Bulletin of the Seismological Society of AmericaBulletin of the Seismological Society of AmericaAftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake,Aftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake,South Korea: implication of fault heterogeneity and post-seismic relaxationSouth Korea: implication of fault heterogeneity and post-seismic relaxation
--Manuscript Draft--
Manuscript Number:Manuscript Number: BSSA-D-20-00059R1
Article Type:Article Type: Article
Section/Category:Section/Category: Observations, Mechanisms and Hazards of Induced Seismicity
Full Title:Full Title: Aftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake, SouthKorea: implication of fault heterogeneity and post-seismic relaxation
Corresponding Author:Corresponding Author: Junkee Rhie, Ph.D.Seoul National UniversitySeoul, KOREA, REPUBLIC OF
Corresponding Author 's Institution:Corresponding Author 's Institution: Seoul National University
Corresponding Author E-Mail:Corresponding Author E-Mail: [email protected]
Order of Authors:Order of Authors: Jeong-Ung Woo, Ph. D
Minook Kim
Junkee Rhie, Ph.D.
Tae-Seob Kang, Ph.D.
Abstract:Abstract: The sequence of foreshocks, mainshock, and aftershocks associated with a fault ruptureare the result of interactions of complex fault systems, the tectonic stress field, and fluidmovement. Analysis of shock sequences can aid our understanding of the spatialdistribution and magnitude of these factors, as well as providing a seismic hazardassessment. The 2017, M W 5.5 Pohang earthquake sequence occurred following fluid-induced seismic activity at a nearby enhanced geothermal system site and is an exampleof reactivation of a critically stressed fault system in the Pohang Basin, South Korea. Wecreated an earthquake catalog based on unsupervised data-mining and measuring theenergy ratio between short- and long-window seismograms recorded by a temporaryseismic network. The spatial distribution of approximately 4,000 relocated aftershocksrevealed four fault segments striking southwestward. We also determined that the threelargest earthquakes ( M L > 4) were located at the boundary of two fault segments. Weinfer that locally concentrated stress at the junctions of the faults caused such largeearthquakes and that their ruptures on multiple segments can explain the high proportionof non-double couple components. The area affected by aftershocks expands to thesouthwest and northeast by 0.5 and 1 km decade -1 , respectively, which may result frompost-seismic deformation or sequentially transferred static Coulomb stress. The b -valuesof the Gutenberg-Richter relationship temporarily increased for the first three days of theaftershock sequence, suggesting that the stress field was perturbed. The b -values weregenerally low (< 1) and locally variable throughout the aftershock area, which may be dueto the complex fault structures and material properties. Furthermore, the mapped p -values of the Omori law vary along strike, which may indicate anisotropic expansionspeeds in the aftershock region.
Author Comments:Author Comments:
Suggested Reviewers:Suggested Reviewers: Kwang-Hee KimPusan National [email protected] is the first author of the paper "Assessing whether the 2017 Mw 5.4 Pohangearthquake in South Korea was an induced event" published at Science.
Francesco [email protected] is the first author of the paper "The November 2017 Mw 5.5 Pohang earthquake: Apossible case of induced seismicity in South Korea" published at Science.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Grzegorz [email protected] is the first author of the paper "Controlling fluid-induced seismicity during a 6.1-km-deep geothermal stimulation in Finland" published at Science Advances.
Chang-Soo ChoKorea Institute for Geosciences and Mineral [email protected] is one of the authors of "Surface Deformations and Rupture Processes Associatedwith the 2017 Mw 5.4 Pohang, Korea, Earthquake" published at BSSA and he did lotsworks on relocated seismicity.
Opposed Reviewers:Opposed Reviewers:
Response to Reviewers:Response to Reviewers: I included the responses of reveiwers' comments in an attached file. We really appreciatethe editor and two reviewers for commenting manuscript in detail.
Additional Information:Additional Information:
QuestionQuestion ResponseResponse
<b>Key Point #1: </b><br><i>Key Pointsare now mandatory for BSSA, and willappear at the front of articles starting in2020. Please submit three COMPLETEsentences addressing the following: 1) whatproblem did you address?; 2) whatconclusions did you come to?; and 3) whatare the implications of your findings? Eachpoint must be 110 characters or less(including spaces).
Three largest M > 4 earthquakes of the 2017 MW 5.5 Pohang sequence was located atjunctions of fault segments.
Key Point #2:Key Point #2: Along-strike expansion of aftershock area was observed, implying afterslips or Coulombstress transfer.
Key Point #3:Key Point #3: Generally low b-values (<1) and variations in p-values along northeast direction weremapped.
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
17 April 2020
Dear Editor:
We would like to thank you, one anonymous reviewer and Sebastian Hainzl, for careful reviews on our
manuscript “Aftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake, South Korea:
implication of fault heterogeneity and post-seismic relaxation”. The comments were thorough and
considerate so that they led us to elaborate the article precisely. Changes of the manuscript are
highlighted in blue. We now hope that you find the paper acceptable for publication.
Sincerely,
Junkee Rhie
School of Earth and Environmental Sciences, Seoul National University, Seoul.
Seoul, 08826, Republic of Korea
+82-2-880-6731
+82-2-871-3269
==============================
REVIEWERS' COMMENTS: [Note that reviewers sometimes upload files as part of their reviews. If
attachments have been uploaded, they can be found either as links at the bottom of this email or when
you log into the online submission system, select "Submissions Needing Revisions," and then select the
action "View Reviewer Attachments."]
Reviewer #1: Report on thy manuscript titled
"Aftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake, South Korea: implication of
fault heterogeneity and post-seismic relaxation"
by J.-U. Woo, M. Kim, J. Rhie and T.-S. Kang
The authors analyzed in detail the aftershock sequence of the M5.5 Pohang mainshock based on a newly
created catalog. For that purpose, they applied a sophisticated method for detection, localization, and
magnitude determination. The resulting catalog is then analyzed with respect to the spatial distribution,
spatial and temporal changes of the frequency-magnitude distribution, spatial variations of the Omori-
decay, and migration patterns. The analysis identifies multiple fault segments and spatial variations of the
aftershock properties.
Overall, I find that the paper is well written, the applied techniques and analysis appropriate and
comprehensive, and the results of interest. I have some comments/suggestions listed below.
Major points:
(1) Short-term incompleteness after the mainshock:
It is suspicious that in Figure 4b, blue (late times), green (early times) and black (all times) distributions
seem to have almost the same slope (b-value) for M>1.8, which might indicate that the observed b-value
differences result from incompleteness problems.
It is well known that earthquake catalogs artificially lack small aftershocks directly after the mainshock
with approximately a linear decrease of Mc with log(t) (see e.g. Hainzl SRL 2016). The authors take this
incompleteness already partially into account by estimating time-dependent Mc-values. However, the Mc-
estimations in time-bins might still underestimate the true Mc-value which could be an explanation for the
Response to Reviews
observed b-value increase with time after the mainshock. To check this possibility, the authors should
present a plot of the earthquake catalog as function of the logarithmic of time after the mainshock, where
they add their estimated step-type Mc-estimation as lines. Alternatively, they might fit a linear log(t)-
dependence of Mc and use dm_i = m_i - Mc(t_i) values to estimate the b-values.
RESPONSE: We agree with your suggestion that the magnitude of completeness (Mc) would be
proportional to log(t). From our analysis, transient Mc decreased from 0.8 at the first temporal bin, and
then decreased to 0.2 at t ~ 4 days. Following your suggestion, we inserted Figure 4a which describes
time-varying Mc.
As you addressed, the estimated b-values can be biased by underestimated Mc-values. To test our
observation of time-varying b-value, we have attached Figure R1 illustrating the magnitude distributions
and GR law fitting lines for four cases (the first temporal bin of green distribution, the second temporal bin,
the sixth temporal bin of red distribution, and the last temporal bin of blue distribution) of which three
cases (the first, the sixth, and the last bin) are already represented in Figure 4c as three examples. For
the first bin, the b-values with > 90% goodness of fit were determined to be in a range of 0.7 to 0.8,
depending on the choice of Mc value. This could have resulted from the time-varying Mc value in the
selected bin. However, the range of b-values with > 90% goodness of fit was still less than those for the
sixth bin.
We also compared the next bin (i.e., the second bin) with the sixth bin, and both of them revealed
stable b-values for the Mc interval of [0.5, 1.0] with > 90% goodness of fit. We observed that the b-values
of the bins are clearly different from each other. The Utsu test between the bins suggests that the
probability of the b-value difference between these bins is statistically not significant, and was estimated
as 0.2%. Therefore, the b-values of the second and the sixth temporal bins, with a more stable Mc
estimation than the first bin, were also different from each other.
We also stated that the transient b-values were also observed for the highest Mc value of 0.8, which
are illustrated as gray dots and error bars in Figure 4b.
For the first temporal bin with green distribution in Figure 4c, we used a green triangle symbol along
with the black circle symbol in order to more clearly illustrate the relatively low b-values of the green
distribution within the interval of [3, 5.5] (see Figure R1a below).
Figure R1. Distribution of earthquake magnitudes and their corresponding Gutenberg-Richter law fitting
lines (left) and the estimation of b-values and Mc (right) for the first, second, sixth, and last temporal bin
illustrated in Figure 4. Green circles on the left panels represent the selected Mc and cumulative number
of earthquakes with M(> Mc−1/2ΔM). Red lines on the right panels represent the 90% thresholds of the
goodness of fit.
(2) The authors fitted an Omori-decay relative to the largest M4.6 aftershock, ignoring the ongoing activity
triggered by the mainshock. They justify this choice by stating that the M4.6 "resets the decay rate of the
mainshock" (line 357). This is not state of the art. The epidemic type aftershock sequence (ETAS) model
is known to work well to describe observed activity as sum of all ongoing aftershock sequences. Thus the
fit of the second sequence should include the ongoing Omori decay of the mainshock (with parameters
K1, c1, p2 fitted in the first period), i.e. the following function should be used
R(t) = K1 / (c1 + t_M4.6 - t_main + t)^p1 + K2 / (c2 + t)^p2
where t is the time after the M4.6 aftershock and K2, c2, p2 are the fit parameters.
RESPONSE: We agree with your suggestion. Taking into account your suggestion, we recalculated the
three Omori parameters (k, c, and p) of the Omori law for the aftershocks of the M 4.6 earthquake (period
B) with the decaying aftershock rates for period A with the estimation of Mc and b-values and > 90%
goodness of fit (lines 383–384). We also rescaled K1, considering that the difference of Mc between
period A and period B:
R(t) = K110b1(𝑀𝑐1−𝑀𝑐2) × (𝑐1 + Δt + t)−p1 + 𝐾2 × (𝑐2 + 𝑡)−𝑝2
where Δt is the onset time difference between the ML 4.6 and ML 5.5 events and the symbols “1” and “2”
are used for periods A and B, respectively. This is because the observed number of earthquakes N(≥
Mc−1/2ΔM) should be rescaled with respect to the Mc values for the period B. However, the updated
formula made no significant change of p-values during period B (Figure 5).
(3) Migration pattern:
I completely understand the motivation to test whether a log(t)-migration can be observed which might be
indicative for afterslip. I have also found such a log(t)-migration for an aftershock sequence triggered by
fluid flow (Hainzl et al., JGR 2016). However, the indicated migration envelope in Fig.6c is not really
convincing. On the bottom side, no obvious migration is observable at all (with a limit around 9 km). On
the top, there is already activation very early on at around 2.3 km which remains active the whole period.
There seems to be an extension with time up to values around 0.5 km, but this extension does not
necessarily look like a log(t)-migration.
As alternative, the authors should also discuss a r ~ sqrt(t) increase which would be indicative for pore
pressure diffusion, because of the known presence of fluids in the triggering process.
RESPONSE: We overlapped the earthquake density under the earthquake distribution in Figure 6c to
visualize the migration trend by illustrating the moment that the tenth earthquake occurred in each 0.25
km spatial bin with a 0.1-km sliding window, following the expression of Wu et al. (2017). To the northeast,
very early seismicity clearly existed at around 2.3 km but our relocated earthquake catalog illustrates that
a more northeastern segment at around 0–1 km was not activated until t ~ 0.1 d. To the southwest, an
aftershock area at around 9–10 km was sequentially expanded with logarithmic time and triggered the
seismicity for ML 4.6 and its aftershocks at t ~ 100 d. Therefore, we suggest that the aftershock area has
generally expanded with time. With the bilateral expansion of aftershock area to the northeast and
southwest, the exceptions such as very early aftershocks at around 2.3 km might be triggered by other
mechanisms such as dynamic or static stress transfer due to the mainshock and the following
aftershocks (lines 431–434).
Following your suggestion, we also tested the density plot with r ~ sqrt(t) with t = 0 at the onset of
mainshock (see Figure R2 below). According to a simple diffusion theory with homogeneous hydraulic
diffusivity, the spatial distribution of earthquakes is expected to expand with the square root of time from
the injection point (in our case, the injection points were the open-hole sections of the PX-1 and PX-2
wells). However, for our study the expansion of the area seems to be irrelevant in the case of the injection
point being at around 5.2 km.
For our case study, we suspect that the observed aftershock expansion may have resulted from the
effect of aseismic afterslip as well as a static/dynamic stress triggering mechanism. We have added and
revised several passages related to this issue (lines 378–418).
Figure R2. Spatiotemporal distribution of seismicity along the E1-E2 of Figure 6a. The abscissa is set to
the squared root of time.
(4) Data availability:
The authors write in lines 533-535 that "The earthquake catalog used in this study will be released at
zenodo website (with doi number) when it is published in journal". What is meant by "when it is published
in journal"? Does it mean that it is not published with this BSSA-article together and only after an
additional publication? This would be not desirable.
RESPONSE: This article and the earthquake catalog described in this article were addressed only in this
original BSSA manuscript. We intended to distribute the earthquake catalog when this article is published.
We modified the data statement in the data/resources section (lines 555–556).
Minor points:
- lines 96-98: Please reformulate the sentence "Using the spatial distribution … parameters b and p",
which is difficult to understand.
RESPONSE: We revised the phrase as the Gutenberg-Richter b-value and the Omori law p-value (lines
98–99).
- line 153: The number of 1357 earthquakes is not comparable to the 174 events detected by Kim et al.
(2018b). Thus, "This is comparable to" should be better replaced by e.g. "These events can be compared
to the 174 earthquakes detected by Kim et al. (2018a), who utilized … : The FAST algorithm …"
RESPONSE: We compared the earthquake catalog of FAST with that of Kim et al. (2018) for their
overlapping period (i.e., from the beginning of 14th November 2017 to 15th November 2017 08:40 UTC).
For the period, the FAST catalog contains 169 earthquakes, while Kim et al. (2018) investigated 217
earthquakes, not 174 events. We have revised the related passages and inserted the expression “their
overlapping period” to avoid confusion (lines 154–157).
- lines 289-291: I do not understand how the length and width of the fault are really calculated from the
following sentence: "The fault length and width were determined as the difference between the 2.5th and
97.5th percentile of the strike and dip components". Please rewrite it.
RESPONSE: We have added a sentence to describe the definition of the strike and dip components. It
will help in understanding how we defined the length and width of faults in this study (lines 291–295).
- line 312/313 and line 809 (Fig. 4 caption): Contradictory Mc-values: 0.8 and 1.0 … which one is correct?
RESPONSE: The value 0.8 is correct. We fixed the description of Figure 4a.
- line 320: "at least 250 earthquakes" … with magnitude above Mc?
RESPONSE: For each spatial bin, we calculated its b-value only if at least 250 earthquakes with
magnitude measurements existed within 1.5 km from the center of each bin. We measured the Mc value
for each bin and obtained b-values based on the Aki’s maximum likelihood method with M ≥ Mc−1/2ΔM
(ΔM represents the magnitude bin (= 0.1)). Among the spatial bins with estimated b-values, the minimum
and maximum number of earthquakes with M ≥ Mc−1/2ΔM were 165 and 480, respectively. We have
revised the passage to better explain the estimation of the Mc value before analyzing the b-values (line
323).
- line 329 and 331: Please add error values to the b-values, e.g. "0.69 +-0.15" instead of "0.69"
RESPONSE: We added the 1σ error following your suggestion (lines 312, 315, 334, and 336).
- line 345-347: It is difficult to follow what is really done here: "We also applied Utsu's test for all pairs of
spatially varying b-values for which the difference is statistically significant with a significance level of 5%
if Δb > 0.1 for half the cases and Δb > 0.135 for all cases (Figure 4d)". Please describe it more clearly.
RESPONSE: We revisited the related description and reformatted the passage into multiple sentences
for a clearer expression of the concept (lines 350–352).
- line 379: "propagating afterslip" instead of "propagating aftershock"
RESPONSE: We have revised this (line 384).
- line 396-399: I understand the first part of the sentence, namely "For regions with low p-value, the slip
velocity decreases relatively slowly", but I do not understand why in this case "the accumulated post-
seismic displacement required to rupture asperities can takes short time compared to that of the regions
with high p-value"?
RESPONSE: As time goes on, the “slowly decreasing slip velocity” can generate deformation which is
larger than that for the case of “rapidly decreasing slip velocity”. The deformation (or slip displacement)
as the definite integral of slip velocity with respect to time on the specific time period [ta, tb] will also be
larger in the case of “slowly decreasing slip velocity”. Therefore, a certain amount of displacement
(required to rupture asperities) can be reached faster than that of the opposite case. We have revised the
related passage to better explain this (lines 401–404).
- line 403: "aftershocks" instead of "aftershock rates"
RESPONSE: We have revised this (line 409).
- line 406-415:
On the one hand it is stated that "no clear evidence of post-seismic deformation was observed from
differential InSAR analysis (Song and Lee, 2019)". On the other hand it is mentioned that "the
descending image of differential InSAR reveals surface deformation during the first day after the
mainshock" and "the ascending image reveals deformation during the next 19 days". This seems to be
contradictory.
RESPONSE: We have revised the related portions of lines 412–415. In Song and Lee (2019), the InSAR
analysis was applied for the post-seismic period of 16-28 November 2017, from one day after the
occurrence of the ML 5.4 mainshock, which occurred on 15 November 2017. Therefore, we suspect that a
possible aseismic deformation during the day of the ML 5.4 mainshock, when most of aftershock
expansion occurred, may not have been resolved by the InSAR analysis.
- Tabe 2: There are empty rows for "Median" and "deviation". Furthermore, please clarify how the "fault
thickness" is defined.
RESPONSE: The “Median absolute deviation” represents one cell, not three cells. We have revised Table
2 by adding a rotated layout. The fault thickness was estimated as the difference between the 2.5th and
97.5th percentiles of the last principal components of events after the PCA analysis, but we decided not to
further address the fault thickness in Table 2 considering that it is not as important as the fault strike and
width.
- Figures are often too small and should be increased to ensure visibility.
RESPONSE: We have reviewed all of the figures and increased their size where appropriate.
- Fig. 4a: The colored points are difficult to see.
RESPONSE: We have increased the thickness of the lines and increased the size of the subfigures.
- Fig: A1, caption: Incomplete sentence: " For each subfigure, earthquakes until Red and gray dots"
RESPONSE: We have revised this.
Sebastian Hainzl
Reviewer #2: 1. line 55 in page 4. : "90 people, and made 1500 homeless" -> "92 people, and made 1797
homeless" according to white paper report officially written by Ministry of the Interior and Safety of Korea.
RESPONSE: We have changed the values (from 90 people and 1500 homeless to 135 people and 1797
homeless) following the annual report 2017 of Korean Meteorological Administration (KMA) and replaced
the citation of the related sentence (line 55). Since the estimated values of injuries and homeless are
actually slightly different from several news sources, we have retained these fixed values of the annual
report 2017 of KMA (See the details on the annual report 2017 of KMA:
https://www.kma.go.kr/download_01/Annual_Report_2017.pdf, last accessed on 2nd, April, 2020).
2. line 102 in page 6 : "Ávila-Barrientos et al., 2008" -> " Ávila-Barrientos et al., 2015"
RESPONSE: We have revised this.
3. line 149 in page 8 : "performance trials and or " -> "performance trials and"
RESPONSE: We have revised this.
4. page 31 : It is necessary to sort references alphabetically.
RESPONSE: We have revised this.
5. page 16 in line 326 : "(Chang et al., 2020)"-> There is none in reference.
RESPONSE: Chang et al. (2020) originally planned to submit but unfortunately, it seems to have taken
more time than expected. Therefore, although Chang et al. (2020) studied a more detailed analysis on
the stress state on the Pohang EGS site, we have changed the related citations to the Ellsworth et al.
(2019) and Lee et al. (2019), which also addressed the stress states of the study area.
6. line 375 in page 18 : "Fukumaya and Ellsworth, 2000"->There is none in reference.
RESPONSE: We have revised this.
7. line 377, 380 in page 18 : " Perfettini et al., 2017"->" Perfettini et al., 2018"
RESPONSE: We have revised these two sentences, which both had the same issue.
8. line 441, page 21 : "Choi et al., 2018"-> " Choi et al., 2019"
RESPONSE: We have revised this.
9. line 622 in page 29 : "Langenbruch, C., Ellsworth, W. L., Woo, J. U., and Wald, D. J. (2019)." ->
"Langenbruch, C., Ellsworth, W. L., Woo, J. U., and Wald, D. J. (2020)."
RESPONSE: We have revised this.
10 line 696 in page 32 : "Trnkoczy, A. (1999). " -> This paper is not alluded in your article.
RESPONSE: We have removed the citation.
11. line 274 in page 13 : focal mechanism (strike: 34°, dip: 52°, rake: 136°) is too different from strike
and F4 table 2. And dip in Fig2(e) is about 50 degree different from that of F4 in table 2. It is
necessary to check it.
RESPONSE: We felt the need to visualize all of the focal mechanisms stated throughout the paper.
Therefore, we have inserted seven beachball diagrams into Figure 2 and Table A3 to help illustrate the
focal mechanism solutions. In Figure 2e, the aftershock distribution is matched with the focal mechanism
solution for the ML 4.6 event.
The dip of F4 in Table 2 has been estimated from the three-dimensional aftershock distribution
during one day from the onset of the ML 4.6 event. We visually identified some events in this period which
occurred along the other fault segment, which could have resulted in biased estimation of strike and dip
for F4. We therefore recalculated the fault geometry for F4 from the earthquakes of which onset times
was within one day from the onset of the ML 4.6 event and for which the location was within 2 km from the
ML 4.6 event in order to limit the outlier points on other faults (lines 286–287). By doing this, we obtained
a more gentle dip of F4. We also changed the projection direction of Figure 2e to coincide with the strike
of the F4.
12 page 47 : Earthquake occurrence rate(quakes/day) should be decreased with day according to omori's
law R(t) in Fig5(c),(d). Is it correct?
RESPONSE: Since Figures 5c and 5d represent the number of cumulated events, we changed the
ordinate title of Fig. 5c and 5d to “Cumulative number of aftershocks”.
13. Authors should have all grant to publish the paper using data from all source(KMA. KHNP, KIGAM
and EGS project)
RESPONSE: We have stated these details in the acknowledgements section.
14. In final comment, Authors described activity of aftershocks of Pohang earthquake well. It seems that
F2 fault is similar to F3. Dip and strike is changed slightly along strike of fault. It seems that F2 and F3
have just different spatial strees status and physical properties. Why do authors insist that F2 and F3 is
different fault? And how about calculating not temporal variation of b value with cumulative events of
increment day but temporal variation of b value a day in Figure 4(a)?
RESPONSE: We have visually identified a curvilinear fault system from the aftershock distribution due to
the slightly different geometry of F2 and F3. The difference of p-values between the two fault segments
represents a property of fault heterogeneity in the fault system and does not indicate that that the two
fault segments are in separate fault systems. We thus used the term “fault segment” instead of the “fault,”
which might have misled to indicating a separate fault system.
Illustrated in Figure 4b, the temporal variation of b-values were calculated with a sliding window of
200 earthquakes. Following your suggestion, we also tested daily b-values via these same procedures
(Figure R3). However, given that the aftershock rates decreased with the Omori law, we were able to
obtain b-values for a few days from the mainshock and the day of the ML 4.6 earthquake. However, the
temporal changes of the daily b-values showed a similar trend as those in Figure 4b (see Figure R3
below; please note that the log-based abscissa was used for the comparison of Figure 4b).
Figure R3. Temporal variations of b-values and Mc obtained with a bin size of 1 day. The standard
deviations of each bin are represented with horizontal and vertical error bars.
1
Aftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake, South Korea: 1
implication of fault heterogeneity and post-seismic relaxation 2
3
Jeong-Ung Woo 1, Minook Kim2a, Junkee Rhie 1*, Tae-Seob Kang 3 4
Corresponding author: Junkee Rhie ([email protected]) 5
School of Earth and Environmental Sciences, Seoul National University, 1 Gwanak-ro, Gwanak-6
gu, Seoul 08826, Republic of Korea 7
Phone: +82-2-880-6731 8
Fax: +82-2-871-3269 9
1 School of Earth and Environmental Sciences, Seoul National University, Seoul 08826, Republic 10
of Korea 11
2 Department of Structural Systems and Site Evaluation, Korea Institute of Nuclear Safety, 12
Daejeon 34412, South Korea 13
3 Division of Earth Environmental System Science, Pukyong National University, Busan 48513, 14
Republic of Korea 15
16
Abstract 17
18
The sequence of foreshocks, mainshock, and aftershocks associated with a fault rupture are the 19
result of interactions of complex fault systems, the tectonic stress field, and fluid movement. 20
Analysis of shock sequences can aid our understanding of the spatial distribution and magnitude 21
a also at Division of Earth Environmental System Science, Pukyong National University, Busan
48513.
Annotated Manuscript Click here to access/download;Annotated Manuscript;manu_with_color.docx
2
of these factors, as well as providing a seismic hazard assessment. The 2017, MW 5.5 Pohang 22
earthquake sequence occurred following fluid-induced seismic activity at a nearby enhanced 23
geothermal system site and is an example of reactivation of a critically stressed fault system in 24
the Pohang Basin, South Korea. We created an earthquake catalog based on unsupervised data-25
mining and measuring the energy ratio between short- and long-window seismograms recorded 26
by a temporary seismic network. The spatial distribution of approximately 4,000 relocated 27
aftershocks revealed four fault segments striking southwestward. We also determined that the 28
three largest earthquakes (ML > 4) were located at the boundary of two fault segments. We infer 29
that locally concentrated stress at the junctions of the faults caused such large earthquakes and 30
that their ruptures on multiple segments can explain the high proportion of non-double couple 31
components. The area affected by aftershocks expands to the southwest and northeast by 0.5 and 32
1 km decade-1, respectively, which may result from post-seismic deformation or sequentially 33
transferred static Coulomb stress. The b-values of the Gutenberg-Richter relationship 34
temporarily increased for the first three days of the aftershock sequence, suggesting that the 35
stress field was perturbed. The b-values were generally low (< 1) and locally variable throughout 36
the aftershock area, which may be due to the complex fault structures and material properties. 37
Furthermore, the mapped p-values of the Omori law vary along strike, which may indicate 38
anisotropic expansion speeds in the aftershock region. 39
40
3
INTRODUCTION 41
On 15 November 2017, a moderate-sized earthquake of moment magnitude (MW) 5.5 or local 42
magnitude (ML) 5.4 struck the city of Pohang, located in the southeastern part of the Korean 43
Peninsula, which damaged infrastructure, injured 135 people, and made 1,797 homeless (Korea 44
Meteorological Administration, 2017). The earthquake (hereafter referred to as the mainshock) 45
was the second-largest earthquake event among earthquakes recorded instrumentally in South 46
Korea since 1978, according to the catalog of the Korea Meteorological Administration (KMA). 47
A close examination of the seismic source characteristics of such a rarely observed moderate-48
sized earthquake and its foreshock-mainshock-aftershock sequence is necessary not only to 49
evaluate the current stress field (Zoback, 1992; Soh et al., 2018) and fault properties, but also to 50
understand aftershock triggering mechanisms (King et al., 1994; Kilb et al., 2000). Estimation of 51
statistical parameters (i.e., the Gutenberg-Richter b-value and the Omori law p-value) from a 52
large number of microearthquakes in conjunction with the seismic source properties of 53
aftershocks can give information on fault heterogeneities, such as crack density, slip distribution, 54
applied shear stress, viscoelastic properties, and heat flow (Wiemer and Katsumata, 1999; Murru 55
et al., 2007). 56
One important point to note is that the mainshock occurred near an enhanced geothermal 57
system (EGS) site (Grigoli et al., 2018; Kim et al., 2018; Lee et al., 2019). A body of evidence 58
supports the claim that the mainshock was triggered by five fluid-injection experiments as well 59
as an associated loss of heavy drilling muds and released tectonic energy on a critically stressed 60
fault (Ellsworth et al., 2019; Woo et al., 2019a). The periods of stimulation experiments 61
conducted on two hydraulic wells (PX-1 and PX-2) were closely correlated with microseismicity 62
observed near the wells. Induced seismicity mapped in the vicinity of the EGS indicated the 63
4
presence of a previously unmapped fault. Microseismicity triggered on this fault migrated to the 64
location of the mainshock. A breakout was observed in the PX-2 well at intervals corresponding 65
to the assumed fault. The groundwater levels of PX-1 and PX-2 decreased abruptly by 121 m and 66
793 m, respectively, immediately after the mainshock but gradually recovered by 0.078 m d-1 67
and 0.198 m d-1, respectively (Lee, 2019). 68
Previous studies of aftershock distributions in the Pohang Basin determined the presence of 69
complex fault geometries (Hong et al., 2018; Kim et al., 2019). Separately, Grigoli et al. (2018) 70
reported that obtaining a significant non-double-couple (non-DC) component when inverting the 71
moment tensor for a mainshock can be attributed to the complexity of the rupture process in a 72
multi-fault system. The spatial pattern of early aftershocks associated with two 2016 Gyeongju 73
earthquakes (ML 5.1 and ML 5.8), which occurred on two sub-parallel faults approximately 40 74
km away from the Pohang mainshock, is differentiated from the presence of two or three fault 75
segments with varying strikes and dips for the early aftershocks associated with the 2017 Pohang 76
earthquakes (Uchide and Song, 2018; Son et al., 2018; Woo et al., 2019b). 77
In this study, we created an earthquake catalog for the foreshock-mainshock-aftershock 78
sequence from data recorded by local permanent seismic networks, temporary seismometers 79
deployed as part of the aftershock monitoring system, and the temporary Pohang EGS 80
monitoring system. Earthquakes were detected using a machine-learning data mining technique 81
for data obtained during the first ten days and a conventional automatic detection algorithm was 82
employed for the aftershock monitoring system as a whole. Each detected earthquake was 83
located by manual picking and visual inspection and then precisely relocated by the double-84
difference method (Waldhauser and Ellsworth, 2000). Using the spatial distribution of over 85
4,000 earthquakes, we modeled fault systems as a series of multiple fault segments by mapping 86
5
the spatio-temporal distribution of the Gutenberg-Richter b-value and the Omori law p-value 87
statistical parameters. 88
Mapping the distribution of earthquake magnitudes provides an independent analysis of the 89
characteristics of aftershock activities and can be used to analyze spatial heterogeneities of 90
material properties, such as stress state, level of asperities, and heat flow rate (Scholz, 1968; 91
Wiemer and Katsumata, 1999; Wiemer and Wyss 1997; Ávila-Barrientos et al., 2015); assess 92
seismic hazards via epidemic-type aftershock sequence modeling (ETAS; Ogata, 1998); and 93
conduct probabilistic seismic hazard analysis (PSHA; Cornell, 1968). In this study, we evaluated 94
the relative magnitude of each earthquake by using amplitude ratios relative to earthquakes of 95
known ML. 96
97
DATA AND METHOD 98
Seismic Networks 99
Continuous seismic waveform data used to detect and analyze seismic source parameters were 100
collected from four different networks (Figure 1). The first data set was obtained from a 101
combined permanent seismic network operated by KMA, the Korea Institute of Geoscience and 102
Mineral Resources (KIGAM) and the Korea Hydro and Nuclear Power (KHNP). The permanent 103
seismic networks of KMA, KIGAM, and KHNP are named KS, KG, and KN, respectively. The 104
second set of continuous waveform data were recorded by nine borehole seismometers installed 105
at depths of between 100 and 150 m, which operated to monitor microseismic events for the 106
Pohang EGS project. Three of the temporary borehole seismometers recorded the mainshock, 107
while the operation of the other borehole seismometers started within the next 2 days; all of them 108
operated until the end of November 2017. The third continuous waveform data set was collected 109
6
by 37 temporary broad-band seismometers installed after the mainshock by the university 110
consortium (Pukyong National University and Seoul National University) and KIGAM 111
independently. The first seismometer installed temporarily for monitoring aftershocks started its 112
operation approximately 1 h after the onset of the mainshock. Lastly, we used waveforms of 214 113
early aftershocks, occurred within four hours from the mainshock, recorded at eight short-period 114
temporary seismometers deployed by Pusan National University (Kim et al., 2018). The 115
seismograph stations of these temporary networks were densely spaced and located within the 116
radius of 20 km from the EGS site, respectively (Figure 1). 117
118
Detection and Hypocenter Determination 119
Since stabilizing temporary seismometers for aftershock monitoring can take many hours, 120
conventional algorithms for earthquake detection, such as STA/LTA (Withers et al., 1998; 121
Trnkoczy 2002), are of limited use for locating early aftershocks because of the incompleteness 122
of the local seismometer network. In this study, we utilized the Fingerprint and Similarity 123
Threshold (FAST) data-mining algorithm that uses waveform similarity to detect such early 124
aftershock sequences (Yoon et al., 2015; Yoon et al., 2017; Bergen et al., 2018) with a 125
conventional energy-based algorithm for the period for aftershock monitoring system. The FAST 126
algorithm finds pairs of waveforms having similar spectrograms without any prior information, 127
allowing us to obtain pairs of earthquake candidates with correlative signals. The performance of 128
the FAST algorithm to discriminate true events from earthquake candidates can be improved by 129
measuring similarity at multiple stations (Bergen et al., 2018). 130
We applied the FAST method to ten days of continuous seismograms recorded between 14 Nov 131
2017 and 23 Nov 2017 to cover the period of operation of the aftershock monitoring system. We 132
7
used three-component seismograms obtained from two short-period (PHA2 and DKJ) and one 133
broadband (CHS) seismometers, which are located within 30 km from the mainshock. The three 134
borehole seismometers of the EGS monitoring system that were operational at the onset of the 135
mainshock were not used in detection due to the high level of ambient background noise and 136
regularly observed pulse-like signals. The sampling rate of the seismograms was fixed at 100 Hz 137
and the frequency range of the bandpass filter was set to 2 – 20 Hz. All parameters employed in 138
the FAST algorithm routines were either determined manually from performance trials and were 139
previously applied values (Yoon et al., 2017; Yoon et al., 2019a) and are summarized in Table 140
A1 and A2. 141
We detected 1,580 candidate events via the FAST search, leading to a subset of 1,357 locatable 142
earthquakes from visual inspection. Compared with the earthquake catalog published by Kim et 143
al. (2018), which utilized eight local seismographs deployed within 3 km of the EGS site for 144
earthquake detection, the FAST algorithm successfully detected 169 out of 217 or 78% of 145
earthquakes for their overlapping period. 146
While the aftershock monitoring network was operational (i.e., from 15 November 2017 to 28 147
February 2018), we applied an automatic algorithm to detect and locate microseismic 148
earthquakes (Sabbione and Velis, 2013). Continuous waveforms were transformed into 149
characteristic functions for measuring the ratio between the short-term average (STA) and the 150
long-term average (LTA). We declared candidate earthquakes when the STA/LTA ratio 151
exceeded 5 for a given time window of 4 s at more than three stations. For each triggered time 152
window, the normalized squared envelope functions of Baer and Kradolfer (1987) were 153
calculated to determine the time at which to maximize the function value (hereafter referred to as 154
the BK function). Since the BK function can be maximized for the arrivals of either the P-wave 155
8
or the S-wave, the maximum value of the BK function was tested to discriminate whether the 156
measured local maximum corresponded to the first arrival. If we observed a local high BK 157
function value before the maximum of the BK function in a given time window, we set two 158
consecutive time samples as the arrivals of the P- and S-waves. Otherwise, we searched for other 159
local maximum after the triggered time window and set the maxima as the P- and S-wave phase 160
arrivals when a secondary maximum was available. The phase arrivals determined in this way 161
were visually confirmed by using a Wadati plot (Wadati, 1933). 162
We determined the initial hypocenters of the detected earthquakes via Hypoellipse (Lahr, 1999), 163
with phase arrival times being determined by manual inspection and a 1-D layered seismic 164
velocity model for the Pohang EGS site (Woo et al., 2019a; Table 1). In this procedure, we 165
combined the earthquakes detected from either the FAST algorithm or the STA/LTA method 166
with events with ML > 2.0 listed in the KMA and Kim et al. (2018) event catalogs. Earthquakes 167
with an onset difference of less than 2 s were regarded as duplicate events. Station corrections 168
were calculated based on a comparison of the theoretical arrival times for five immediate 169
foreshocks reported by Woo et al. (2019a) and their picked arrival times. 170
Initial hypocenters were relocated with hypoDD (Waldhauser and Ellsworth, 2000) by using 171
travel time differences obtained from waveform cross-correlation measurements as well as 172
picked phase times as inputs to the double-difference algorithm. The 1-D velocity model of Woo 173
et al. (2019a) was applied for the relocation procedure, again (Table 1). All relocated events were 174
shifted by 39 m, 28 m, and 96 m in eastwards, northwards, and downwards, respectively, to 175
match the centroid of the five immediate foreshocks with the results of Woo et al. (2019a), of 176
which recordings at 17 PX-2 borehole chains were applied to obtain accurate hypocenters. We 177
resampled waveforms to 1,000 Hz with a cubic spline after first having applied a 2–10 Hz 178
9
bandpass filter. Each seismogram was reduced to a 1 s time window centered at each phase 179
arrival time. We allow a time shift up to 0.1 s for the cross-correlation measurements. Time 180
shifts that maximized the cross-correlation coefficient (CC) between two pairs of waveforms 181
were used only if the maximum CCs were greater than 0.85. The squared maximum CCs were 182
used to weight the measurements. The relative locations were calculated by least-squares fitting 183
of the data and the location uncertainties were evaluated by using bootstrapping analysis 184
(Waldhauser and Ellsworth, 2000). Synthetic travel time differences between paired events were 185
reconstructed by random selection of a set of residuals and relative locations for these synthetic 186
travel times were calculated 200 times. 187
188
Magnitude Estimation and Statistical Analysis 189
Waveform similarity can be assessed to estimate the relative magnitudes of earthquakes (Shelly 190
et al., 2016; Yoon et al., 2019b). We adopted a simple magnitude-amplitude relationship 191
modified from the equation of Shelly et al. (2016) that considers the differences in hypocentral 192
distance between two earthquakes: 193
dm = clog10(a/r), (1) 194
where dm, a, and r represent the ratios of magnitude, amplitude, and hypocenteral distance, and c 195
is a coefficient for the magnitude-amplitude relationship (Shelly et al. 2016). The coefficient c in 196
Equation 1 varies with the earthquake magnitude scale that is used: for example, Shelly et al. 197
(2016) reported that c = 1 for ML and c = 2/3 for MW. In this study, we used a set of MLs of 198
aftershocks and Equation (1) to estimate the coefficient c, following the method of Woo et al. 199
(2019b). If the CC of a waveform pair was greater than 0.85, we calculated the amplitude ratio as 200
the slope of the eigenvector for the largest eigenvalue of the covariance matrix of the two 201
10
waveforms (Shelly et al., 2016). Thus, for earthquakes with known values of ML, we were able to 202
estimate the parameter c. 203
We can also determine relative magnitudes of earthquakes by using our estimated value of c in 204
Equation (1). Estimated relative magnitudes (MRel) were arithmetically averaged to produce a 205
representative value and uncertainties were obtained from their standard deviations. 206
The Gutenberg-Richter law (G-R law) describes the relationship between earthquake frequency 207
and magnitude. Its statistical properties are widely accepted and applied to the investigation of 208
seismo-tectonic properties in a specific region over a certain time period. Examples of 209
application of the G-R law include work on aftershock sequences by Wiemer and Katsumata 210
(1999) and Woo et al. (2019b), on earthquake swarms by Farrell et al. (2009), on induced 211
seismicity by Shapiro (2007), and in laboratory experiments by Scholz (1968). The earthquake 212
frequency distribution with magnitude can be written as: 213
log10 N(≥ M) = a - bM, (2) 214
where N is the number of earthquakes equal to or greater than a magnitude M, and a and b are 215
scaling constants. a is proportional to the overall seismicity in a given spatio-temporal interval, 216
whereas b represents the ratio of the number of large earthquakes to small earthquakes. The 217
behavior of b-values has been attributed to crack density (Mogi, 1962), stress drop (Wyss, 1973), 218
and tectonic stresses (Schorlemmer et al., 2005, Scholz, 1968), and slip distribution (Wiemer and 219
Katsumata, 1999). 220
We determined the magnitude of completeness (MC) for 3,521 magnitudes based on a modified 221
goodness-of-fit method of Wiemer and Wyss (2000), following Woo et al. (2019b). Then, we 222
evaluated the b-value for a set of magnitudes using the maximum likelihood method of Aki 223
11
(1965) with a magnitude bin of 0.1. The uncertainty of b-values was estimated with the method 224
of Shi and Bolt (1982). 225
Omori’s law describes the decay rate of aftershocks. Its parameters are also broadly applied to 226
interpret regional seismic and tectonic properties (Omori, 1894; Utsu, 1961). The extended form 227
of Omori’s law can be written as: 228
R(t) = K(t+c)-p, (3) 229
where K, c, and p are the scaling coefficients that describe the aftershock decay rates in a given 230
region. p, which represents the power of the aftershock decay rates, has a range of 0.6 to 1.8 and 231
is considered to be a function of stress and temperature in the crust (Utsu and Ogata, 1995; 232
Wiemer and Katsumura 1999). We mapped the spatial variation of p-values by binning 250 233
magnitudes and by selecting magnitudes greater than MC. The three parameters and their 234
associated uncertainties were determined following the maximum likelihood method presented 235
by Ogata (1983). 236
237
RESULTS 238
Of the 4,446 earthquakes with initial locations, we relocated seven foreshocks, the mainshock, 239
and 3,938 aftershocks using hypoDD (Waldhauser and Ellsworth, 2000), having excluded 240
earthquakes with fewer than seven traveltime difference measurements. Uncertainties of relative 241
locations to within two standard deviations were estimated as 25 m in the eastwest direction, 18 242
m in the northsouth direction, and 37 m vertically. Figure 2 presents the spatial distribution of 243
aftershocks, both in map view and cross-sections, four in the dip direction (A1-A2, B1-B2, C1-C2, 244
and D1-D2) and one in the strike direction (E1-E2). From the map, we determined the apparent 245
strike of aftershocks (crossline of E1-E2) to be 210°, which corresponds to the azimuth of the first 246
12
principal vector obtained from two-dimensional principal component analysis (PCA) (Jollifle, 247
2002). From cross-sections in the dip direction (A1-A2 to D1-D2), we observed that the spatial 248
distribution of aftershocks delineates at least four different fault segments (Figure 2). In the most 249
northeastern part of the study area, a sub-vertical fault was identified from the aftershock 250
distribution. An ML 3.5 earthquake on 21:05:15, 19 November 2017; UTC with a focal 251
mechanism (strike: 234°, dip: 85°, rake: -174°) is consistent with the inferred fault (Figure 2b, 252
Table A3). Among the relocated earthquakes, the first observed event on the fault plane occurred 253
within 72 s of the onset of the mainshock (Figure A1), which indicates that reactivation of the 254
fault segment was initiated by the mainshock rupture or soon afterward. Two slightly different 255
fault geometries, both dipping northwestward, are distinguished in the middle of sections B1-B2 256
and C1-C2 from the spatial distribution of the aftershocks. The aftershock distribution along B1-257
B2 has a wider range of focal depths, a shallower dip, and a strike closer to north-south than that 258
of C1-C2. Both the mainshock and the ML 4.3 aftershock are located adjacent to a virtual 259
boundary of B1-B2 and C1-C2 and their focal mechanisms are consistent with the observed fault 260
geometry. Earthquakes in the southwestern part of D1-D2 occurred after the largest aftershock 261
(ML 4.6) (Figure A1) and their focal depths deepened to the south-east, dipping in the opposite 262
direction to the three other fault segments observed on A1-A2, B1-B2 and C1-C2. Such a conjugate 263
fault geometry is matched with one nodal plane of the focal mechanism (strike: 34°, dip: 52°, 264
rake: 136°) of the largest ML 4.6 aftershock (Figure 2e, Table A3). 265
From the complex fault geometry delineated by the four cross-sections, we constructed a 266
simplified fault model to describe the observed aftershock distribution. For the three segments 267
that reactivated with the occurrence of the mainshock, we described their geometry using the 268
aftershocks that occurred within one day of the mainshock. Because the mainshock was situated 269
13
on a virtual boundary between two faults (F2 and F3) with slightly different strikes and dips, we 270
divided the aftershock area based on the hypocenter of the mainshock and an apparent strike of 271
210°, which we estimated from PCA of data in map view. The aftershocks on the most 272
northeasterly fault segment (F1) were de-clustered from the adjoining fault (F2) using the simple 273
assumption that the Heunghae Fault (i.e., Song et al., 2015; Yun et al., 1991) vertically intersects 274
them both. Earthquakes that occurred up to 1 day after the largest ML 4.6 aftershock and are 275
located within 2 km from the event were used to investigate the most southwesterly fault 276
segment (F4). Faults F1F4 were used to divide the study area into four regions and earthquakes 277
were assigned to a region on the basis of the location of their hypocenter. We applied PCA 278
analysis with bootstrapping to earthquakes that were resampled 200 times to estimate strike, dip, 279
fault length, and fault width. We rotated the first and second principal components to the two 280
unit direction vectors for strike and dip; thus defining the strike and dip components of 281
earthquakes as these projections. The fault length and width were then determined as the 282
difference between the 2.5th and 97.5th percentiles of the strike and dip components. The 283
resulting fault geometry is summarized in Table 2. 284
We determined c using the 266 relocated earthquakes with known ML. We evaluated c as 0.85 285
by PCA (Figure 3), which is larger than the case for the MW magnitude scale (c = 2/3) scale but 286
smaller than the case for the ML magnitude scale (c = 1). The difference in c implies that the ML 287
magnitude does not naturally match MW for earthquakes within the range of magnitudes included 288
in this study, filtered to a frequency range of 2 – 10 Hz. The observed value of c is relatively 289
high compared with 0.7 that was estimation using the MLs of the Gyeongju aftershock sequences 290
(Woo et al., 2019b), which may be the result of systematic differences between ML and KMA 291
magnitude. 292
14
We estimated the magnitudes of 3,521 earthquakes with measurements ≥ 5. Figures 3b 293
illustrates the comparison of MRel with ML and the variations of MRel with time. Since MRel is 294
exactly proportional to ML without any scaling parameters, we propose that MRel can replace ML 295
as the magnitude scale for subsequent analysis. 296
We examined temporal variations of seismic b-values by binning 600 earthquake magnitudes 297
(MRel) into a set (Figure 4a). There was an overlap of four hundred earthquakes between two 298
consecutive bins. The MC decreased from 0.8 to 0.2 during the first 3 days of the early aftershock 299
sequence, which is indicative of a decrease in the background noise level for that period (Hainzl 300
2016). The b-value for the first bin was evaluated as 0.66 ± 0.03, which is consistent with b-301
values for earthquakes detected during fluid injection into the Pohang EGS site before the 302
occurrence of the mainshock (Woo et al., 2019a). The b-value increased with time for the first 303
three days up to a maximum of 0.98 ± 0.05 and fluctuated during a month. After three months, it 304
decreased to 0.77 ± 0.04 when the largest aftershock of ML 4.6 occurred. We tested the temporal 305
changes of b-values with a fixed MC of 0.8, corresponding to the maximum values over the 306
whole period, to investigate whether the observed temporal variations of b-values were biased by 307
the choice of MC (gray dots of Figure 4b) and confirmed that the main features were not 308
significantly changed. Figure 4c illustrates the magnitude-frequency distributions of three data 309
sets highlighted in Figure 4b. 310
The spatial variation of b-values was investigated for the vertical cross-section along the 311
apparent strike of 210°. Earthquakes within 1.0 km from the center of each 0.5 0.5 km grid cell 312
on the cross-section were binned. We analyzed the MC and b-value only if each bin contained at 313
least 250 earthquakes. Figure 4d illustrates the spatial distribution of b-values on the vertical 314
cross-section. The estimated b-values are between 0.63 and 0.91, all of which are lower than the 315
15
typically assumed b-value of 1 (Wyss, 1973). Since ML is approximated by MRel, such low b-316
values can be interpreted as an increase in applied shear stress and effective stress (Scholz, 1968; 317
Wyss 1973), low material heterogeneity (Mogi, 1962), or a high stress drop (Wyss, 1973). 318
Considering that the slip tendency of the mainshock is indicative of a critically stressed fault 319
(Ellsworth et al., 2019; Lee et al., 2019) and the stress drop of 5.6 MPa for the mainshock is not 320
higher than that of other earthquakes in South Korea (Rhee and Sheen 2016; Woo et al., 2019a), 321
our preferred interpretation is that the generally low b-values in the aftershock area may result 322
from high applied stress in this region. We estimated a b-value of 0.73 ± 0.04 near the 323
hypocenter of the mainshock, which is comparable to the values observed for the earthquakes 324
during the fluid injection (= 0.66 ± 0.08). 325
The significance of temporal and spatial differences in b-values can be verified by Utsu’s test 326
(Utsu, 1992), in which the probability that the b-values between two sets of earthquakes are the 327
same is defined via Akaike Information Criterion (Akaike, 1974). We first tested the statistical 328
significance of the temporal differences of b-values among early (< 1 d), intermediate (~ 3 d), 329
and late aftershocks (~ 90 d), which are highlighted in green, red, and blue, respectively, in 330
Figures 4b and 4c. The probability that the b-value for the intermediate period is not significantly 331
higher than those of the early and late aftershocks was estimated as 7.3×10-7 and 1.6×10-3, 332
respectively, indicating that the temporal increase and decrease of b-values are statistically 333
reasonable with a significance level of 5%. Similar variations of b-values with time can be found 334
for the 2016 Gyeongju earthquake (Woo et al., 2019b) and other cases (Smith, 1981; Chan et al., 335
2012; Gulia et al., 2018), which can be interpreted as local stress changes due to the mainshock 336
rupture or a mixed effect of a changing spatial distribution of b-value and a heterogeneous 337
population of aftershocks with time (Figure 4d). 338
16
We also applied Utsu’s test to all pairs of spatially varying b-values and measured the 339
distribution of significant levels with the b-value difference (Δb) bin of 0.01. A significance level 340
of 5% was held for Δb > 0.14 in half of the cases and Δb > 0.18 in all cases (Figure 4e). 341
Therefore, we roughly designated three sub-regions: R1 with relatively low b-values, and R2 and 342
R3 with high b-values (Figure 4d). The ML 4.3 and ML 4.6 earthquakes are located near R2 and 343
R3, which can be interpreted as indicating material heterogeneity with respect to the conjugate 344
fault system (Figures 2c and 2e). Alternatively, spatial variations of pore pressure or applied 345
stress may contribute to b-value heterogeneity. 346
The p-values were estimated for two data sets: (1) period A, between the onset of the 347
mainshock and the ML 4.6 aftershock; and (2) period B, after the onset of the largest aftershock 348
of ML 4.6 (Figure 5). This grouping was chosen because the occurrence of the largest aftershock 349
at ~87 days resulted in increased seismicity, which reset the decay rate for the mainshock. For 350
each data set, we estimated the p-value that represents the whole data set and the spatial variation 351
of p-values at the cross-sections along the apparent strike of 210°, with the same bins used for 352
estimating the spatial variations of b-values. The p-values of the period B were estimated with 353
the consideration of decaying aftershock rates of the period A. The p-value of period A was 354
estimated as 1.10, which is larger than the value for period B (= 0.88). Such a difference may 355
result from differing initial stress levels for periods A and B with respect to the stress 356
perturbation of the mainshock sequence, spatial heterogeneity of the internal structure for the 357
conjugate fault system (Figure 2e; Wiemer and Katsumata, 1999), or just an insufficient number 358
of earthquakes in the calculation of p-values for period B. With the exception of p-values for 359
period B, the p-values of the period A were higher in the southwestern region than those in the 360
northeastern region (Figure 5a). This could be indicative of a spatial variation of heat flow 361
17
(Kisslinger and Jones, 1991), heterogeneity of fault strength (Mikumo and Miyatake, 1979) or an 362
insufficient number of aftershocks to allow accurate fitting of the aftershock power decay law for 363
the southwestern aftershock region prior to the occurrence of the ML 4.6 aftershock. 364
365
DISCUSSION 366
Expansion of aftershock areas with time 367
Expansion of early aftershock sequences is widely observed (Tajima and Kanamori 1985; 368
Fukuyama et al., 2003; Peng and Zhao, 2009; Kato and Obara, 2014; Hainzl et al., 2016). Some 369
temporal evolution of aftershock areas have been interpreted to be the result of afterslip or post-370
seismic deformation (Helmstetter and Shaw 2009; Peng and Zhao 2009; Ross et al., 2017; 371
Perfettini et al., 2018). Speeds of along-strike expansion of the aftershock zone were measured 372
on a logarithmic time scale and showed that propagating afterslip can cause the expansion of 373
aftershocks (Peng and Zhao 2009; Frank et al., 2017; Perfettini et al., 2018; Ross et al., 2017). In 374
the present study, we examined the spatio-temporal distribution of aftershocks on a logarithmic 375
time scale to consider possible post-seismic deformation following the mainshock (Figure 6). In 376
a map view, we observed that the aftershock zone has roughly expanded along the apparent 377
strike direction, especially during the first day (Figures 6a and c), whereas no clear trends were 378
observed in a vertical sense (Figure 6b). 379
The general speed of virtual aftershock migration fronts for the bilateral expansion along the 380
strike direction were ~1 km decade-1 northeastward and ~0.5 km decade-1 southwestward (Figure 381
6c), which may indicate post-seismic deformations related to aseismic afterslip (Peng and Zhao 382
2009; Perfettini et al., 2018). The difference in the migration speeds can be attributed to different 383
rate-and-state parameters described by Dieterich (1994) following the equations published by 384
18
Perfettini et al. (2018). However, in our case, we also observe a significant p-value variation in 385
the northeastern and southwestern parts of the study area (Figure 5a). Such variations of p-value 386
require a different model rather than the rate-dependent friction law (Helmstetter and Shaw 2009; 387
Mignan 2015). Assuming that the power-law rheology governs post-seismic velocity which is 388
proportional to (1+t/t*)-p (Montési, 2004), where t* is a characteristic time of the aftershock, the 389
slip velocity or the aftershock occurrence rate decays with time as a power of p. The slowly 390
decreasing slip velocity with a lower p-value generate a larger accumulated displacement than 391
that with a higher p-value in a given time period, and thus the time required to rupture asperities 392
is shorter than that with a higher p-value. 393
Therefore, the p-value variation observed for the aftershock area during the period A may be 394
related to the different seismic migration speeds (Figures 5a and 6c). We did not further compare 395
p-values and the migration speed in this study, since it may require more complex analysis than a 396
simplified form of the Omori’s law (Narteau et al., 2002). Furthermore, there is an absence of 397
data for very early (« 1 d) or late (> 100 d) aftershocks. 398
The expansion of the aftershock zone can also be explained by a cascade of sequentially 399
triggered aftershocks in terms of changes to the static Coulomb stress (Ellsworth and Bulut, 400
2018). These mechanisms can also explain very early aftershocks deviated from the expansion of 401
aftershock area at around 3 km. Since no clear evidence of post-seismic deformation was 402
observed in the differential InSAR analysis during 12 days of the post-seismic period (Song and 403
Lee, 2019), the observed expansion of aftershocks could possibly be attributed to changes to the 404
static stress field caused by the aftershock sequences rather than a result of aseismic deformation. 405
However, post-seismic deformation during the first day of the aftershock period might not be 406
19
captured in InSAR data because most of the expansion of the aftershock would be limited to 407
observations within 1 d from the mainshock. 408
409
High percentage of non-DC components observed for the mainshock and two largest 410
aftershocks 411
The moment tensor solutions of the mainshock and two largest earthquakes have high 412
percentages (> 30%) of non-DC components (Grigoli et al., 2018; Hong et al., 2018), in contrast 413
to the normally observed moment tensor solutions in South Korea. Such high non-DC 414
components of the moment tensor solutions of the three largest earthquakes can result from 415
complex shear faulting of multiple DCs, tensile opening/closing, and shear faulting in anisotropic 416
and heterogeneous media (Miller et al., 1998). It has already been established that the spatial 417
distribution of the Pohang earthquake sequence indicates that multiple fault segments were 418
reactivated in a complex fault system and the faulting types of the focal mechanism vary 419
throughout the aftershock area (i.e., Kim et al., 2019). Hence, a combination of multiple DC 420
moment tensor solutions with varying senses of slip motion could be one of the causes of the 421
three largest earthquakes having high non-DC components. 422
We propose the following sequence of events to explain the mainshock and major aftershock 423
sequence associated with the MW 5.5 Pohang earthquake. We infer that the nucleation of the 424
mainshock rupture was initiated at the junction between F2 and F3 and that the rupture 425
propagated along F2 and F3 with possible intervention of F1. Later, the ML 4.3 earthquake was 426
initiated between two adjacent conjugate faults dipping southwestward and northeastward in the 427
deeper aftershock region below the mainshock (Figure 2c). Finally, the ML 4.6 earthquake 428
nucleated at the southwestern tip of the aftershock area and subsequent aftershocks occurred on a 429
20
previously unrecorded southeastward dipping fault, suggesting that the rupture of the ML 4.6 430
earthquake sequences was initiated at the intersection of conjugate faults F3 and F4. 431
Although the three earthquakes were located at the intersection of multiple fault planes, it is hard 432
to envisage that all the earthquakes located in the surrounding area ruptured on multiple fault 433
planes. Some MRel 3 3.6 earthquakes without non-DC components were located in the vicinity 434
of the interconnecting faults (Choi et al., 2019), which may suggest that a certain amount of 435
seismic energy is required for the simultaneous movement of multiple fault segments. The fault 436
dimensions for the three largest earthquakes are inferred to be greater than 1 km, based on the 437
assumption of a constant stress drop of 5.6 MPa on a circular crack (i.e., Figure 2f), leading us to 438
propose that a kilometer rupture scale is the threshold to rupture multiple fault planes. Low b-439
values observed throughout the aftershock area can be considered as stress concentrations within 440
areas of high asperities (Wiemer and Wyss, 1997). High asperities in the regions adjoining two 441
or more fault segments may concentrate tectonic energy either as an earthquake nucleation point 442
or as barriers to rupture propagation. This may explain why only ML > 4 non-DC component 443
earthquakes were observed. The sonic log data of the PX-2 borehole recorded the existence of 444
anisotropic structures in the Pohang Basin (Ellsworth et al., 2019). Such anisotropic materials 445
can also cause earthquakes with high non-DC components. However, it is our preferred 446
interpretation that non-DC components in the three largest earthquakes result from the fault 447
complexity because low, non-DC earthquakes for ML 3 – 3.6 earthquakes were also observed. 448
449
Comparison between aftershock activities and induced seismicity at the EGS site during 450
stimulation. 451
21
The seismicity recorded during the five hydraulic stimulation experiments at the Pohang EGS 452
site and the inferred focal mechanisms revealed a fault plane located near the PX-2 well (Woo et 453
al., 2019a). PX-2 seismicity was clustered on a plane with a strike of 214° and a dip of 43° and 454
migrated southwestward, heading toward the location of the mainshock (Woo et al., 2019a). 455
However, the fault geometry for the induced earthquakes related to the PX-2 well has a 20° 456
shallower dip angle than the moment tensor solution of the mainshock and aftershocks. It 457
suggests that complex fault segments exist locally throughout the aftershock region and that a 458
simple fault plane does not explain the detailed fault structures. The ML 4.3 earthquakes have 459
deeper focal depths and their focal mechanism has steeper dips than that of the mainshock, which 460
can also be regarded as a result of complex fault geometry. Observation of various types of focal 461
mechanisms in aftershock sequences (Kim et al., 2019) are also a manifestation of the complex 462
geometry, which is in contrast to the nearly identical focal mechanisms for the PX-2 seismicity 463
(Woo et al., 2019a). 464
The b-values observed during the Pohang EGS project have insignificant variations, with an 465
average value of 0.66 ± 0.08 (Woo et al., 2019a, Langenbruch et al., 2020); whereas, the b-466
values estimated for the early aftershock sequences are statistically different from the b-values 467
for a bin of approximately 3 days after the mainshock (Figure 4b). If we assume that b-values act 468
as a stress-meter (Scholz, 2015; Rigo et al., 2018; Woo et al., 2019b) and temporal variation of 469
b-values during the aftershock period represents the level of stress state, the invariant b-values 470
observed during the stimulation period suggest that stress perturbations caused by fluid injection 471
may be far lower than the accumulated tectonic stress, indicating the existence of a critically 472
stressed fault system before the mainshock. 473
474
22
Reactivation of a multi-segment fault system and spatial variations of b-values and p-values 475
The complexity of the Pohang aftershock distributions was modeled as four fault segments, 476
following the approach of Hong et al. (2018) and Kim et al. (2018, 2019) (Figure 7). The 477
seismicity along a subvertical fault, F1, in the northeastern of the study area clearly represents 478
migration of the aftershock front northeastward during the first day of the aftershock sequences 479
(Figure 6). Although this fault plane is located about 3 km away from the mainshock hypocenter, 480
it may have been reactivated as a part of the mainshock rupture process. Alternatively, it may 481
have been dynamically triggered by the mainshock considering circumstantial evidence that 482
aftershock activity on the fault segments was initiated within just 2 min (Figure A1) and the slip 483
distribution of the mainshock calculated from the static deformations with InSAR data is largest 484
in the northeastern part of the fault model (Song and Lee, 2019). Aftershocks on F1 are bounded 485
by the Heunghae Fault, which has surface expression (Figure 1), detaching F1 from F2 and F3. 486
Therefore, in either case, the reactivation of F1 may require a certain stress threshold to be 487
ruptured preferentially to F2 and F3. 488
Two slightly different geometries of F2 and F3 are suggested by Hong et al. (2018), reflecting a 489
complex fault system near the Pohang EGS site. While the b-values vary slightly on F2, the 490
observed p-values were higher for F3, at least until the occurrence of the ML 4.6 event. The 491
different behaviors of the two statistical parameters imply that the two fault segments exist under 492
different physical conditions, such as: differential stress states (Scholz, 1968), local 493
heterogeneity of the rock matrix that may interact with viscous materials (Wyss, 1973; Bayrak et 494
al., 2013), or variable spatial distribution of heat flow (Kisslinger and Jones, 1991). 495
The b-values decreased to ~0.7 when fault segment F4 was reactivated by the ML 4.6 aftershock. 496
The lower b-values may indicate F4 was already highly stressed when the ML 4.6 earthquake was 497
23
triggered. The observed p-values for period B were generally much lower than those for period A, 498
which may be the result of using short time periods for analysis during period B or just uneven 499
seismicity observed for periods A and B. 500
501
CONCLUSIONS 502
In this study, we detected over 4,000 earthquakes related to the MW 5.5 (ML 5.4) Pohang 503
earthquake by using both unsupervised data-mining and a conventional automatic earthquake 504
detection method. From the spatio-temporal distribution of relocated seismicity, we observed 505
that four fault segments were responsible for the aftershocks. All the faults strike 506
northeastsouthwest, but have different dip angles and dip directions. The three largest 507
earthquakes are located at the boundaries of two adjoining fault segments, which may have 508
focused the stress released by multiple faults, resulting in high, non-DC earthquake mechanisms. 509
By measuring amplitude ratios between two similar earthquakes, we estimated relative 510
magnitudes of earthquakes to infer the statistical parameters related to earthquake frequency and 511
magnitude. The observed spatio-temporal distribution of b-value indicates that they were 512
spatially variable, but generally as low as ~0.7, and temporarily increased with time. The 513
observed p-values were different for the northeastern and southwestern parts of the study area, 514
implying that heterogeneities in material properties such as frictional heat can lead to two 515
different speeds of aftershock expansion rate with logarithmic time. The complexity of faulting 516
in the aftershock zone will influence the duration and magnitude of seismic activity that is 517
caused by the locally perturbed stress field that is a result of the mainshock. We hope that our 518
findings can be applied to an interpretation of aftershock mechanisms under the general complex 519
24
fault systems and can be utilized to perform a seismic hazard assessment lowering the epistemic 520
uncertainty about the characteristics of the fault sources and their contemporary seismic activity. 521
522
Data and Resources 523
524
The earthquake catalog used in this study is available at https://github.com/Jeong-525
Ung/PH_aftershock. 526
527
Acknowledgments 528
529
We thank the Korea Institute of Geoscience and Mineral resources (KIGAM), the Korea 530
Meteorological Administration (KMA), the Korea Hydro & Nuclear Power (KHNP), and the 531
K.‐ H. Kim for providing seismic data used in this study. We appreciate C. E. Yoon and G. C. 532
Beroza for comments on FAST usage, W. L. Ellsworth for advice on visualizing the seismicity, J. 533
Song for discussion on non-DC earthquakes. This work was conducted during the Korean 534
Government Commission (KGC) on the relations between the 2017 Pohang earthquake and EGS 535
project, funded by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) 536
grant from the South Korean government (MOTIE) (no. 2018‐ 3010111860). This work was 537
supported by the Nuclear Safety Research Program through the Korea Foundation of Nuclear 538
Safety (KoFONS) using the financial resource granted by the Nuclear Safety and Security 539
Commission (NSSC) of the Republic of Korea (No. 1705010). 540
541
References 542
25
543
Akaike, H. (1974). A new look at the statistical model identification, IEEE Trans. 544
Automat. Control 19, 716–723. 545
Aki K. (1965). Maximum likelihood estimate of b in the formula logN=a−bM and its 546
confidence limits, Bull. Earthq. Res. Inst. Univ. Tokyo 43, 237–239. 547
Ávila-Barrientos, L., Zúñiga, F. R., Rodríguez-Pérez, Q., and Guzmán-Speziale, M. 548
(2015). Variation of b and p values from aftershocks sequences along the Mexican subduction 549
zone and their relation to plate characteristics. J. S. Am. Earth Sci. 63, 162–171. 550
Baer, M., and Kradolfer, U. (1987). An automatic phase picker for local and teleseismic 551
events. Bull. Seismol. Soc. Am. 77, 1437–1445. 552
Bayrak, Y., Yadav, R. B. S., Kalafat, D., Tsapanos, T. M., Çınar, H., Singh, A. P., Bayrak 553
E., Yılmaz, Ş., Öcal, F, and Koravos, G. (2013). Seismogenesis and earthquake triggering during 554
the Van (Turkey) 2011 seismic sequence. Tectonophysics 601, 163–176. 555
Bergen, K. J., and Beroza, G. C. (2018). Detecting earthquakes over a seismic network 556
using single-station similarity measures. Geophys. J. Int. 213, 1984–1998. 557
Chan, C. H., Wu, Y. M., Tseng, T. L., Lin, T. L., and Chen, C. C. (2012). Spatial and 558
temporal evolution of b-values before large earthquakes in Taiwan. Tectonophysics 532, 215–222. 559
Choi, J. H., Ko, K., Gihm, Y. S., Cho, C. S., Lee, H., Song, S. G., Bang, E.-S., Lee, H.-J., 560
Bae, H.-K., Kim S. W., Choi S.-J., Lee, S. S., and Lee, S. R. (2019). Surface deformations and 561
rupture processes associated with the 2017 M w 5.4 Pohang, Korea, earthquake. Bull. Seismol. 562
Soc. Am. 109, 756–769. 563
Cornell, C. A. (1968). Engineering seismic risk analysis. Bull. Seismol. Soc. Am. 58, 564
1583–1606. 565
26
Dieterich, J. (1994). A constitutive law for rate of earthquake production and its 566
application to earthquake clustering. J. Geophys. Res.: Solid Earth 99, 2601–2618. 567
Ellsworth, W. L., and Bulut, F. (2018). Nucleation of the 1999 Izmit earthquake by a 568
triggered cascade of foreshocks. Nat. Geosci. 11, 531–535. 569
Ellsworth, W. L., Giardini, D., Townend, J., Ge, S., and Shimamoto, T. (2019). 570
Triggering of the Pohang, Korea, Earthquake (MW5.5) by Enhanced Geothermal System 571
Stimulation. Seismol. Res. Lett. 90, 1844–1858. 572
Farrell, J., Husen, S., and Smith, R. B. (2009). Earthquake swarm and b-value 573
characterization of the Yellowstone volcano-tectonic system. J. Volcanol. Geotherm. Res. 188, 574
260–276. 575
Frank, W. B., Poli, P., and Perfettini, H. (2017). Mapping the rheology of the Central 576
Chile subduction zone with aftershocks. Geophys. Res. Lett. 44, 5374–5382. 577
Fukuyama, E., Ellsworth, W. L., Waldhauser, F., and Kubo, A. (2003). Detailed fault 578
structure of the 2000 western Tottori, Japan, earthquake sequence. Bull. Seismol. Soc. Am. 93, 579
1468–1478. 580
Grigoli, F., Cesca, S., Rinaldi, A. P., Manconi, A., Lopez-Comino, J. A., Clinton, J. F., 581
Westaway, R., Cauzzi, C., Dahm, T., Wiemer, S. (2018). The November 2017 Mw 5.5 Pohang 582
earthquake: A possible case of induced seismicity in South Korea. Science 360, 1003–1006. 583
Gulia, L., Rinaldi, A. P., Tormann, T., Vannucci, G., Enescu, B., and Wiemer, S. (2018). 584
The Effect of a Mainshock on the Size Distribution of the Aftershocks. Geophys. Res. Lett. 45, 585
13–277. 586
Hainzl, S. (2016). Rate‐ dependent incompleteness of earthquake catalogs. Seismol. Res. 587
Lett. 87, 337-344. 588
27
Hainzl, S., Fischer, T., Čermáková, H., Bachura, M., and Vlček, J. (2016). Aftershocks 589
triggered by fluid intrusion: Evidence for the aftershock sequence occurred 2014 in West 590
Bohemia/Vogtland. J. Geophys. Res.: Solid Earth 121, 2575-2590. 591
Helmstetter, A., and Shaw, B. E. (2009). Afterslip and aftershocks in the rate‐ and‐ state 592
friction law. J. Geophys. Res.: Solid Earth 114, B01308 593
Hong, T. K., Lee, J., Park, S., and Kim, W. (2018). Time-advanced occurrence of 594
moderate-size earthquakes in a stable intraplate region after a megathrust earthquake and their 595
seismic properties. Sci. Rep. 8, 1–8. 596
Jolliffe I. T. (2002). Principal Component Analysis, Second Ed., Springer, New York. 597
Kato, A., and Obara, K. (2014). Step‐ like migration of early aftershocks following the 598
2007 Mw 6.7 Noto‐ Hanto earthquake, Japan. Geophys. Res. Lett. 41, 3864–3869. 599
Kilb, D., Gomberg, J., and Bodin, P. (2000). Triggering of earthquake aftershocks by 600
dynamic stresses. Nature 408, 570–574. 601
Kim, K. H., Ree, J. H., Kim, Y., Kim, S., Kang, S. Y., and Seo, W. (2018). Assessing 602
whether the 2017 MW 5.4 Pohang earthquake in South Korea was an induced event. Science 360, 603
1007–1009. 604
Kim, K. H., Seo, W., Han, J., Kwon, J., Kang, S. Y., Ree, J. H., Kim, S., and Liu, K. 605
(2019). The 2017 ML 5.4 Pohang earthquake sequence, Korea, recorded by a dense seismic 606
network. Tectonophysics 774, 228306. 607
King, G. C., Stein, R. S., and Lin, J. (1994). Static stress changes and the triggering of 608
earthquakes. Bull. Seismol. Soc. Am. 84, 935–953. 609
Kisslinger, C., and Jones, L. M. (1991). Properties of aftershock sequences in southern 610
California. J. Geophys. Res.: Solid Earth 96, 11947–11958. 611
28
Korea Meteorological Administration (2017). Annual report 2017. Seoul, Republic of 612
Korea, 35 pp. 613
Lahr J. C.1999. HYPOELLIPSE: A computer program for determining local earthquake 614
hypocentral parameters, magnitude and first‐ motion pattern (Y2K compliant version), version 615
1.0, U.S. Geol. Surv. Open‐ File Rept. 99‐ 23, On‐ Line Edition. 616
Langenbruch, C., Ellsworth, W. L., Woo, J. U., and Wald, D. J. (2020). Value at Induced 617
Risk: Injection‐ induced seismic risk from low‐ probability, high‐ impact events. Geophys. Res. 618
Lett. 47, c2019GL085878. 619
Lee, K.-K. (2019). Final Report of the Korean Government Commission on Relations 620
between the 2017 Pohang Earthquake and EGS Project. Geological Society of Korea. 621
https://doi.org/10.22719/KETEP-2019043001. 622
Lee, K.-K., Ellsworth, W. L., Giardini, D., Townend, J., Ge, S., Shimamoto, T., Yeo, I.-623
W., Kang, T.-S., Rhie, J., Sheen, D.-H., Chang, C., Woo, J.-U., Langenbruch, C. (2019). 624
Managing injection-induced seismic risks. Science 364, 730–732. 625
Mignan, A. (2015). Modeling aftershocks as a stretched exponential relaxation. Geophys. 626
Res. Lett. 42, 9726–9732. 627
Mikumo, T., and Miyatake, T. (1979). Earthquake sequences on a frictional fault model 628
with non-uniform strengths and relaxation times. Geophys. J. Int. 59, 497-522. 629
Miller, A. D., Foulger, G. R., and Julian, B. R. (1998). Non‐ double‐ couple earthquakes 630
2. Observations. Rev. Geophys. 36, 551–568. 631
Mogi, K. (1962). Study of the elastic shocks caused by the fracture of heterogeneous 632
materials and its relation to earthquakes phenomena. Bull. Earthq. Res. Inst. 40, 125–173. 633
29
Montési, L. G. (2004). Controls of shear zone rheology and tectonic loading on 634
postseismic creep J. Geophys. Res.: Solid Earth 109, B10404. 635
Murru, M., Console, R., Falcone, G., Montuori, C., and Sgroi, T. (2007). Spatial 636
mapping of the b value at Mount Etna, Italy, using earthquake data recorded from 1999 to 2005. 637
J. Geophys. Res.: Solid Earth 112, B12303 638
Narteau, C., Shebalin, P., and Holschneider, M. (2002). Temporal limits of the power law 639
aftershock decay rate. J. Geophys. Res.: Solid Earth 107, ESE–12. 640
Ogata, Y. (1983). Estimation of the parameters in the modified Omori formula for 641
aftershock frequencies by the maximum likelihood procedure. J. Phys. Earth 31, 115–124. 642
Ogata, Y. (1998). Space-time point-process models for earthquake occurrences. Ann. Inst. 643
Stati. Math. 50, 379–402. 644
Omori F. (1894). On after‐ shocks of earthquakes, J. Coll. Sci. Imp. Univ. Tokyo 7, 111–645
200. 646
Peng, Z., and Zhao, P. (2009). Migration of early aftershocks following the 2004 647
Parkfield earthquake. Nat. Geosci. 2, 877. 648
Perfettini, H., Frank, W. B., Marsan, D., and Bouchon, M. (2018). A model of aftershock 649
migration driven by afterslip. Geophys. Res. Lett. 45, 2283–2293. 650
Rhee, H. M., and Sheen, D. H. (2016). Lateral variation in the source parameters of 651
earthquakes in the Korean Peninsula. Bull. Seismol. Soc. Am. 106, 2266–2274. 652
Rigo, A., Souriau, A., and Sylvander, M. (2018). Spatial variations of b-value and crustal 653
stress in the Pyrenees. J. Seismol. 22, 337–352. 654
30
Rong, K., Yoon, C. E., Bergen, K. J., Elezabi, H., Bailis, P., Levis, P., and Beroza, G. C. 655
(2018). Locality-sensitive hashing for earthquake detection: a case study of scaling data-driven 656
science. Proceedings of the VLDB Endowment, 11, 1674-1687. 657
Ross, Z. E., Rollins, C., Cochran, E. S., Hauksson, E., Avouac, J. P., and Ben‐ Zion, Y. 658
(2017). Aftershocks driven by afterslip and fluid pressure sweeping through a fault‐ fracture 659
mesh. Geophys. Res. Lett. 44, 8260–8267. 660
Sabbione, J. I., and Velis, D. R. (2013). A robust method for microseismic event 661
detection based on automatic phase pickers. J. Appl. Geophys. 99, 42-50. 662
Scholz, C. H. (1968). The frequency-magnitude relation of microfracturing in rock and 663
its relation to earthquakes. Bull. Seismol. Soc. Am. 58, 399–415. 664
Scholz, C. H. (2015). On the stress dependence of the earthquake b value. Geophys. Res. 665
Lett. 42, 1399–1402. 666
Schorlemmer, D., Wiemer, S., and Wyss, M. (2005). Variations in earthquake-size 667
distribution across different stress regimes. Nature 437, 539–542. 668
Shapiro, S. A., Dinske, C., and Kummerow, J. (2007). Probability of a given‐ magnitude 669
earthquake induced by a fluid injection. Geophys. Res. Lett. 34, L22314. 670
Shelly, D. R., Hardebeck, J. L., Ellsworth, W. L., and Hill, D. P. (2016). A new strategy 671
for earthquake focal mechanisms using waveform‐ correlation‐ derived relative polarities and 672
cluster analysis: Application to the 2014 Long Valley Caldera earthquake swarm. J. Geophys. 673
Res.: Solid Earth 121, 8622–8641. 674
Shi, Y., and Bolt, B. A. (1982). The standard error of the magnitude-frequency b value. 675
Bull. Seismol. Soc. Am. 72, 1677–1687. 676
Smith, W. D. (1981). The b-value as an earthquake precursor. Nature 289, 136–139. 677
31
Soh, I., Chang, C., Lee, J., Hong, T. K., and Park, E. S. (2018). Tectonic stress 678
orientations and magnitudes, and friction of faults, deduced from earthquake focal mechanism 679
inversions over the Korean Peninsula. Geophys. J. Int. 213, 1360–1373. 680
Son, M., Cho, C. S., Shin, J. S., Rhee, H. M., and Sheen, D. H. (2018). Spatiotemporal 681
Distribution of Events during the First Three Months of the 2016 Gyeongju, Korea, Earthquake 682
Sequence. Bull. Seismol. Soc. Am. 108, 210–217. 683
Song, C. W., Son M., Sohn, Y. K., Han R., Shinn Y. J., and Kim J.-C. (2015). A study on 684
potential geologic facility sites for carbon dioxide storage in the Miocene Pohang Basin, SE 685
Korea. J. Geol. Soc. Korea 51, 53–66 (in Korean). 686
Song, S. G., and Lee, H. (2019). Static slip model of the 2017 Mw 5.4 Pohang, South 687
Korea, earthquake constrained by the InSAR data. Seismol. Res. Lett. 90, 140–148. 688
Tajima, F., and Kanamori, H. (1985). Global survey of aftershock area expansion 689
patterns. Phys. Earth Planet. Inter. 40, 77–134. 690
Trnkoczy A. (2002). Understanding and parameter setting of STA/LTA trigger algorithm, 691
in IASPEI New Manual of Seismological Observatory Practice (NMSOP), Bormann P. (Editor), 692
Vol. 2, GeoForschungsZentrum, Potsdam, Germany, 1–19. 693
Uchide, T., and Song, S. G. (2018). Fault rupture model of the 2016 Gyeongju, South 694
Korea, earthquake and its implication for the underground fault system. Geophys. Res. Lett. 45, 695
2257–2264. 696
Utsu, T. (1961). A statistical study on the occurrence of aftershocks, Geophys. Mag. 30, 697
521–605. 698
Utsu, T. (1992), On Seismicity, Report of the Joint Research Insititute for Statistical 699
Mathematics, Inst. Stat. Math., Tokyo 34, 139–157. 700
32
Utsu, T., Y. Ogata, and Matsu’ura, R. S. (1995). The centenary of the Omori formula for 701
a decay law of aftershock activity, J. Phys. Earth 43, 1–33. 702
Wadati, K., and Oki, S. (1933). On the travel time of earthquake waves, Part II, Geophys. 703
Mag. 7, 101–111. 704
Waldhauser, F., and Ellsworth, W. L. (2000). A double-difference earthquake location 705
algorithm: Method and application to the northern Hayward fault, California. Bull. Seismol. Soc. 706
Am. 90, 1353–1368. 707
Wiemer, S., and Katsumata, K. (1999). Spatial variability of seismicity parameters in 708
aftershock zones. J. Geophys. Res.: Solid Earth 104, 13135–13151. 709
Wiemer, S., and Wyss, M. (1997). Mapping the frequency‐ magnitude distribution in 710
asperities: An improved technique to calculate recurrence times?. J. Geophys. Res.: Solid 711
Earth 102, 15115–15128. 712
Wiemer, S., and Wyss, M. (2000). Minimum magnitude of completeness in earthquake 713
catalogs: Examples from Alaska, the western United States, and Japan. Bull. Seismol. Soc. 714
Am. 90, 859–869. 715
Withers, M., Aster, R., Young, C., Beiriger, J., Harris, M., Moore, S., and Trujillo, J. 716
(1998). A comparison of select trigger algorithms for automated global seismic phase and event 717
detection. Bull. Seismol. Soc. Am. 88, 95–106. 718
Woo, J. U., Kim, M., Sheen, D. H., Kang, T. S., Rhie, J., Grigoli, F., Ellsworth, W., L., 719
and Giardini, D. (2019a). An in‐ depth seismological analysis revealing a causal link between 720
the 2017 MW 5.5 Pohang earthquake and EGS project. J. Geophys. Res.: Solid Earth. 124, 721
13060–13078. 722
33
Woo, J. U., Rhie, J., Kim, S., Kang, T. S., Kim, K. H., and Kim, Y. (2019b). The 2016 723
Gyeongju earthquake sequence revisited: aftershock interactions within a complex fault system. 724
Geophys. J. Int., 217, 58–74. 725
Wyss, M. (1973). Towards a physical understanding of the earthquake frequency 726
distribution. Geophys. J. R. Astron. Soc., 31, 341–359. 727
Yoon, C. E., Bergen, K. J., Rong, K., Elezabi, H., Ellsworth, W. L., Beroza, G. C., Bailis 728
P. and Levis, P. (2019a). Unsupervised Large‐ Scale Search for Similar Earthquake Signals. Bull. 729
Seismol. Soc. Am. 109, 1451–1468. 730
Yoon, C. E., Huang, Y., Ellsworth, W. L., and Beroza, G. C. (2017). Seismicity during 731
the initial stages of the Guy‐ Greenbrier, Arkansas, earthquake sequence. J. Geophys. Res.: Solid 732
Earth 122, 9253–9274. 733
Yoon, C. E., O’Reilly, O., Bergen, K. J., and Beroza, G. C. (2015). Earthquake detection 734
through computationally efficient similarity search. Sci. Adv. 1, e1501057. 735
Yoon, C. E., Yoshimitsu, N., Ellsworth, W. L., and Beroza, G. C. (2019b). Foreshocks 736
and Mainshock Nucleation of the 1999 M w 7.1 Hector Mine, California, Earthquake. J. 737
Geophys. Res.: Solid Earth 124, 1569–1582. 738
Yun H., Min, K. D., Moon, H.-S., Lee H. K., and Yi, S. S. (1991). Biostratigraphic, 739
Chemostratigraphic, Paleomagnetostratigraphic, and Tertiary formations in southern part of 740
Korea: regional tectonics and its stratigraphical implication in the Pohang basin. J. Paleont. Soc. 741
Korea 7, 1–12 (in Korean). 742
Zoback, M. L. (1992). First‐ and second‐ order patterns of stress in the lithosphere: The 743
World Stress Map Project. J. Geophys. Res.: Solid Earth 97, 11703–11728. 744
745
34
Full mailing address for each author 746
1 School of Earth and Environmental Sciences, Seoul National University, Seoul 08826, 747
Republic of Korea 748
J.-U.W., [email protected] 749
J.R., [email protected] 750
2 Department of Structural Systems and Site Evaluation, Korea Institute of Nuclear Safety, 751
Daejeon 34412, South Korea 752
M.K., [email protected] 753
3 Division of Earth Environmental System Science, Pukyong National University, Busan 48513, 754
Republic of Korea 755
T.-S.K., [email protected] 756
757
758
35
Tables 759
760
Table 1. The 1-D layered seismic velocity structure for the Pohang EGS site. 761
Depth to the top of the layer (km) P-wave velocity (kms-1) S-wave velocity (kms-1)
0.0 1.67 0.48
0.203 4.01 2.21
0.67 5.08 3.03
2.4 5.45 3.07
3.4 5.85 3.31
7.7 5.91 3.51
12 6.44 3.70
34 8.05 4.60
762
763
36
Table 2. Parameters of the faults involved in the aftershock sequences. 764
Properties Fault 1 (F1) Fault 2 (F2) Fault 3 (F3) Fault 4 (F4)
Strike (°)
Median 222.7 207.4 223.1 17.0
Median absolute deviation 1.1 1.2 0.4 1.3
Dip (°)
Median 77.4 59.8 61.2 62.0
Median absolute deviation 2.0 1.3 0.6 2.2
Fault length (km) 2.8 2.4 3.4 1.9
Fault width (km) 1.9 3.5 2.9 1.3
765
37
Figures 766
767
768
Figure 1. Map of (a) temporary and (b) permanent seismic stations used for analysis of source 769
parameters, geologic lineaments, faults, and relocated hypocenters. Three surface ruptures near 770
the study area are illustrated in (a) (Song et al., 2015; Yun et al., 1991). The focal mechanism of 771
the mainshock that was determined from the polarity of first arrivals is illustrated in (b). (c) 772
shows the location of the Gyeongsang Basin (GB) and the Yeonil Basin (YB) where many 773
NENNE sinistral strike-slip surface ruptures and NW transfer faults have developed. The red 774
boxes in (b) and (c) represent the domain of (a) and (b), respectively. 775
38
776
Figure 2. (a) Distribution of the 3946 epicenters relocated via hypoDD (Waldhauser and 777
Ellsworth, 2000) by using traveltime differences. The earthquakes projected onto each of the 778
cross-sections A1-A2 to E1-E2 shown in (b) to (f) fall within the rectangles denoted by dashed 779
black lines in (a). The trajectory of two stimulation wells PX-1 and PX-2 are illustrated as gray 780
lines in (c) with open sections colored in blue and red. (b–f) Depth distribution of the relocated 781
39
hypocenters along the cross-sections of A1-A2 to E1-E2. Spatial distribution of seven focal 782
mechanism solutions are illustrated in (a) to (e). The compressional quadrants of the focal 783
mechanisms of the mainshock and two largest aftershocks (ML 4.3 and ML 4.6) are colored in red, 784
blue, and green, respectively. Possible interpretations for delineated faults from the aftershock 785
distribution are marked as gray lines in (b), (c), (d), and (e). The circles in (f) represent the 786
rupture radii of earthquakes with MRel > 1.5, assuming a stress drop of 5.6 MPa, which 787
corresponds to an approximated value for the mainshock estimated by the spectral ratio method 788
(Woo et al., 2019a). The red, blue, and green circles in (f) indicate the rupture size of the three 789
largest earthquakes with ML 5.4, 4.3, and 4.6, respectively. 790
791
40
792
Figure 3. (a) Determination of the scaling parameter c in Equation (1) from known ML 793
magnitudes. The amplitude ratios measured from two similar waveforms observed at a station 794
are measured and counted to estimate the scaling parameter for given MLs. The red line indicates 795
the scaling parameter c of 0.85 calculated from the slope of the first principal components 796
between magnitude differences and the ratio of amplitude divided by hypocenteral distances. (b) 797
Comparison between ML and MRel. The red line indicates identity relation. 798
799
41
800
Figure 4. (a) The distribution of earthquake magnitudes with their logarithmic origin time. The 801
three largest earthquakes (ML 5.4, 4.3, and 4.6) are denoted as red, blue, and green stars, 802
respectively. The time interval of each bin to measure the temporal changes of b-values and their 803
corresponding Mcs are illustrated as orange lines and squares. A set of 600 earthquakes 804
constitute a bin for measuring b-values and there is an overlap of 400 earthquakes between two 805
consecutive bins. We combined MRels obtained from Equation (1) with MLs of the three largest 806
earthquakes. (b) Temporal variations of seismic b-values. The standard deviations of each of the 807
magnitude bins are represented as vertical and horizontal error bars. The black, red, and blue dots 808
indicate the three bins illustrated in (c). The gray dots and error bars represent b-values and their 809
standard errors calculated based on a maximum MC of 0.8 for all bins. (c) Four examples of the 810
magnitude distribution and their corresponding G-R law fitting lines (black for the entire data set 811
42
and green, red, and blue for the three selected temporal bins of (b)). The filled circles indicate the 812
cutoff magnitudes of MC that honor Equation (1) for larger magnitudes. (d) Two dimensional 813
spatial variations of b-values at a vertical profile along both the apparent strike of 210°. 814
Hypocenters of three largest earthquakes are denoted as red, blue, and green stars, respectively. 815
Dashed lines and solid lines indicate the significant difference level of 5% with median and 816
conservative thresholds. (e) Two-σ interval distribution of probabilities that differences of paired 817
b-values (Δb) in (c) is insignificant. The Δbs are binned by 0.1. The difference is statistically 818
significant with a significance level of 5% if Δb > 0.14 for half the cases and Δb > 0.18 for all 819
cases. 820
43
821
Figure 5. (a) Two dimensional spatial variations of p-values for a cross-section along the 822
apparent strike 210° for earthquake sequences in period A. (b) Two dimensional spatial 823
variations of p-values for the same depth profile shown in (a), but for seismic sequences in 824
period B. The p-values of period B were estimated using the given aftershock rates of period A 825
when they were available (denoted by the two spatial bins contoured by black lines). (c and d) 826
Cumulative number of aftershocks with time and their corresponding Omori law plots for the 827
case of whole data sets (RA(t) and RB(t)) and the three examples of the spatial bins contoured by 828
blue, green, and red lines in (a). 829
830
44
831
Figure 6. Temporal distribution of seismicity presented in (a) plan view, (b) a depth profile along 832
the apparent strike of 210°, and (c) a unidirectional projection along the apparent strike. We 833
illustrate the radius of earthquakes with MRel ≥ 1.5 by assuming a circular crack rupture and a 834
stress drop of 5.6 MPa from Woo et al. (2019a). In (a) and (b), the location of the mainshock and 835
the two largest consecutive aftershocks of ML 4.6 and ML 5.4 are represented as red, blue and 836
green stars, respectively. The rupture radii of the three largest earthquakes are displayed in (c) 837
with colors to match the star symbols in (a) and (b). The trajectory of two stimulation wells PX-1 838
and PX-2 are illustrated as gray lines in (a) and (b). In (c), the earthquake density of each point 839
was measured as the number of earthquakes within a circle of 0.25-unit radius (showing decades 840
along the x-axis and km along the y-axis) from the given point. The x symbol in (c) represents 841
45
the moment at which the number of earthquakes in the 0.5-km bin reached 10 with a 0.1-km 842
sliding window. 843
46
844
Figure 7. Schematic diagram that illustrates four fault segments inferred from the hydraulic 845
stimulation wells of PX-1 and PX-2 and the distribution of aftershocks. The three cubes colored 846
in red, blue, and green show the hypocenters of the mainshock and the two largest aftershocks of 847
ML 4.3 and 4.6, respectively. The occurrence of the mainshock triggered seismicity on fault 848
segments F2 and F3, and possibly affected the re-activation of F1. Fault segment F4, located to 849
the southwest of F3, was not delineated until the largest aftershock (ML 4.6) occurred. 850
851
852
47
Appendices 853
854
Table A1. Fingerprint extraction parameters for the FAST algorithm to detect earthquakes with 855
waveform similarity. 856
Fingerprint extraction parameter Value
Time-series window length of generated spectrogram 6.0 s
Time-series window lag of generated spectrogram 0.1 s
Spectral image window length 64
Spectral image window lag 10
Fingerprint sparsity 400
Final spectral image width 32
Number of hash functions per hash table 4
Number of hash tables 100
Number of votes 2
Near-repeat exclusion parameters 5
857
Table A2. Input parameters for the network detection in the FAST algorithm (Bergen and Beroza, 858
2018; Rong et al., 2018) 859
Event-pair extraction, pruning, and network detection parameters values
Time gap along diagonal 3 s
Time gap adjacent diagonal 3 s
Adjacent diagonal merge iteration 2
48
Number of votes 10
Minimum fingerprint pairs 3
Maximum bounding-box width 5 s
Minimum number of stations for detection 1
Arrival time constraint: maximum time gap 5
860
Table A3. Focal mechanisms illustrated in Figure 2. 861
Origin time
(UTC, dd/mm/yy
HH:MM:SS.SS)
Latitude
(º)
Longitude
(º)
Depth
(km)
Strike
(º)
Dip
(º)
Rake
(º)
15/11/2017 05:29:31.61 36.10592 129.37215 4.245 211 40 128
15/11/2017 06:09:49.88 36.08742 129.34946 4.318 91 74 -132
15/11/2017 07:49:30.37 36.11412 129.36825 5.450 201 65 110
19/11/2017 14:45:47.79 36.11303 129.38023 3.865 34 85 149
19/11/2017 21:05:15.48 36.13384 129.37956 3.851 234 85 -174
25/12/2017 07:19:22.58 36.10206 129.36582 5.291 39 81 165
10/02/2018 20:03:03.74 36.07862 129.34237 4.447 34 52 136
862
49
Figure A1. Snapshots of the distribution maps of aftershocks at (a) 10 min, (b) 1 h, (c) 5 h, (d) 1 863
d, (e) 30 d, and (f) 4 mo from the onset of the mainshock. 864
865
1
Aftershock sequence and statistics of the 2017 MW 5.5 Pohang earthquake, South Korea: 1
implication of fault heterogeneity and post-seismic relaxation 2
3
Jeong-Ung Woo 1, Minook Kim2a, Junkee Rhie 1*, Tae-Seob Kang 3 4
Corresponding author: Junkee Rhie ([email protected]) 5
School of Earth and Environmental Sciences, Seoul National University, 1 Gwanak-ro, Gwanak-6
gu, Seoul 08826, Republic of Korea 7
Phone: +82-2-880-6731 8
Fax: +82-2-871-3269 9
1 School of Earth and Environmental Sciences, Seoul National University, Seoul 08826, Republic 10
of Korea 11
2 Department of Structural Systems and Site Evaluation, Korea Institute of Nuclear Safety, 12
Daejeon 34412, South Korea 13
3 Division of Earth Environmental System Science, Pukyong National University, Busan 48513, 14
Republic of Korea 15
16
Abstract 17
18
The sequence of foreshocks, mainshock, and aftershocks associated with a fault rupture are the 19
result of interactions of complex fault systems, the tectonic stress field, and fluid movement. 20
Analysis of shock sequences can aid our understanding of the spatial distribution and magnitude 21
a also at Division of Earth Environmental System Science, Pukyong National University, Busan
48513.
Manuscript Click here to access/download;Manuscript;manu_without_color.docx
2
of these factors, as well as providing a seismic hazard assessment. The 2017, MW 5.5 Pohang 22
earthquake sequence occurred following fluid-induced seismic activity at a nearby enhanced 23
geothermal system site and is an example of reactivation of a critically stressed fault system in 24
the Pohang Basin, South Korea. We created an earthquake catalog based on unsupervised data-25
mining and measuring the energy ratio between short- and long-window seismograms recorded 26
by a temporary seismic network. The spatial distribution of approximately 4,000 relocated 27
aftershocks revealed four fault segments striking southwestward. We also determined that the 28
three largest earthquakes (ML > 4) were located at the boundary of two fault segments. We infer 29
that locally concentrated stress at the junctions of the faults caused such large earthquakes and 30
that their ruptures on multiple segments can explain the high proportion of non-double couple 31
components. The area affected by aftershocks expands to the southwest and northeast by 0.5 and 32
1 km decade-1, respectively, which may result from post-seismic deformation or sequentially 33
transferred static Coulomb stress. The b-values of the Gutenberg-Richter relationship 34
temporarily increased for the first three days of the aftershock sequence, suggesting that the 35
stress field was perturbed. The b-values were generally low (< 1) and locally variable throughout 36
the aftershock area, which may be due to the complex fault structures and material properties. 37
Furthermore, the mapped p-values of the Omori law vary along strike, which may indicate 38
anisotropic expansion speeds in the aftershock region. 39
40
3
INTRODUCTION 41
On 15 November 2017, a moderate-sized earthquake of moment magnitude (MW) 5.5 or local 42
magnitude (ML) 5.4 struck the city of Pohang, located in the southeastern part of the Korean 43
Peninsula, which damaged infrastructure, injured 135 people, and made 1,797 homeless (Korea 44
Meteorological Administration, 2017). The earthquake (hereafter referred to as the mainshock) 45
was the second-largest earthquake event among earthquakes recorded instrumentally in South 46
Korea since 1978, according to the catalog of the Korea Meteorological Administration (KMA). 47
A close examination of the seismic source characteristics of such a rarely observed moderate-48
sized earthquake and its foreshock-mainshock-aftershock sequence is necessary not only to 49
evaluate the current stress field (Zoback, 1992; Soh et al., 2018) and fault properties, but also to 50
understand aftershock triggering mechanisms (King et al., 1994; Kilb et al., 2000). Estimation of 51
statistical parameters (i.e., the Gutenberg-Richter b-value and the Omori law p-value) from a 52
large number of microearthquakes in conjunction with the seismic source properties of 53
aftershocks can give information on fault heterogeneities, such as crack density, slip distribution, 54
applied shear stress, viscoelastic properties, and heat flow (Wiemer and Katsumata, 1999; Murru 55
et al., 2007). 56
One important point to note is that the mainshock occurred near an enhanced geothermal 57
system (EGS) site (Grigoli et al., 2018; Kim et al., 2018; Lee et al., 2019). A body of evidence 58
supports the claim that the mainshock was triggered by five fluid-injection experiments as well 59
as an associated loss of heavy drilling muds and released tectonic energy on a critically stressed 60
fault (Ellsworth et al., 2019; Woo et al., 2019a). The periods of stimulation experiments 61
conducted on two hydraulic wells (PX-1 and PX-2) were closely correlated with microseismicity 62
observed near the wells. Induced seismicity mapped in the vicinity of the EGS indicated the 63
4
presence of a previously unmapped fault. Microseismicity triggered on this fault migrated to the 64
location of the mainshock. A breakout was observed in the PX-2 well at intervals corresponding 65
to the assumed fault. The groundwater levels of PX-1 and PX-2 decreased abruptly by 121 m and 66
793 m, respectively, immediately after the mainshock but gradually recovered by 0.078 m d-1 67
and 0.198 m d-1, respectively (Lee, 2019). 68
Previous studies of aftershock distributions in the Pohang Basin determined the presence of 69
complex fault geometries (Hong et al., 2018; Kim et al., 2019). Separately, Grigoli et al. (2018) 70
reported that obtaining a significant non-double-couple (non-DC) component when inverting the 71
moment tensor for a mainshock can be attributed to the complexity of the rupture process in a 72
multi-fault system. The spatial pattern of early aftershocks associated with two 2016 Gyeongju 73
earthquakes (ML 5.1 and ML 5.8), which occurred on two sub-parallel faults approximately 40 74
km away from the Pohang mainshock, is differentiated from the presence of two or three fault 75
segments with varying strikes and dips for the early aftershocks associated with the 2017 Pohang 76
earthquakes (Uchide and Song, 2018; Son et al., 2018; Woo et al., 2019b). 77
In this study, we created an earthquake catalog for the foreshock-mainshock-aftershock 78
sequence from data recorded by local permanent seismic networks, temporary seismometers 79
deployed as part of the aftershock monitoring system, and the temporary Pohang EGS 80
monitoring system. Earthquakes were detected using a machine-learning data mining technique 81
for data obtained during the first ten days and a conventional automatic detection algorithm was 82
employed for the aftershock monitoring system as a whole. Each detected earthquake was 83
located by manual picking and visual inspection and then precisely relocated by the double-84
difference method (Waldhauser and Ellsworth, 2000). Using the spatial distribution of over 85
4,000 earthquakes, we modeled fault systems as a series of multiple fault segments by mapping 86
5
the spatio-temporal distribution of the Gutenberg-Richter b-value and the Omori law p-value 87
statistical parameters. 88
Mapping the distribution of earthquake magnitudes provides an independent analysis of the 89
characteristics of aftershock activities and can be used to analyze spatial heterogeneities of 90
material properties, such as stress state, level of asperities, and heat flow rate (Scholz, 1968; 91
Wiemer and Katsumata, 1999; Wiemer and Wyss 1997; Ávila-Barrientos et al., 2015); assess 92
seismic hazards via epidemic-type aftershock sequence modeling (ETAS; Ogata, 1998); and 93
conduct probabilistic seismic hazard analysis (PSHA; Cornell, 1968). In this study, we evaluated 94
the relative magnitude of each earthquake by using amplitude ratios relative to earthquakes of 95
known ML. 96
97
DATA AND METHOD 98
Seismic Networks 99
Continuous seismic waveform data used to detect and analyze seismic source parameters were 100
collected from four different networks (Figure 1). The first data set was obtained from a 101
combined permanent seismic network operated by KMA, the Korea Institute of Geoscience and 102
Mineral Resources (KIGAM) and the Korea Hydro and Nuclear Power (KHNP). The permanent 103
seismic networks of KMA, KIGAM, and KHNP are named KS, KG, and KN, respectively. The 104
second set of continuous waveform data were recorded by nine borehole seismometers installed 105
at depths of between 100 and 150 m, which operated to monitor microseismic events for the 106
Pohang EGS project. Three of the temporary borehole seismometers recorded the mainshock, 107
while the operation of the other borehole seismometers started within the next 2 days; all of them 108
operated until the end of November 2017. The third continuous waveform data set was collected 109
6
by 37 temporary broad-band seismometers installed after the mainshock by the university 110
consortium (Pukyong National University and Seoul National University) and KIGAM 111
independently. The first seismometer installed temporarily for monitoring aftershocks started its 112
operation approximately 1 h after the onset of the mainshock. Lastly, we used waveforms of 214 113
early aftershocks, occurred within four hours from the mainshock, recorded at eight short-period 114
temporary seismometers deployed by Pusan National University (Kim et al., 2018). The 115
seismograph stations of these temporary networks were densely spaced and located within the 116
radius of 20 km from the EGS site, respectively (Figure 1). 117
118
Detection and Hypocenter Determination 119
Since stabilizing temporary seismometers for aftershock monitoring can take many hours, 120
conventional algorithms for earthquake detection, such as STA/LTA (Withers et al., 1998; 121
Trnkoczy 2002), are of limited use for locating early aftershocks because of the incompleteness 122
of the local seismometer network. In this study, we utilized the Fingerprint and Similarity 123
Threshold (FAST) data-mining algorithm that uses waveform similarity to detect such early 124
aftershock sequences (Yoon et al., 2015; Yoon et al., 2017; Bergen et al., 2018) with a 125
conventional energy-based algorithm for the period for aftershock monitoring system. The FAST 126
algorithm finds pairs of waveforms having similar spectrograms without any prior information, 127
allowing us to obtain pairs of earthquake candidates with correlative signals. The performance of 128
the FAST algorithm to discriminate true events from earthquake candidates can be improved by 129
measuring similarity at multiple stations (Bergen et al., 2018). 130
We applied the FAST method to ten days of continuous seismograms recorded between 14 Nov 131
2017 and 23 Nov 2017 to cover the period of operation of the aftershock monitoring system. We 132
7
used three-component seismograms obtained from two short-period (PHA2 and DKJ) and one 133
broadband (CHS) seismometers, which are located within 30 km from the mainshock. The three 134
borehole seismometers of the EGS monitoring system that were operational at the onset of the 135
mainshock were not used in detection due to the high level of ambient background noise and 136
regularly observed pulse-like signals. The sampling rate of the seismograms was fixed at 100 Hz 137
and the frequency range of the bandpass filter was set to 2 – 20 Hz. All parameters employed in 138
the FAST algorithm routines were either determined manually from performance trials and were 139
previously applied values (Yoon et al., 2017; Yoon et al., 2019a) and are summarized in Table 140
A1 and A2. 141
We detected 1,580 candidate events via the FAST search, leading to a subset of 1,357 locatable 142
earthquakes from visual inspection. Compared with the earthquake catalog published by Kim et 143
al. (2018), which utilized eight local seismographs deployed within 3 km of the EGS site for 144
earthquake detection, the FAST algorithm successfully detected 169 out of 217 or 78% of 145
earthquakes for their overlapping period. 146
While the aftershock monitoring network was operational (i.e., from 15 November 2017 to 28 147
February 2018), we applied an automatic algorithm to detect and locate microseismic 148
earthquakes (Sabbione and Velis, 2013). Continuous waveforms were transformed into 149
characteristic functions for measuring the ratio between the short-term average (STA) and the 150
long-term average (LTA). We declared candidate earthquakes when the STA/LTA ratio 151
exceeded 5 for a given time window of 4 s at more than three stations. For each triggered time 152
window, the normalized squared envelope functions of Baer and Kradolfer (1987) were 153
calculated to determine the time at which to maximize the function value (hereafter referred to as 154
the BK function). Since the BK function can be maximized for the arrivals of either the P-wave 155
8
or the S-wave, the maximum value of the BK function was tested to discriminate whether the 156
measured local maximum corresponded to the first arrival. If we observed a local high BK 157
function value before the maximum of the BK function in a given time window, we set two 158
consecutive time samples as the arrivals of the P- and S-waves. Otherwise, we searched for other 159
local maximum after the triggered time window and set the maxima as the P- and S-wave phase 160
arrivals when a secondary maximum was available. The phase arrivals determined in this way 161
were visually confirmed by using a Wadati plot (Wadati, 1933). 162
We determined the initial hypocenters of the detected earthquakes via Hypoellipse (Lahr, 1999), 163
with phase arrival times being determined by manual inspection and a 1-D layered seismic 164
velocity model for the Pohang EGS site (Woo et al., 2019a; Table 1). In this procedure, we 165
combined the earthquakes detected from either the FAST algorithm or the STA/LTA method 166
with events with ML > 2.0 listed in the KMA and Kim et al. (2018) event catalogs. Earthquakes 167
with an onset difference of less than 2 s were regarded as duplicate events. Station corrections 168
were calculated based on a comparison of the theoretical arrival times for five immediate 169
foreshocks reported by Woo et al. (2019a) and their picked arrival times. 170
Initial hypocenters were relocated with hypoDD (Waldhauser and Ellsworth, 2000) by using 171
travel time differences obtained from waveform cross-correlation measurements as well as 172
picked phase times as inputs to the double-difference algorithm. The 1-D velocity model of Woo 173
et al. (2019a) was applied for the relocation procedure, again (Table 1). All relocated events were 174
shifted by 39 m, 28 m, and 96 m in eastwards, northwards, and downwards, respectively, to 175
match the centroid of the five immediate foreshocks with the results of Woo et al. (2019a), of 176
which recordings at 17 PX-2 borehole chains were applied to obtain accurate hypocenters. We 177
resampled waveforms to 1,000 Hz with a cubic spline after first having applied a 2–10 Hz 178
9
bandpass filter. Each seismogram was reduced to a 1 s time window centered at each phase 179
arrival time. We allow a time shift up to 0.1 s for the cross-correlation measurements. Time 180
shifts that maximized the cross-correlation coefficient (CC) between two pairs of waveforms 181
were used only if the maximum CCs were greater than 0.85. The squared maximum CCs were 182
used to weight the measurements. The relative locations were calculated by least-squares fitting 183
of the data and the location uncertainties were evaluated by using bootstrapping analysis 184
(Waldhauser and Ellsworth, 2000). Synthetic travel time differences between paired events were 185
reconstructed by random selection of a set of residuals and relative locations for these synthetic 186
travel times were calculated 200 times. 187
188
Magnitude Estimation and Statistical Analysis 189
Waveform similarity can be assessed to estimate the relative magnitudes of earthquakes (Shelly 190
et al., 2016; Yoon et al., 2019b). We adopted a simple magnitude-amplitude relationship 191
modified from the equation of Shelly et al. (2016) that considers the differences in hypocentral 192
distance between two earthquakes: 193
dm = clog10(a/r), (1) 194
where dm, a, and r represent the ratios of magnitude, amplitude, and hypocenteral distance, and c 195
is a coefficient for the magnitude-amplitude relationship (Shelly et al. 2016). The coefficient c in 196
Equation 1 varies with the earthquake magnitude scale that is used: for example, Shelly et al. 197
(2016) reported that c = 1 for ML and c = 2/3 for MW. In this study, we used a set of MLs of 198
aftershocks and Equation (1) to estimate the coefficient c, following the method of Woo et al. 199
(2019b). If the CC of a waveform pair was greater than 0.85, we calculated the amplitude ratio as 200
the slope of the eigenvector for the largest eigenvalue of the covariance matrix of the two 201
10
waveforms (Shelly et al., 2016). Thus, for earthquakes with known values of ML, we were able to 202
estimate the parameter c. 203
We can also determine relative magnitudes of earthquakes by using our estimated value of c in 204
Equation (1). Estimated relative magnitudes (MRel) were arithmetically averaged to produce a 205
representative value and uncertainties were obtained from their standard deviations. 206
The Gutenberg-Richter law (G-R law) describes the relationship between earthquake frequency 207
and magnitude. Its statistical properties are widely accepted and applied to the investigation of 208
seismo-tectonic properties in a specific region over a certain time period. Examples of 209
application of the G-R law include work on aftershock sequences by Wiemer and Katsumata 210
(1999) and Woo et al. (2019b), on earthquake swarms by Farrell et al. (2009), on induced 211
seismicity by Shapiro (2007), and in laboratory experiments by Scholz (1968). The earthquake 212
frequency distribution with magnitude can be written as: 213
log10 N(≥ M) = a - bM, (2) 214
where N is the number of earthquakes equal to or greater than a magnitude M, and a and b are 215
scaling constants. a is proportional to the overall seismicity in a given spatio-temporal interval, 216
whereas b represents the ratio of the number of large earthquakes to small earthquakes. The 217
behavior of b-values has been attributed to crack density (Mogi, 1962), stress drop (Wyss, 1973), 218
and tectonic stresses (Schorlemmer et al., 2005, Scholz, 1968), and slip distribution (Wiemer and 219
Katsumata, 1999). 220
We determined the magnitude of completeness (MC) for 3,521 magnitudes based on a modified 221
goodness-of-fit method of Wiemer and Wyss (2000), following Woo et al. (2019b). Then, we 222
evaluated the b-value for a set of magnitudes using the maximum likelihood method of Aki 223
11
(1965) with a magnitude bin of 0.1. The uncertainty of b-values was estimated with the method 224
of Shi and Bolt (1982). 225
Omori’s law describes the decay rate of aftershocks. Its parameters are also broadly applied to 226
interpret regional seismic and tectonic properties (Omori, 1894; Utsu, 1961). The extended form 227
of Omori’s law can be written as: 228
R(t) = K(t+c)-p, (3) 229
where K, c, and p are the scaling coefficients that describe the aftershock decay rates in a given 230
region. p, which represents the power of the aftershock decay rates, has a range of 0.6 to 1.8 and 231
is considered to be a function of stress and temperature in the crust (Utsu and Ogata, 1995; 232
Wiemer and Katsumura 1999). We mapped the spatial variation of p-values by binning 250 233
magnitudes and by selecting magnitudes greater than MC. The three parameters and their 234
associated uncertainties were determined following the maximum likelihood method presented 235
by Ogata (1983). 236
237
RESULTS 238
Of the 4,446 earthquakes with initial locations, we relocated seven foreshocks, the mainshock, 239
and 3,938 aftershocks using hypoDD (Waldhauser and Ellsworth, 2000), having excluded 240
earthquakes with fewer than seven traveltime difference measurements. Uncertainties of relative 241
locations to within two standard deviations were estimated as 25 m in the eastwest direction, 18 242
m in the northsouth direction, and 37 m vertically. Figure 2 presents the spatial distribution of 243
aftershocks, both in map view and cross-sections, four in the dip direction (A1-A2, B1-B2, C1-C2, 244
and D1-D2) and one in the strike direction (E1-E2). From the map, we determined the apparent 245
strike of aftershocks (crossline of E1-E2) to be 210°, which corresponds to the azimuth of the first 246
12
principal vector obtained from two-dimensional principal component analysis (PCA) (Jollifle, 247
2002). From cross-sections in the dip direction (A1-A2 to D1-D2), we observed that the spatial 248
distribution of aftershocks delineates at least four different fault segments (Figure 2). In the most 249
northeastern part of the study area, a sub-vertical fault was identified from the aftershock 250
distribution. An ML 3.5 earthquake on 21:05:15, 19 November 2017; UTC with a focal 251
mechanism (strike: 234°, dip: 85°, rake: -174°) is consistent with the inferred fault (Figure 2b, 252
Table A3). Among the relocated earthquakes, the first observed event on the fault plane occurred 253
within 72 s of the onset of the mainshock (Figure A1), which indicates that reactivation of the 254
fault segment was initiated by the mainshock rupture or soon afterward. Two slightly different 255
fault geometries, both dipping northwestward, are distinguished in the middle of sections B1-B2 256
and C1-C2 from the spatial distribution of the aftershocks. The aftershock distribution along B1-257
B2 has a wider range of focal depths, a shallower dip, and a strike closer to north-south than that 258
of C1-C2. Both the mainshock and the ML 4.3 aftershock are located adjacent to a virtual 259
boundary of B1-B2 and C1-C2 and their focal mechanisms are consistent with the observed fault 260
geometry. Earthquakes in the southwestern part of D1-D2 occurred after the largest aftershock 261
(ML 4.6) (Figure A1) and their focal depths deepened to the south-east, dipping in the opposite 262
direction to the three other fault segments observed on A1-A2, B1-B2 and C1-C2. Such a conjugate 263
fault geometry is matched with one nodal plane of the focal mechanism (strike: 34°, dip: 52°, 264
rake: 136°) of the largest ML 4.6 aftershock (Figure 2e, Table A3). 265
From the complex fault geometry delineated by the four cross-sections, we constructed a 266
simplified fault model to describe the observed aftershock distribution. For the three segments 267
that reactivated with the occurrence of the mainshock, we described their geometry using the 268
aftershocks that occurred within one day of the mainshock. Because the mainshock was situated 269
13
on a virtual boundary between two faults (F2 and F3) with slightly different strikes and dips, we 270
divided the aftershock area based on the hypocenter of the mainshock and an apparent strike of 271
210°, which we estimated from PCA of data in map view. The aftershocks on the most 272
northeasterly fault segment (F1) were de-clustered from the adjoining fault (F2) using the simple 273
assumption that the Heunghae Fault (i.e., Song et al., 2015; Yun et al., 1991) vertically intersects 274
them both. Earthquakes that occurred up to 1 day after the largest ML 4.6 aftershock and are 275
located within 2 km from the event were used to investigate the most southwesterly fault 276
segment (F4). Faults F1F4 were used to divide the study area into four regions and earthquakes 277
were assigned to a region on the basis of the location of their hypocenter. We applied PCA 278
analysis with bootstrapping to earthquakes that were resampled 200 times to estimate strike, dip, 279
fault length, and fault width. We rotated the first and second principal components to the two 280
unit direction vectors for strike and dip; thus defining the strike and dip components of 281
earthquakes as these projections. The fault length and width were then determined as the 282
difference between the 2.5th and 97.5th percentiles of the strike and dip components. The 283
resulting fault geometry is summarized in Table 2. 284
We determined c using the 266 relocated earthquakes with known ML. We evaluated c as 0.85 285
by PCA (Figure 3), which is larger than the case for the MW magnitude scale (c = 2/3) scale but 286
smaller than the case for the ML magnitude scale (c = 1). The difference in c implies that the ML 287
magnitude does not naturally match MW for earthquakes within the range of magnitudes included 288
in this study, filtered to a frequency range of 2 – 10 Hz. The observed value of c is relatively 289
high compared with 0.7 that was estimation using the MLs of the Gyeongju aftershock sequences 290
(Woo et al., 2019b), which may be the result of systematic differences between ML and KMA 291
magnitude. 292
14
We estimated the magnitudes of 3,521 earthquakes with measurements ≥ 5. Figures 3b 293
illustrates the comparison of MRel with ML and the variations of MRel with time. Since MRel is 294
exactly proportional to ML without any scaling parameters, we propose that MRel can replace ML 295
as the magnitude scale for subsequent analysis. 296
We examined temporal variations of seismic b-values by binning 600 earthquake magnitudes 297
(MRel) into a set (Figure 4a). There was an overlap of four hundred earthquakes between two 298
consecutive bins. The MC decreased from 0.8 to 0.2 during the first 3 days of the early aftershock 299
sequence, which is indicative of a decrease in the background noise level for that period (Hainzl 300
2016). The b-value for the first bin was evaluated as 0.66 ± 0.03, which is consistent with b-301
values for earthquakes detected during fluid injection into the Pohang EGS site before the 302
occurrence of the mainshock (Woo et al., 2019a). The b-value increased with time for the first 303
three days up to a maximum of 0.98 ± 0.05 and fluctuated during a month. After three months, it 304
decreased to 0.77 ± 0.04 when the largest aftershock of ML 4.6 occurred. We tested the temporal 305
changes of b-values with a fixed MC of 0.8, corresponding to the maximum values over the 306
whole period, to investigate whether the observed temporal variations of b-values were biased by 307
the choice of MC (gray dots of Figure 4b) and confirmed that the main features were not 308
significantly changed. Figure 4c illustrates the magnitude-frequency distributions of three data 309
sets highlighted in Figure 4b. 310
The spatial variation of b-values was investigated for the vertical cross-section along the 311
apparent strike of 210°. Earthquakes within 1.0 km from the center of each 0.5 0.5 km grid cell 312
on the cross-section were binned. We analyzed the MC and b-value only if each bin contained at 313
least 250 earthquakes. Figure 4d illustrates the spatial distribution of b-values on the vertical 314
cross-section. The estimated b-values are between 0.63 and 0.91, all of which are lower than the 315
15
typically assumed b-value of 1 (Wyss, 1973). Since ML is approximated by MRel, such low b-316
values can be interpreted as an increase in applied shear stress and effective stress (Scholz, 1968; 317
Wyss 1973), low material heterogeneity (Mogi, 1962), or a high stress drop (Wyss, 1973). 318
Considering that the slip tendency of the mainshock is indicative of a critically stressed fault 319
(Ellsworth et al., 2019; Lee et al., 2019) and the stress drop of 5.6 MPa for the mainshock is not 320
higher than that of other earthquakes in South Korea (Rhee and Sheen 2016; Woo et al., 2019a), 321
our preferred interpretation is that the generally low b-values in the aftershock area may result 322
from high applied stress in this region. We estimated a b-value of 0.73 ± 0.04 near the 323
hypocenter of the mainshock, which is comparable to the values observed for the earthquakes 324
during the fluid injection (= 0.66 ± 0.08). 325
The significance of temporal and spatial differences in b-values can be verified by Utsu’s test 326
(Utsu, 1992), in which the probability that the b-values between two sets of earthquakes are the 327
same is defined via Akaike Information Criterion (Akaike, 1974). We first tested the statistical 328
significance of the temporal differences of b-values among early (< 1 d), intermediate (~ 3 d), 329
and late aftershocks (~ 90 d), which are highlighted in green, red, and blue, respectively, in 330
Figures 4b and 4c. The probability that the b-value for the intermediate period is not significantly 331
higher than those of the early and late aftershocks was estimated as 7.3×10-7 and 1.6×10-3, 332
respectively, indicating that the temporal increase and decrease of b-values are statistically 333
reasonable with a significance level of 5%. Similar variations of b-values with time can be found 334
for the 2016 Gyeongju earthquake (Woo et al., 2019b) and other cases (Smith, 1981; Chan et al., 335
2012; Gulia et al., 2018), which can be interpreted as local stress changes due to the mainshock 336
rupture or a mixed effect of a changing spatial distribution of b-value and a heterogeneous 337
population of aftershocks with time (Figure 4d). 338
16
We also applied Utsu’s test to all pairs of spatially varying b-values and measured the 339
distribution of significant levels with the b-value difference (Δb) bin of 0.01. A significance level 340
of 5% was held for Δb > 0.14 in half of the cases and Δb > 0.18 in all cases (Figure 4e). 341
Therefore, we roughly designated three sub-regions: R1 with relatively low b-values, and R2 and 342
R3 with high b-values (Figure 4d). The ML 4.3 and ML 4.6 earthquakes are located near R2 and 343
R3, which can be interpreted as indicating material heterogeneity with respect to the conjugate 344
fault system (Figures 2c and 2e). Alternatively, spatial variations of pore pressure or applied 345
stress may contribute to b-value heterogeneity. 346
The p-values were estimated for two data sets: (1) period A, between the onset of the 347
mainshock and the ML 4.6 aftershock; and (2) period B, after the onset of the largest aftershock 348
of ML 4.6 (Figure 5). This grouping was chosen because the occurrence of the largest aftershock 349
at ~87 days resulted in increased seismicity, which reset the decay rate for the mainshock. For 350
each data set, we estimated the p-value that represents the whole data set and the spatial variation 351
of p-values at the cross-sections along the apparent strike of 210°, with the same bins used for 352
estimating the spatial variations of b-values. The p-values of the period B were estimated with 353
the consideration of decaying aftershock rates of the period A. The p-value of period A was 354
estimated as 1.10, which is larger than the value for period B (= 0.88). Such a difference may 355
result from differing initial stress levels for periods A and B with respect to the stress 356
perturbation of the mainshock sequence, spatial heterogeneity of the internal structure for the 357
conjugate fault system (Figure 2e; Wiemer and Katsumata, 1999), or just an insufficient number 358
of earthquakes in the calculation of p-values for period B. With the exception of p-values for 359
period B, the p-values of the period A were higher in the southwestern region than those in the 360
northeastern region (Figure 5a). This could be indicative of a spatial variation of heat flow 361
17
(Kisslinger and Jones, 1991), heterogeneity of fault strength (Mikumo and Miyatake, 1979) or an 362
insufficient number of aftershocks to allow accurate fitting of the aftershock power decay law for 363
the southwestern aftershock region prior to the occurrence of the ML 4.6 aftershock. 364
365
DISCUSSION 366
Expansion of aftershock areas with time 367
Expansion of early aftershock sequences is widely observed (Tajima and Kanamori 1985; 368
Fukuyama et al., 2003; Peng and Zhao, 2009; Kato and Obara, 2014; Hainzl et al., 2016). Some 369
temporal evolution of aftershock areas have been interpreted to be the result of afterslip or post-370
seismic deformation (Helmstetter and Shaw 2009; Peng and Zhao 2009; Ross et al., 2017; 371
Perfettini et al., 2018). Speeds of along-strike expansion of the aftershock zone were measured 372
on a logarithmic time scale and showed that propagating afterslip can cause the expansion of 373
aftershocks (Peng and Zhao 2009; Frank et al., 2017; Perfettini et al., 2018; Ross et al., 2017). In 374
the present study, we examined the spatio-temporal distribution of aftershocks on a logarithmic 375
time scale to consider possible post-seismic deformation following the mainshock (Figure 6). In 376
a map view, we observed that the aftershock zone has roughly expanded along the apparent 377
strike direction, especially during the first day (Figures 6a and c), whereas no clear trends were 378
observed in a vertical sense (Figure 6b). 379
The general speed of virtual aftershock migration fronts for the bilateral expansion along the 380
strike direction were ~1 km decade-1 northeastward and ~0.5 km decade-1 southwestward (Figure 381
6c), which may indicate post-seismic deformations related to aseismic afterslip (Peng and Zhao 382
2009; Perfettini et al., 2018). The difference in the migration speeds can be attributed to different 383
rate-and-state parameters described by Dieterich (1994) following the equations published by 384
18
Perfettini et al. (2018). However, in our case, we also observe a significant p-value variation in 385
the northeastern and southwestern parts of the study area (Figure 5a). Such variations of p-value 386
require a different model rather than the rate-dependent friction law (Helmstetter and Shaw 2009; 387
Mignan 2015). Assuming that the power-law rheology governs post-seismic velocity which is 388
proportional to (1+t/t*)-p (Montési, 2004), where t* is a characteristic time of the aftershock, the 389
slip velocity or the aftershock occurrence rate decays with time as a power of p. The slowly 390
decreasing slip velocity with a lower p-value generate a larger accumulated displacement than 391
that with a higher p-value in a given time period, and thus the time required to rupture asperities 392
is shorter than that with a higher p-value. 393
Therefore, the p-value variation observed for the aftershock area during the period A may be 394
related to the different seismic migration speeds (Figures 5a and 6c). We did not further compare 395
p-values and the migration speed in this study, since it may require more complex analysis than a 396
simplified form of the Omori’s law (Narteau et al., 2002). Furthermore, there is an absence of 397
data for very early (« 1 d) or late (> 100 d) aftershocks. 398
The expansion of the aftershock zone can also be explained by a cascade of sequentially 399
triggered aftershocks in terms of changes to the static Coulomb stress (Ellsworth and Bulut, 400
2018). These mechanisms can also explain very early aftershocks deviated from the expansion of 401
aftershock area at around 3 km. Since no clear evidence of post-seismic deformation was 402
observed in the differential InSAR analysis during 12 days of the post-seismic period (Song and 403
Lee, 2019), the observed expansion of aftershocks could possibly be attributed to changes to the 404
static stress field caused by the aftershock sequences rather than a result of aseismic deformation. 405
However, post-seismic deformation during the first day of the aftershock period might not be 406
19
captured in InSAR data because most of the expansion of the aftershock would be limited to 407
observations within 1 d from the mainshock. 408
409
High percentage of non-DC components observed for the mainshock and two largest 410
aftershocks 411
The moment tensor solutions of the mainshock and two largest earthquakes have high 412
percentages (> 30%) of non-DC components (Grigoli et al., 2018; Hong et al., 2018), in contrast 413
to the normally observed moment tensor solutions in South Korea. Such high non-DC 414
components of the moment tensor solutions of the three largest earthquakes can result from 415
complex shear faulting of multiple DCs, tensile opening/closing, and shear faulting in anisotropic 416
and heterogeneous media (Miller et al., 1998). It has already been established that the spatial 417
distribution of the Pohang earthquake sequence indicates that multiple fault segments were 418
reactivated in a complex fault system and the faulting types of the focal mechanism vary 419
throughout the aftershock area (i.e., Kim et al., 2019). Hence, a combination of multiple DC 420
moment tensor solutions with varying senses of slip motion could be one of the causes of the 421
three largest earthquakes having high non-DC components. 422
We propose the following sequence of events to explain the mainshock and major aftershock 423
sequence associated with the MW 5.5 Pohang earthquake. We infer that the nucleation of the 424
mainshock rupture was initiated at the junction between F2 and F3 and that the rupture 425
propagated along F2 and F3 with possible intervention of F1. Later, the ML 4.3 earthquake was 426
initiated between two adjacent conjugate faults dipping southwestward and northeastward in the 427
deeper aftershock region below the mainshock (Figure 2c). Finally, the ML 4.6 earthquake 428
nucleated at the southwestern tip of the aftershock area and subsequent aftershocks occurred on a 429
20
previously unrecorded southeastward dipping fault, suggesting that the rupture of the ML 4.6 430
earthquake sequences was initiated at the intersection of conjugate faults F3 and F4. 431
Although the three earthquakes were located at the intersection of multiple fault planes, it is hard 432
to envisage that all the earthquakes located in the surrounding area ruptured on multiple fault 433
planes. Some MRel 3 3.6 earthquakes without non-DC components were located in the vicinity 434
of the interconnecting faults (Choi et al., 2019), which may suggest that a certain amount of 435
seismic energy is required for the simultaneous movement of multiple fault segments. The fault 436
dimensions for the three largest earthquakes are inferred to be greater than 1 km, based on the 437
assumption of a constant stress drop of 5.6 MPa on a circular crack (i.e., Figure 2f), leading us to 438
propose that a kilometer rupture scale is the threshold to rupture multiple fault planes. Low b-439
values observed throughout the aftershock area can be considered as stress concentrations within 440
areas of high asperities (Wiemer and Wyss, 1997). High asperities in the regions adjoining two 441
or more fault segments may concentrate tectonic energy either as an earthquake nucleation point 442
or as barriers to rupture propagation. This may explain why only ML > 4 non-DC component 443
earthquakes were observed. The sonic log data of the PX-2 borehole recorded the existence of 444
anisotropic structures in the Pohang Basin (Ellsworth et al., 2019). Such anisotropic materials 445
can also cause earthquakes with high non-DC components. However, it is our preferred 446
interpretation that non-DC components in the three largest earthquakes result from the fault 447
complexity because low, non-DC earthquakes for ML 3 – 3.6 earthquakes were also observed. 448
449
Comparison between aftershock activities and induced seismicity at the EGS site during 450
stimulation. 451
21
The seismicity recorded during the five hydraulic stimulation experiments at the Pohang EGS 452
site and the inferred focal mechanisms revealed a fault plane located near the PX-2 well (Woo et 453
al., 2019a). PX-2 seismicity was clustered on a plane with a strike of 214° and a dip of 43° and 454
migrated southwestward, heading toward the location of the mainshock (Woo et al., 2019a). 455
However, the fault geometry for the induced earthquakes related to the PX-2 well has a 20° 456
shallower dip angle than the moment tensor solution of the mainshock and aftershocks. It 457
suggests that complex fault segments exist locally throughout the aftershock region and that a 458
simple fault plane does not explain the detailed fault structures. The ML 4.3 earthquakes have 459
deeper focal depths and their focal mechanism has steeper dips than that of the mainshock, which 460
can also be regarded as a result of complex fault geometry. Observation of various types of focal 461
mechanisms in aftershock sequences (Kim et al., 2019) are also a manifestation of the complex 462
geometry, which is in contrast to the nearly identical focal mechanisms for the PX-2 seismicity 463
(Woo et al., 2019a). 464
The b-values observed during the Pohang EGS project have insignificant variations, with an 465
average value of 0.66 ± 0.08 (Woo et al., 2019a, Langenbruch et al., 2020); whereas, the b-466
values estimated for the early aftershock sequences are statistically different from the b-values 467
for a bin of approximately 3 days after the mainshock (Figure 4b). If we assume that b-values act 468
as a stress-meter (Scholz, 2015; Rigo et al., 2018; Woo et al., 2019b) and temporal variation of 469
b-values during the aftershock period represents the level of stress state, the invariant b-values 470
observed during the stimulation period suggest that stress perturbations caused by fluid injection 471
may be far lower than the accumulated tectonic stress, indicating the existence of a critically 472
stressed fault system before the mainshock. 473
474
22
Reactivation of a multi-segment fault system and spatial variations of b-values and p-values 475
The complexity of the Pohang aftershock distributions was modeled as four fault segments, 476
following the approach of Hong et al. (2018) and Kim et al. (2018, 2019) (Figure 7). The 477
seismicity along a subvertical fault, F1, in the northeastern of the study area clearly represents 478
migration of the aftershock front northeastward during the first day of the aftershock sequences 479
(Figure 6). Although this fault plane is located about 3 km away from the mainshock hypocenter, 480
it may have been reactivated as a part of the mainshock rupture process. Alternatively, it may 481
have been dynamically triggered by the mainshock considering circumstantial evidence that 482
aftershock activity on the fault segments was initiated within just 2 min (Figure A1) and the slip 483
distribution of the mainshock calculated from the static deformations with InSAR data is largest 484
in the northeastern part of the fault model (Song and Lee, 2019). Aftershocks on F1 are bounded 485
by the Heunghae Fault, which has surface expression (Figure 1), detaching F1 from F2 and F3. 486
Therefore, in either case, the reactivation of F1 may require a certain stress threshold to be 487
ruptured preferentially to F2 and F3. 488
Two slightly different geometries of F2 and F3 are suggested by Hong et al. (2018), reflecting a 489
complex fault system near the Pohang EGS site. While the b-values vary slightly on F2, the 490
observed p-values were higher for F3, at least until the occurrence of the ML 4.6 event. The 491
different behaviors of the two statistical parameters imply that the two fault segments exist under 492
different physical conditions, such as: differential stress states (Scholz, 1968), local 493
heterogeneity of the rock matrix that may interact with viscous materials (Wyss, 1973; Bayrak et 494
al., 2013), or variable spatial distribution of heat flow (Kisslinger and Jones, 1991). 495
The b-values decreased to ~0.7 when fault segment F4 was reactivated by the ML 4.6 aftershock. 496
The lower b-values may indicate F4 was already highly stressed when the ML 4.6 earthquake was 497
23
triggered. The observed p-values for period B were generally much lower than those for period A, 498
which may be the result of using short time periods for analysis during period B or just uneven 499
seismicity observed for periods A and B. 500
501
CONCLUSIONS 502
In this study, we detected over 4,000 earthquakes related to the MW 5.5 (ML 5.4) Pohang 503
earthquake by using both unsupervised data-mining and a conventional automatic earthquake 504
detection method. From the spatio-temporal distribution of relocated seismicity, we observed 505
that four fault segments were responsible for the aftershocks. All the faults strike 506
northeastsouthwest, but have different dip angles and dip directions. The three largest 507
earthquakes are located at the boundaries of two adjoining fault segments, which may have 508
focused the stress released by multiple faults, resulting in high, non-DC earthquake mechanisms. 509
By measuring amplitude ratios between two similar earthquakes, we estimated relative 510
magnitudes of earthquakes to infer the statistical parameters related to earthquake frequency and 511
magnitude. The observed spatio-temporal distribution of b-value indicates that they were 512
spatially variable, but generally as low as ~0.7, and temporarily increased with time. The 513
observed p-values were different for the northeastern and southwestern parts of the study area, 514
implying that heterogeneities in material properties such as frictional heat can lead to two 515
different speeds of aftershock expansion rate with logarithmic time. The complexity of faulting 516
in the aftershock zone will influence the duration and magnitude of seismic activity that is 517
caused by the locally perturbed stress field that is a result of the mainshock. We hope that our 518
findings can be applied to an interpretation of aftershock mechanisms under the general complex 519
24
fault systems and can be utilized to perform a seismic hazard assessment lowering the epistemic 520
uncertainty about the characteristics of the fault sources and their contemporary seismic activity. 521
522
Data and Resources 523
524
The earthquake catalog used in this study is available at https://github.com/Jeong-525
Ung/PH_aftershock. 526
527
Acknowledgments 528
529
We thank the Korea Institute of Geoscience and Mineral resources (KIGAM), the Korea 530
Meteorological Administration (KMA), the Korea Hydro & Nuclear Power (KHNP), and the 531
K.‐ H. Kim for providing seismic data used in this study. We appreciate C. E. Yoon and G. C. 532
Beroza for comments on FAST usage, W. L. Ellsworth for advice on visualizing the seismicity, J. 533
Song for discussion on non-DC earthquakes. This work was conducted during the Korean 534
Government Commission (KGC) on the relations between the 2017 Pohang earthquake and EGS 535
project, funded by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) 536
grant from the South Korean government (MOTIE) (no. 2018‐ 3010111860). This work was 537
supported by the Nuclear Safety Research Program through the Korea Foundation of Nuclear 538
Safety (KoFONS) using the financial resource granted by the Nuclear Safety and Security 539
Commission (NSSC) of the Republic of Korea (No. 1705010). 540
541
References 542
25
543
Akaike, H. (1974). A new look at the statistical model identification, IEEE Trans. 544
Automat. Control 19, 716–723. 545
Aki K. (1965). Maximum likelihood estimate of b in the formula logN=a−bM and its 546
confidence limits, Bull. Earthq. Res. Inst. Univ. Tokyo 43, 237–239. 547
Ávila-Barrientos, L., Zúñiga, F. R., Rodríguez-Pérez, Q., and Guzmán-Speziale, M. 548
(2015). Variation of b and p values from aftershocks sequences along the Mexican subduction 549
zone and their relation to plate characteristics. J. S. Am. Earth Sci. 63, 162–171. 550
Baer, M., and Kradolfer, U. (1987). An automatic phase picker for local and teleseismic 551
events. Bull. Seismol. Soc. Am. 77, 1437–1445. 552
Bayrak, Y., Yadav, R. B. S., Kalafat, D., Tsapanos, T. M., Çınar, H., Singh, A. P., Bayrak 553
E., Yılmaz, Ş., Öcal, F, and Koravos, G. (2013). Seismogenesis and earthquake triggering during 554
the Van (Turkey) 2011 seismic sequence. Tectonophysics 601, 163–176. 555
Bergen, K. J., and Beroza, G. C. (2018). Detecting earthquakes over a seismic network 556
using single-station similarity measures. Geophys. J. Int. 213, 1984–1998. 557
Chan, C. H., Wu, Y. M., Tseng, T. L., Lin, T. L., and Chen, C. C. (2012). Spatial and 558
temporal evolution of b-values before large earthquakes in Taiwan. Tectonophysics 532, 215–222. 559
Choi, J. H., Ko, K., Gihm, Y. S., Cho, C. S., Lee, H., Song, S. G., Bang, E.-S., Lee, H.-J., 560
Bae, H.-K., Kim S. W., Choi S.-J., Lee, S. S., and Lee, S. R. (2019). Surface deformations and 561
rupture processes associated with the 2017 M w 5.4 Pohang, Korea, earthquake. Bull. Seismol. 562
Soc. Am. 109, 756–769. 563
Cornell, C. A. (1968). Engineering seismic risk analysis. Bull. Seismol. Soc. Am. 58, 564
1583–1606. 565
26
Dieterich, J. (1994). A constitutive law for rate of earthquake production and its 566
application to earthquake clustering. J. Geophys. Res.: Solid Earth 99, 2601–2618. 567
Ellsworth, W. L., and Bulut, F. (2018). Nucleation of the 1999 Izmit earthquake by a 568
triggered cascade of foreshocks. Nat. Geosci. 11, 531–535. 569
Ellsworth, W. L., Giardini, D., Townend, J., Ge, S., and Shimamoto, T. (2019). 570
Triggering of the Pohang, Korea, Earthquake (MW5.5) by Enhanced Geothermal System 571
Stimulation. Seismol. Res. Lett. 90, 1844–1858. 572
Farrell, J., Husen, S., and Smith, R. B. (2009). Earthquake swarm and b-value 573
characterization of the Yellowstone volcano-tectonic system. J. Volcanol. Geotherm. Res. 188, 574
260–276. 575
Frank, W. B., Poli, P., and Perfettini, H. (2017). Mapping the rheology of the Central 576
Chile subduction zone with aftershocks. Geophys. Res. Lett. 44, 5374–5382. 577
Fukuyama, E., Ellsworth, W. L., Waldhauser, F., and Kubo, A. (2003). Detailed fault 578
structure of the 2000 western Tottori, Japan, earthquake sequence. Bull. Seismol. Soc. Am. 93, 579
1468–1478. 580
Grigoli, F., Cesca, S., Rinaldi, A. P., Manconi, A., Lopez-Comino, J. A., Clinton, J. F., 581
Westaway, R., Cauzzi, C., Dahm, T., Wiemer, S. (2018). The November 2017 Mw 5.5 Pohang 582
earthquake: A possible case of induced seismicity in South Korea. Science 360, 1003–1006. 583
Gulia, L., Rinaldi, A. P., Tormann, T., Vannucci, G., Enescu, B., and Wiemer, S. (2018). 584
The Effect of a Mainshock on the Size Distribution of the Aftershocks. Geophys. Res. Lett. 45, 585
13–277. 586
Hainzl, S. (2016). Rate‐ dependent incompleteness of earthquake catalogs. Seismol. Res. 587
Lett. 87, 337-344. 588
27
Hainzl, S., Fischer, T., Čermáková, H., Bachura, M., and Vlček, J. (2016). Aftershocks 589
triggered by fluid intrusion: Evidence for the aftershock sequence occurred 2014 in West 590
Bohemia/Vogtland. J. Geophys. Res.: Solid Earth 121, 2575-2590. 591
Helmstetter, A., and Shaw, B. E. (2009). Afterslip and aftershocks in the rate‐ and‐ state 592
friction law. J. Geophys. Res.: Solid Earth 114, B01308 593
Hong, T. K., Lee, J., Park, S., and Kim, W. (2018). Time-advanced occurrence of 594
moderate-size earthquakes in a stable intraplate region after a megathrust earthquake and their 595
seismic properties. Sci. Rep. 8, 1–8. 596
Jolliffe I. T. (2002). Principal Component Analysis, Second Ed., Springer, New York. 597
Kato, A., and Obara, K. (2014). Step‐ like migration of early aftershocks following the 598
2007 Mw 6.7 Noto‐ Hanto earthquake, Japan. Geophys. Res. Lett. 41, 3864–3869. 599
Kilb, D., Gomberg, J., and Bodin, P. (2000). Triggering of earthquake aftershocks by 600
dynamic stresses. Nature 408, 570–574. 601
Kim, K. H., Ree, J. H., Kim, Y., Kim, S., Kang, S. Y., and Seo, W. (2018). Assessing 602
whether the 2017 MW 5.4 Pohang earthquake in South Korea was an induced event. Science 360, 603
1007–1009. 604
Kim, K. H., Seo, W., Han, J., Kwon, J., Kang, S. Y., Ree, J. H., Kim, S., and Liu, K. 605
(2019). The 2017 ML 5.4 Pohang earthquake sequence, Korea, recorded by a dense seismic 606
network. Tectonophysics 774, 228306. 607
King, G. C., Stein, R. S., and Lin, J. (1994). Static stress changes and the triggering of 608
earthquakes. Bull. Seismol. Soc. Am. 84, 935–953. 609
Kisslinger, C., and Jones, L. M. (1991). Properties of aftershock sequences in southern 610
California. J. Geophys. Res.: Solid Earth 96, 11947–11958. 611
28
Korea Meteorological Administration (2017). Annual report 2017. Seoul, Republic of 612
Korea, 35 pp. 613
Lahr J. C.1999. HYPOELLIPSE: A computer program for determining local earthquake 614
hypocentral parameters, magnitude and first‐ motion pattern (Y2K compliant version), version 615
1.0, U.S. Geol. Surv. Open‐ File Rept. 99‐ 23, On‐ Line Edition. 616
Langenbruch, C., Ellsworth, W. L., Woo, J. U., and Wald, D. J. (2020). Value at Induced 617
Risk: Injection‐ induced seismic risk from low‐ probability, high‐ impact events. Geophys. Res. 618
Lett. 47, c2019GL085878. 619
Lee, K.-K. (2019). Final Report of the Korean Government Commission on Relations 620
between the 2017 Pohang Earthquake and EGS Project. Geological Society of Korea. 621
https://doi.org/10.22719/KETEP-2019043001. 622
Lee, K.-K., Ellsworth, W. L., Giardini, D., Townend, J., Ge, S., Shimamoto, T., Yeo, I.-623
W., Kang, T.-S., Rhie, J., Sheen, D.-H., Chang, C., Woo, J.-U., Langenbruch, C. (2019). 624
Managing injection-induced seismic risks. Science 364, 730–732. 625
Mignan, A. (2015). Modeling aftershocks as a stretched exponential relaxation. Geophys. 626
Res. Lett. 42, 9726–9732. 627
Mikumo, T., and Miyatake, T. (1979). Earthquake sequences on a frictional fault model 628
with non-uniform strengths and relaxation times. Geophys. J. Int. 59, 497-522. 629
Miller, A. D., Foulger, G. R., and Julian, B. R. (1998). Non‐ double‐ couple earthquakes 630
2. Observations. Rev. Geophys. 36, 551–568. 631
Mogi, K. (1962). Study of the elastic shocks caused by the fracture of heterogeneous 632
materials and its relation to earthquakes phenomena. Bull. Earthq. Res. Inst. 40, 125–173. 633
29
Montési, L. G. (2004). Controls of shear zone rheology and tectonic loading on 634
postseismic creep J. Geophys. Res.: Solid Earth 109, B10404. 635
Murru, M., Console, R., Falcone, G., Montuori, C., and Sgroi, T. (2007). Spatial 636
mapping of the b value at Mount Etna, Italy, using earthquake data recorded from 1999 to 2005. 637
J. Geophys. Res.: Solid Earth 112, B12303 638
Narteau, C., Shebalin, P., and Holschneider, M. (2002). Temporal limits of the power law 639
aftershock decay rate. J. Geophys. Res.: Solid Earth 107, ESE–12. 640
Ogata, Y. (1983). Estimation of the parameters in the modified Omori formula for 641
aftershock frequencies by the maximum likelihood procedure. J. Phys. Earth 31, 115–124. 642
Ogata, Y. (1998). Space-time point-process models for earthquake occurrences. Ann. Inst. 643
Stati. Math. 50, 379–402. 644
Omori F. (1894). On after‐ shocks of earthquakes, J. Coll. Sci. Imp. Univ. Tokyo 7, 111–645
200. 646
Peng, Z., and Zhao, P. (2009). Migration of early aftershocks following the 2004 647
Parkfield earthquake. Nat. Geosci. 2, 877. 648
Perfettini, H., Frank, W. B., Marsan, D., and Bouchon, M. (2018). A model of aftershock 649
migration driven by afterslip. Geophys. Res. Lett. 45, 2283–2293. 650
Rhee, H. M., and Sheen, D. H. (2016). Lateral variation in the source parameters of 651
earthquakes in the Korean Peninsula. Bull. Seismol. Soc. Am. 106, 2266–2274. 652
Rigo, A., Souriau, A., and Sylvander, M. (2018). Spatial variations of b-value and crustal 653
stress in the Pyrenees. J. Seismol. 22, 337–352. 654
30
Rong, K., Yoon, C. E., Bergen, K. J., Elezabi, H., Bailis, P., Levis, P., and Beroza, G. C. 655
(2018). Locality-sensitive hashing for earthquake detection: a case study of scaling data-driven 656
science. Proceedings of the VLDB Endowment, 11, 1674-1687. 657
Ross, Z. E., Rollins, C., Cochran, E. S., Hauksson, E., Avouac, J. P., and Ben‐ Zion, Y. 658
(2017). Aftershocks driven by afterslip and fluid pressure sweeping through a fault‐ fracture 659
mesh. Geophys. Res. Lett. 44, 8260–8267. 660
Sabbione, J. I., and Velis, D. R. (2013). A robust method for microseismic event 661
detection based on automatic phase pickers. J. Appl. Geophys. 99, 42-50. 662
Scholz, C. H. (1968). The frequency-magnitude relation of microfracturing in rock and 663
its relation to earthquakes. Bull. Seismol. Soc. Am. 58, 399–415. 664
Scholz, C. H. (2015). On the stress dependence of the earthquake b value. Geophys. Res. 665
Lett. 42, 1399–1402. 666
Schorlemmer, D., Wiemer, S., and Wyss, M. (2005). Variations in earthquake-size 667
distribution across different stress regimes. Nature 437, 539–542. 668
Shapiro, S. A., Dinske, C., and Kummerow, J. (2007). Probability of a given‐ magnitude 669
earthquake induced by a fluid injection. Geophys. Res. Lett. 34, L22314. 670
Shelly, D. R., Hardebeck, J. L., Ellsworth, W. L., and Hill, D. P. (2016). A new strategy 671
for earthquake focal mechanisms using waveform‐ correlation‐ derived relative polarities and 672
cluster analysis: Application to the 2014 Long Valley Caldera earthquake swarm. J. Geophys. 673
Res.: Solid Earth 121, 8622–8641. 674
Shi, Y., and Bolt, B. A. (1982). The standard error of the magnitude-frequency b value. 675
Bull. Seismol. Soc. Am. 72, 1677–1687. 676
Smith, W. D. (1981). The b-value as an earthquake precursor. Nature 289, 136–139. 677
31
Soh, I., Chang, C., Lee, J., Hong, T. K., and Park, E. S. (2018). Tectonic stress 678
orientations and magnitudes, and friction of faults, deduced from earthquake focal mechanism 679
inversions over the Korean Peninsula. Geophys. J. Int. 213, 1360–1373. 680
Son, M., Cho, C. S., Shin, J. S., Rhee, H. M., and Sheen, D. H. (2018). Spatiotemporal 681
Distribution of Events during the First Three Months of the 2016 Gyeongju, Korea, Earthquake 682
Sequence. Bull. Seismol. Soc. Am. 108, 210–217. 683
Song, C. W., Son M., Sohn, Y. K., Han R., Shinn Y. J., and Kim J.-C. (2015). A study on 684
potential geologic facility sites for carbon dioxide storage in the Miocene Pohang Basin, SE 685
Korea. J. Geol. Soc. Korea 51, 53–66 (in Korean). 686
Song, S. G., and Lee, H. (2019). Static slip model of the 2017 Mw 5.4 Pohang, South 687
Korea, earthquake constrained by the InSAR data. Seismol. Res. Lett. 90, 140–148. 688
Tajima, F., and Kanamori, H. (1985). Global survey of aftershock area expansion 689
patterns. Phys. Earth Planet. Inter. 40, 77–134. 690
Trnkoczy A. (2002). Understanding and parameter setting of STA/LTA trigger algorithm, 691
in IASPEI New Manual of Seismological Observatory Practice (NMSOP), Bormann P. (Editor), 692
Vol. 2, GeoForschungsZentrum, Potsdam, Germany, 1–19. 693
Uchide, T., and Song, S. G. (2018). Fault rupture model of the 2016 Gyeongju, South 694
Korea, earthquake and its implication for the underground fault system. Geophys. Res. Lett. 45, 695
2257–2264. 696
Utsu, T. (1961). A statistical study on the occurrence of aftershocks, Geophys. Mag. 30, 697
521–605. 698
Utsu, T. (1992), On Seismicity, Report of the Joint Research Insititute for Statistical 699
Mathematics, Inst. Stat. Math., Tokyo 34, 139–157. 700
32
Utsu, T., Y. Ogata, and Matsu’ura, R. S. (1995). The centenary of the Omori formula for 701
a decay law of aftershock activity, J. Phys. Earth 43, 1–33. 702
Wadati, K., and Oki, S. (1933). On the travel time of earthquake waves, Part II, Geophys. 703
Mag. 7, 101–111. 704
Waldhauser, F., and Ellsworth, W. L. (2000). A double-difference earthquake location 705
algorithm: Method and application to the northern Hayward fault, California. Bull. Seismol. Soc. 706
Am. 90, 1353–1368. 707
Wiemer, S., and Katsumata, K. (1999). Spatial variability of seismicity parameters in 708
aftershock zones. J. Geophys. Res.: Solid Earth 104, 13135–13151. 709
Wiemer, S., and Wyss, M. (1997). Mapping the frequency‐ magnitude distribution in 710
asperities: An improved technique to calculate recurrence times?. J. Geophys. Res.: Solid 711
Earth 102, 15115–15128. 712
Wiemer, S., and Wyss, M. (2000). Minimum magnitude of completeness in earthquake 713
catalogs: Examples from Alaska, the western United States, and Japan. Bull. Seismol. Soc. 714
Am. 90, 859–869. 715
Withers, M., Aster, R., Young, C., Beiriger, J., Harris, M., Moore, S., and Trujillo, J. 716
(1998). A comparison of select trigger algorithms for automated global seismic phase and event 717
detection. Bull. Seismol. Soc. Am. 88, 95–106. 718
Woo, J. U., Kim, M., Sheen, D. H., Kang, T. S., Rhie, J., Grigoli, F., Ellsworth, W., L., 719
and Giardini, D. (2019a). An in‐ depth seismological analysis revealing a causal link between 720
the 2017 MW 5.5 Pohang earthquake and EGS project. J. Geophys. Res.: Solid Earth. 124, 721
13060–13078. 722
33
Woo, J. U., Rhie, J., Kim, S., Kang, T. S., Kim, K. H., and Kim, Y. (2019b). The 2016 723
Gyeongju earthquake sequence revisited: aftershock interactions within a complex fault system. 724
Geophys. J. Int., 217, 58–74. 725
Wyss, M. (1973). Towards a physical understanding of the earthquake frequency 726
distribution. Geophys. J. R. Astron. Soc., 31, 341–359. 727
Yoon, C. E., Bergen, K. J., Rong, K., Elezabi, H., Ellsworth, W. L., Beroza, G. C., Bailis 728
P. and Levis, P. (2019a). Unsupervised Large‐ Scale Search for Similar Earthquake Signals. Bull. 729
Seismol. Soc. Am. 109, 1451–1468. 730
Yoon, C. E., Huang, Y., Ellsworth, W. L., and Beroza, G. C. (2017). Seismicity during 731
the initial stages of the Guy‐ Greenbrier, Arkansas, earthquake sequence. J. Geophys. Res.: Solid 732
Earth 122, 9253–9274. 733
Yoon, C. E., O’Reilly, O., Bergen, K. J., and Beroza, G. C. (2015). Earthquake detection 734
through computationally efficient similarity search. Sci. Adv. 1, e1501057. 735
Yoon, C. E., Yoshimitsu, N., Ellsworth, W. L., and Beroza, G. C. (2019b). Foreshocks 736
and Mainshock Nucleation of the 1999 M w 7.1 Hector Mine, California, Earthquake. J. 737
Geophys. Res.: Solid Earth 124, 1569–1582. 738
Yun H., Min, K. D., Moon, H.-S., Lee H. K., and Yi, S. S. (1991). Biostratigraphic, 739
Chemostratigraphic, Paleomagnetostratigraphic, and Tertiary formations in southern part of 740
Korea: regional tectonics and its stratigraphical implication in the Pohang basin. J. Paleont. Soc. 741
Korea 7, 1–12 (in Korean). 742
Zoback, M. L. (1992). First‐ and second‐ order patterns of stress in the lithosphere: The 743
World Stress Map Project. J. Geophys. Res.: Solid Earth 97, 11703–11728. 744
745
34
Full mailing address for each author 746
1 School of Earth and Environmental Sciences, Seoul National University, Seoul 08826, 747
Republic of Korea 748
J.-U.W., [email protected] 749
J.R., [email protected] 750
2 Department of Structural Systems and Site Evaluation, Korea Institute of Nuclear Safety, 751
Daejeon 34412, South Korea 752
M.K., [email protected] 753
3 Division of Earth Environmental System Science, Pukyong National University, Busan 48513, 754
Republic of Korea 755
T.-S.K., [email protected] 756
757
758
35
Tables 759
760
Table 1. The 1-D layered seismic velocity structure for the Pohang EGS site. 761
Depth to the top of the layer (km) P-wave velocity (kms-1) S-wave velocity (kms-1)
0.0 1.67 0.48
0.203 4.01 2.21
0.67 5.08 3.03
2.4 5.45 3.07
3.4 5.85 3.31
7.7 5.91 3.51
12 6.44 3.70
34 8.05 4.60
762
763
36
Table 2. Parameters of the faults involved in the aftershock sequences. 764
Properties Fault 1 (F1) Fault 2 (F2) Fault 3 (F3) Fault 4 (F4)
Strike (°)
Median 222.7 207.4 223.1 17.0
Median absolute deviation 1.1 1.2 0.4 1.3
Dip (°)
Median 77.4 59.8 61.2 62.0
Median absolute deviation 2.0 1.3 0.6 2.2
Fault length (km) 2.8 2.4 3.4 1.9
Fault width (km) 1.9 3.5 2.9 1.3
765
37
Figures 766
767
768
Figure 1. Map of (a) temporary and (b) permanent seismic stations used for analysis of source 769
parameters, geologic lineaments, faults, and relocated hypocenters. Three surface ruptures near 770
the study area are illustrated in (a) (Song et al., 2015; Yun et al., 1991). The focal mechanism of 771
the mainshock that was determined from the polarity of first arrivals is illustrated in (b). (c) 772
shows the location of the Gyeongsang Basin (GB) and the Yeonil Basin (YB) where many 773
NENNE sinistral strike-slip surface ruptures and NW transfer faults have developed. The red 774
boxes in (b) and (c) represent the domain of (a) and (b), respectively. 775
38
776
Figure 2. (a) Distribution of the 3946 epicenters relocated via hypoDD (Waldhauser and 777
Ellsworth, 2000) by using traveltime differences. The earthquakes projected onto each of the 778
cross-sections A1-A2 to E1-E2 shown in (b) to (f) fall within the rectangles denoted by dashed 779
black lines in (a). The trajectory of two stimulation wells PX-1 and PX-2 are illustrated as gray 780
lines in (c) with open sections colored in blue and red. (b–f) Depth distribution of the relocated 781
39
hypocenters along the cross-sections of A1-A2 to E1-E2. Spatial distribution of seven focal 782
mechanism solutions are illustrated in (a) to (e). The compressional quadrants of the focal 783
mechanisms of the mainshock and two largest aftershocks (ML 4.3 and ML 4.6) are colored in red, 784
blue, and green, respectively. Possible interpretations for delineated faults from the aftershock 785
distribution are marked as gray lines in (b), (c), (d), and (e). The circles in (f) represent the 786
rupture radii of earthquakes with MRel > 1.5, assuming a stress drop of 5.6 MPa, which 787
corresponds to an approximated value for the mainshock estimated by the spectral ratio method 788
(Woo et al., 2019a). The red, blue, and green circles in (f) indicate the rupture size of the three 789
largest earthquakes with ML 5.4, 4.3, and 4.6, respectively. 790
791
40
792
Figure 3. (a) Determination of the scaling parameter c in Equation (1) from known ML 793
magnitudes. The amplitude ratios measured from two similar waveforms observed at a station 794
are measured and counted to estimate the scaling parameter for given MLs. The red line indicates 795
the scaling parameter c of 0.85 calculated from the slope of the first principal components 796
between magnitude differences and the ratio of amplitude divided by hypocenteral distances. (b) 797
Comparison between ML and MRel. The red line indicates identity relation. 798
799
41
800
Figure 4. (a) The distribution of earthquake magnitudes with their logarithmic origin time. The 801
three largest earthquakes (ML 5.4, 4.3, and 4.6) are denoted as red, blue, and green stars, 802
respectively. The time interval of each bin to measure the temporal changes of b-values and their 803
corresponding Mcs are illustrated as orange lines and squares. A set of 600 earthquakes 804
constitute a bin for measuring b-values and there is an overlap of 400 earthquakes between two 805
consecutive bins. We combined MRels obtained from Equation (1) with MLs of the three largest 806
earthquakes. (b) Temporal variations of seismic b-values. The standard deviations of each of the 807
magnitude bins are represented as vertical and horizontal error bars. The black, red, and blue dots 808
indicate the three bins illustrated in (c). The gray dots and error bars represent b-values and their 809
standard errors calculated based on a maximum MC of 0.8 for all bins. (c) Four examples of the 810
magnitude distribution and their corresponding G-R law fitting lines (black for the entire data set 811
42
and green, red, and blue for the three selected temporal bins of (b)). The filled circles indicate the 812
cutoff magnitudes of MC that honor Equation (1) for larger magnitudes. (d) Two dimensional 813
spatial variations of b-values at a vertical profile along both the apparent strike of 210°. 814
Hypocenters of three largest earthquakes are denoted as red, blue, and green stars, respectively. 815
Dashed lines and solid lines indicate the significant difference level of 5% with median and 816
conservative thresholds. (e) Two-σ interval distribution of probabilities that differences of paired 817
b-values (Δb) in (c) is insignificant. The Δbs are binned by 0.1. The difference is statistically 818
significant with a significance level of 5% if Δb > 0.14 for half the cases and Δb > 0.18 for all 819
cases. 820
43
821
Figure 5. (a) Two dimensional spatial variations of p-values for a cross-section along the 822
apparent strike 210° for earthquake sequences in period A. (b) Two dimensional spatial 823
variations of p-values for the same depth profile shown in (a), but for seismic sequences in 824
period B. The p-values of period B were estimated using the given aftershock rates of period A 825
when they were available (denoted by the two spatial bins contoured by black lines). (c and d) 826
Cumulative number of aftershocks with time and their corresponding Omori law plots for the 827
case of whole data sets (RA(t) and RB(t)) and the three examples of the spatial bins contoured by 828
blue, green, and red lines in (a). 829
830
44
831
Figure 6. Temporal distribution of seismicity presented in (a) plan view, (b) a depth profile along 832
the apparent strike of 210°, and (c) a unidirectional projection along the apparent strike. We 833
illustrate the radius of earthquakes with MRel ≥ 1.5 by assuming a circular crack rupture and a 834
stress drop of 5.6 MPa from Woo et al. (2019a). In (a) and (b), the location of the mainshock and 835
the two largest consecutive aftershocks of ML 4.6 and ML 5.4 are represented as red, blue and 836
green stars, respectively. The rupture radii of the three largest earthquakes are displayed in (c) 837
with colors to match the star symbols in (a) and (b). The trajectory of two stimulation wells PX-1 838
and PX-2 are illustrated as gray lines in (a) and (b). In (c), the earthquake density of each point 839
was measured as the number of earthquakes within a circle of 0.25-unit radius (showing decades 840
along the x-axis and km along the y-axis) from the given point. The x symbol in (c) represents 841
45
the moment at which the number of earthquakes in the 0.5-km bin reached 10 with a 0.1-km 842
sliding window. 843
46
844
Figure 7. Schematic diagram that illustrates four fault segments inferred from the hydraulic 845
stimulation wells of PX-1 and PX-2 and the distribution of aftershocks. The three cubes colored 846
in red, blue, and green show the hypocenters of the mainshock and the two largest aftershocks of 847
ML 4.3 and 4.6, respectively. The occurrence of the mainshock triggered seismicity on fault 848
segments F2 and F3, and possibly affected the re-activation of F1. Fault segment F4, located to 849
the southwest of F3, was not delineated until the largest aftershock (ML 4.6) occurred. 850
851
852
47
Appendices 853
854
Table A1. Fingerprint extraction parameters for the FAST algorithm to detect earthquakes with 855
waveform similarity. 856
Fingerprint extraction parameter Value
Time-series window length of generated spectrogram 6.0 s
Time-series window lag of generated spectrogram 0.1 s
Spectral image window length 64
Spectral image window lag 10
Fingerprint sparsity 400
Final spectral image width 32
Number of hash functions per hash table 4
Number of hash tables 100
Number of votes 2
Near-repeat exclusion parameters 5
857
Table A2. Input parameters for the network detection in the FAST algorithm (Bergen and Beroza, 858
2018; Rong et al., 2018) 859
Event-pair extraction, pruning, and network detection parameters values
Time gap along diagonal 3 s
Time gap adjacent diagonal 3 s
Adjacent diagonal merge iteration 2
48
Number of votes 10
Minimum fingerprint pairs 3
Maximum bounding-box width 5 s
Minimum number of stations for detection 1
Arrival time constraint: maximum time gap 5
860
Table A3. Focal mechanisms illustrated in Figure 2. 861
Origin time
(UTC, dd/mm/yy
HH:MM:SS.SS)
Latitude
(º)
Longitude
(º)
Depth
(km)
Strike
(º)
Dip
(º)
Rake
(º)
15/11/2017 05:29:31.61 36.10592 129.37215 4.245 211 40 128
15/11/2017 06:09:49.88 36.08742 129.34946 4.318 91 74 -132
15/11/2017 07:49:30.37 36.11412 129.36825 5.450 201 65 110
19/11/2017 14:45:47.79 36.11303 129.38023 3.865 34 85 149
19/11/2017 21:05:15.48 36.13384 129.37956 3.851 234 85 -174
25/12/2017 07:19:22.58 36.10206 129.36582 5.291 39 81 165
10/02/2018 20:03:03.74 36.07862 129.34237 4.447 34 52 136
862
49
Figure A1. Snapshots of the distribution maps of aftershocks at (a) 10 min, (b) 1 h, (c) 5 h, (d) 1 863
d, (e) 30 d, and (f) 4 mo from the onset of the mainshock. 864
865
129.2˚E 129.3˚E 129.4˚E 129.5˚E
36˚N
36.1˚N
36.2˚N
0 km 5 km
23456
Depth (km)
PHB1
PHB2
PHB3
PHB4
PHB5
PHB6
PHB7
PHB8
EXP1
POHB
POH1
POH2 POH3
POH4
POH5
POH7
PH01
PH03
PH04PH05
PH06
PH07
PH08PH09
KN01 50KP11NKKN02SS06
KN03
KN04SS08
PK01 PK11
PK03PK08
SS02
SS03 PK04
SS04
SS07
SS09
SS10
SS13
SS15
SS23
TP02
TP03
TP04
TP06
TP07
TP08
TP09
TP10
TP11
TP13
TP14
50 km
CHS
DKJHAK
HDBMKL
YSB
GKP1
WAG
WBG
WCGWDG
ADO2 CSO
PHA2
USN
YOCB
YODB
CIGB
DAG2
EUSB
ULJ2YEYB
MIYA
100 km
GB YB
Yang
san
faul
t
Gokgang Fault
Heunghae Fault
Pohang city
KSKGKN
Aftershock monitoring (SNU&PKNU)Kim et al. (2018)EGS monitoring Aftershock monitoring (KIGAM)Permanent
(a) (b)
(c)
Figure 1 Click here to access/download;Figure;Figure 1.pdf
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚2 km
23456
Depth (km)
A2
A1
B2
B1
C2
C1
D2
D1
E2
E1 2
3
4
5
6
7
Dep
th (
km)
0 1 2A1-A2 (km)
9090
2
3
4
5
6
7
Dep
th (
km)
0 1 2 3 4B1-B2 (km)
5050
2
3
4
5
6
7
Dep
th (
km)
0 1 2 3C1-C2 (km)
57
2
3
4
5
6
7
Dep
th (
km)
0 1 2D1-D2 (km)
60
2
3
4
5
6
7
Dep
th (
km)
0 1 2 3 4 5 6 7 8 9 10 11E1-E2 (km)
Heunghae fault50º70º90º
50º70º90º
50º70º90º
(a) (b) (c)
(d) (e)
(f)
Figure 2 Click here to access/download;Figure;Figure 2.pdf
−4
−3
−2
−1
0
1
2
3
4M
agni
tude
diff
eren
ce
−4 −3 −2 −1 0 1 2 3 4(Amplitude ratio)/(Hypocenteral Distance ratio)
0
500
1000
Fre
quen
cy
0
1
2
3
4
Rel
ativ
e M
agni
tude
(M
RE
L)
0 1 2 3 4Local Magnitude (ML)
(a) (b)Figure 3 Click here to access/download;Figure;Figure 3.pdf
0
1
2
3
4
5
Mag
nitu
de
0.001 0.01 0.1 1 10 100Time after E1(day)
0.6
0.7
0.8
0.9
1.0
1.1
1.2
b−va
lue
0.1 1 10 100Time after E1(day)
2
3
4
5
6
7
Dep
th (
km)
−5 −4 −3 −2 −1 0 1 2 3 4 5Distance along strike (km)
0
1
2
3
4
log 1
0(N
)
0 1 2 3 4 5Magnitude
0.65
0.70
0.75
0.80
0.85
0.90
b−value
0.0
0.1
0.2
0.3
0.4
Pro
babi
lity
0.0 0.1 0.2 0.3b−value difference
R1
R2
R3
(a) (b) (c)
(d) (e)
log10(N)=3.08-0.66M
log10(N)=2.74-0.77Mlog10(N)=2.85-0.98M
log10(N)=3.59-0.73M
significant level0.05
Figure 4 Click here to access/download;Figure;Figure 4.pdf
2
3
4
5
6
7
Dep
th (
km)
−5 −4 −3 −2 −1 0 1 2 3 4 5Distance along strike (km)
2
3
4
5
6
7
Dep
th (
km)
2 3 4 5Distance along strike (km)
1
10
100
1000
Cum
mul
ativ
e nu
mbe
r of
afte
rsho
cks
0.001 0.01 0.1 1 10 100Time after E1 (day)
1
10
100C
umm
ulat
ive
num
ber
of a
fters
hock
s
0.001 0.01 0.1 1 10Time after E3(day)
0.9
1.0
1.1
1.2
1.3
1.4p−value(a) (b)
(c) (d)
RB(t)=RA’(t+tdiff)+45/(0.018+t)0.88RA(t)=398/(0.225+t)1.10
R1(t)=32/(0.10+t)0.90
R3(t)=79/(0.43+t)1.35R2(t)=63/(0.15+t)1.10
Figure 5 Click here to access/download;Figure;Figure 5.pdf
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
E2
E1
3
6
Dep
th (
km)
0 3 6 9E1-E2 (km)
-2 -1 0 1 2
log10(Days after mainshock)
0
1
2
3
4
5
6
7
8
9
10
11
E1-
E2(
km)
-3 -2 -1 0 1 2log10(Days after mainshock)
0 1 2
log10(Earthquake density)
(a)
(b)
(c)2km decade-1
1km decade-1
0.5km decade-1
2km decade-11km decade-1
0.5km decade-1
Figure 6 Click here to access/download;Figure;Figure 6.pdf
Heung
hae
faul
t
F4
F1
F2
F3
Figure 7 Click here to access/download;Figure;Figure 7.pdf
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
129.325˚ 129.35˚ 129.375˚ 129.4˚
36.075˚
36.1˚
36.125˚
36.15˚
2 km
Heunghae fault Heunghae fault
Heunghae fault Heunghae fault
Heunghae faultHeunghae fault
(a) (b)
(c)
(e)
(d)
(f)
0~10 min10~60 min
0~1 mo1~4 mo
0~1 hr1~5 hr
0~1 day1~30 day
0~5 hr5~24 hr
During EGS0~10 min
PX-1PX-2
Figure A1 Click here to access/download;Figure;FIgure A1.pdf
0
1
2
3
4lo
g 10(
N)
0 1 2 3 4 5Magnitude
0.50.60.70.80.91.01.11.21.31.41.5
b−va
lue
−0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0Magnitude of completeness
0.5
0.6
0.7
0.8
0.9
1.0
Goo
dnes
s of
fit
0
1
2
3
4
log 1
0(N
)
0 1 2 3 4 5Magnitude
0.50.60.70.80.91.01.11.21.31.41.5
b−va
lue
−0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0Magnitude of completeness
0.5
0.6
0.7
0.8
0.9
1.0
Goo
dnes
s of
fit
0
1
2
3
4
log 1
0(N
)
0 1 2 3 4 5Magnitude
0.50.60.70.80.91.01.11.21.31.41.5
b−va
lue
−0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0Magnitude of completeness
0.5
0.6
0.7
0.8
0.9
1.0
Goo
dnes
s of
fit
0
1
2
3
4
log 1
0(N
)
0 1 2 3 4 5Magnitude
0.50.60.70.80.91.01.11.21.31.41.5
b−va
lue
−0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0Magnitude of completeness
0.5
0.6
0.7
0.8
0.9
1.0
Goo
dnes
s of
fit
(a) First temporal bin
(b) Second temporal bin
(c) Sixth temporal bin
(d) Last temporal bin
Figure R1 Click here to access/download;Figure;Figure R1.pdf
0
1
2
3
4
5
6
7
8
9
10
11
E1-
E2(
km)
0 1 2 3 4 5 6 7 8 9 10log10(Days after mainshock)
0 1 2log10(Earthquake density)
Figure R2Click here to access/download;Figure;Figure R2.pdf
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3b−value
0.11
10100
Tim
e after E1(day)
−0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
MC
Figure R3Click here to access/download;Figure;Figure R3.pdf