. 2020. Use of SAR/InSAR in mining deformation moni-
toring, parameter inversion, and forward predictions: a review. IEEE Geosci Remote Sens
Mag. 8(1):71–90. doi:10.1109/MGRS.2019.2954824.

You might also like