Geothermics 71 (2018) 225–233

Contents lists available at ScienceDirect

Geothermics
journal homepage: www.elsevier.com/locate/geothermics

Seismic velocity structure and characteristics of induced seismicity at the MARK

Geysers Geothermal Field, eastern California
⁎
Guoqing Lina, , Bateer Wub
a
Department of Marine Geosciences, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149, USA
b
Institute of Disaster Prevention, Xueyuan Road, Village Yanjiao, Sanhe, Hebei Province 065201, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: We present a newly developed three-dimensional seismic velocity model, high-precision earthquake relocations
The Geysers and focal mechanisms near the Geysers Geothermal Field in eastern California using seismic data recorded by the
Induced seismicity Northern California Earthquake Data Center from 1984 to 2015. The velocity model generally agrees with those
Crustal structure in previous studies with the Vp model mainly corresponding to rock composition and the Vp/Vs model more
correlated with fluid content. The dominating low Vp/Vs anomalies observed from 0 to 4 km depth below sea
level is a reflection of the geothermal reservoir. The waveform cross-correlation relocated seismicity shows both
spatial and temporal correlations with the geothermal operations in the study area, indicating that they are
induced seismicity. Although the focal solutions are dominated by normal and strike-slip regimes throughout our
entire study time period, there has been an increase in reverse faulting since 2008, which may be caused by the
thermal contraction associated with the Northwest Geysers Enhanced Geothermal System project between 2008
and 2012. Our study provides a groundwork for future seismological studies in The Geysers.

1. Introduction
The Geysers Geothermal Field (GGF) in eastern California is situated
approximately 110 km north of San Francisco and is the largest geo-
thermal field in the world. Although intensively exploited, it still pro-
vides about 6% of California's electrical power. The Geysers is also one
of the most seismically active regions in California and is dominated by
shallow induced seismicity. It is located on the eastern side of the
Maacama Fault Zone and the Healdsburg Fault, bounded by the right-
lateral strike-slip Mercuryville Fault and Collayomi Fault, and sur-
rounded by several Mountains, such as the Geyser Peak, the Cobb Mtn.,
the Boggs Mtn., the Seigler Mtn., and the Mt. Hannah (Fig. 1). Since the
geothermal operations commenced in the 1960s, numerous studies
have been conducted to investigate their effects on hydrologic and
tectonic features in the area. The tectonic setting and structure in The
Geysers have been reviewed by previous researchers (e.g., McLaughlin,
1981; Thompson, 1989, 1992). The spatial/temporal variations in
seismic velocities have been intensively investigated (e.g., Majer and
Mcevilly, 1979; O’Connell and Johnson, 1991; Zucca et al., 1994;
Romero et al., 1995; Alwyn, 1996; Julian et al., 1996; Kirkpatrick et al.,
1997; Boitnott and Kirkpatrick, 1997; Foulger et al., 1997; Gunasekera
et al., 2003; Gritto et al., 2013; Gritto and Jarpe, 2014). The majority of
these studies show that the shallow velocity structure correlates with
the surface geology and the geologic and hydrologic features of the
reservoir. Although some studies found no consistent correlation be-
tween injection and seismicity, it has been widely accepted that there is
a correlation between the two parameters (Majer and Mcevilly, 1979;
Denlinger and Bufe, 1980; Bufe and Shearer, 1981; Allis, 1982;
Eberhart-Phillips and Oppenheimer, 1984; O’Connell and Johnson,
1991; Stark, 1992; Romero et al., 1995; Alwyn, 1996; Trugman et al.,
2016). The stress state study by Boyle and Zoback (2014) revealed a
normal/strike-slip faulting regime within and below the reservoir and
also suggested that the geothermal operations have no significant effect
on the local stress field.
In this paper, we aim to investigate the structural heterogeneities
and examine the spatial and temporal variations and characteristics of
induced seismicity in The Geysers Geothermal Field by using a newly
developed high-resolution three-dimensional (3-D) Vp and Vp/Vs model,
precise earthquake relocations, and focal mechanisms. This study is also
motivated by a recent work on remotely triggered earthquakes by
Zhang et al. (2017). They found that although the adjacent fault areas
are responsive, the two hot water systems themselves, the Coso and the
Salton Sea Geothermal Fields, are insusceptible to large distant earth-
quakes. The Geysers Geothermal Field is an under-pressurized, vapor-
dominated field, which may have different response to remote earth-
quake triggering. Our results in this study will provide essential

⁎
Corresponding author.
E-mail address: glin@rsmas.miami.edu (G. Lin).

http://dx.doi.org/10.1016/j.geothermics.2017.10.003
Received 23 March 2017; Received in revised form 29 September 2017; Accepted 8 October 2017
Available online 17 October 2017
0375-6505/ © 2017 Elsevier Ltd. All rights reserved.
G. Lin, B. Wu
Fig. 1. Major geological structures in the study area, including the surface fault lines
(black lines) and several mountain ranges (pink squares). Gray dots represent ∼246,000
earthquakes between 1984 and 2015 recorded by the NCSN (black triangles). Blue and
white triangles stand for the EGS and the temporary stations, respectively. The back-
ground is the topography base map from the U. S. Geological Survey. The red box in the
inset map shows the location of our study area in California. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of the
article.)
groundwork for the future fine-scale remote triggering analyses and
other seismological studies in The Geysers.
2. Data
We obtained all the seismic data used in this study from the
Northern California Earthquake Data Center (NCEDC, 2014). These
data are mainly recorded by the Northern California Seismic Network
(NCSN) stations (black triangles in Fig. 1), consisting of ∼246,000 local
earthquakes (gray dots in Fig. 1) from 1984 to 2015 with 3.7 million
compressional (P) and 0.18 million shear (S) wave first-arrivals and all
their waveform data. The majority of these events are confined within
the steam-production zone of the Geysers Geothermal Field, bounded
by the Mercuryville Fault to the southwestern side and the Collayomi
Fault to the northeastern side.
For the tomographic inversion, due to the small number of S-picks
available in the NCSN catalog, we also take advantage of the seismic
record from the Enhanced Geothermal Systems (EGS) stations (blue
triangles in Fig. 1), operated by the US Department of Energy Geo-
thermal Technologies Program. The EGS data include ∼444,000
earthquakes since April 2003 with over 5 million P- and 2 million S-
picks. Note that the EGS catalog events are only used in the velocity
tomography, not in the relocation or focal mechanism inversions.
For the selection of the master events used in the tomographic in-
version, we apply different requirements for the NCSN and EGS net-
work catalogs. For the EGS catalog events, we require more than 8 P
picks and 4 S-P times per event. The resulting 1937 events (gray dots in
Fig. 2) consist of 26,975 P and 12,243 S picks. Due to the very few
number of S-picks in the NCSN catalog, we require at least 8 P picks and
no constraints on the number of S picks for each event. The 2038 events
(red dots in Fig. 2) comprise 34,617 P and 333 S picks. The inclusion of
the events outside of the GGF are helpful to improve deep velocity
structure resolution. These total 3975 events and their corresponding
picks are used in our final velocity model inversion. P-wave arrival
times for 3 active-source data (big blue dots in Fig. 2), recorded by
226
Geothermics 71 (2018) 225–233
Fig. 2. Events (dots) and inversion grid nodes (green squares) used in the tomographic
inversion. Gray dots stand for earthquakes that are only recorded by the EGS stations, red
ones for those recorded by the NCSN stations, and two big blue ones for the three ex-
plosion events (with two at the same location). The Cartesian coordinate is rotated 45°
counterclockwise with respect to latitude and longitude, with the X axis pointing to the
southwest and Y axis to the northwest direction. The origin is located at (38.8°, −122.8°)
(shown by the yellow star). Dark blue squares denote the same mountains in Fig. 1. The
two yellow straight lines are the profiles for the cross-sectional views in the following
figures. The inset figure shows the 1-D initial Vp model for the tomographic inversion,
which is used for the NCSN daily operation in the Geysers area. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of the
article.)
temporary seismic stations (white triangles in Fig. 1) and collected by
previous investigators (Meador et al., 1985), were also included in the
tomographic inversion to constrain the shallow crustal structure and
absolute hypocenter locations.
3. Results
3.1. 3-D seismic velocity model
The simul2000 algorithm is one of the most widely applied tomo-
graphic programs, which solves for 3-D Vp, Vp/Vs models and earth-
quake locations simultaneously (Thurber, 1983, 1993; Eberhart-
Phillips, 1990; Thurber and Eberhart-Phillips, 1999). It combines
parameter separation and damped least squares for model perturba-
tions. The original simul2000 algorithm allows a constant minimum
earthquake depth, i.e., none of the earthquake locations can be shal-
lower than the given minimum depth. In this study, we apply a mod-
ified version of the simul2000 that replaces the constant minimum
depth with the specific topographic information to avoid artificial ef-
fects (Lin, 2015).
The horizontal grid spacing in our 3-D velocity model is 2 km (ro-
tated squares in Fig. 2). The vertical node intervals range between 1 and
4 km from −1 to 20 km depth. Zero depth corresponds to mean sea
level. Note that in this paper all depths are relative to mean sea level,
unless otherwise stated. A one-dimensional (1-D) Vp model (inset in
Fig. 2), which is used for the NCSN daily operation in the Geysers area,
is applied as the initial model for the 3-D tomographic inversion. The
starting value for the Vp/Vs inversion is a constant ratio of 1.732.
In order to investigate the model recovery, we performed a check-
erboard resolution test similar to previous studies (e.g., Thurber et al.,
2009; Lin et al., 2010, 2014; Lin, 2013, 2015). We created a synthetic
velocity model based on the 1-D starting model with ± 5% Vp and ∓5%
Vp/Vs perturbations (i.e., opposite signs) that alternate at different
depths and across two grid nodes. Synthetic times were computed
G. Lin, B. Wu Geothermics 71 (2018) 225–233

Fig. 3. Checkerboard resolution test for the Vp model. The synthetic times are computed from the 1-D starting velocity model with ± 5% velocity anomalies across two grid nodes.
(a1–a4) Map views of the synthetic model. (b1–b4) Map views of the inverted model. (c1–c2) Cross sections of the synthetic model along profiles 1-1′ and A-A′ in (a1) and Fig. 2. (d1–d2)
Cross sections of the inverted model along the same profiles. The green contours enclose the well-resolved area with the diagonal element of the resolution matrix greater than 0.3. Dotted
curves at top of cross sections illustrate the local topography. MF is short for the Mercuryville Fault and CF for the Collayomi Fault.

through this model for the same event hypocenters, station locations,
and ray geometry as those in the real data. Fig. 3 shows map view and
cross-sectional comparisons between the synthetic and inverted Vp
models in the test. The green contours enclose the well-resolved area
with the diagonal element of the resolution matrix greater than 0.3,
where 1.0 represents the best resolution and 0.0 not resolved at all. At
−1 km depth (i.e., 1 km above sea level), the Vp model is poorly re-
solved, with only the central part of the steam-production zone re-
covered, due to the lack of seismicity at this shallow depth. At sea level,
the model resolution is significantly improved within the geothermal
field. The next two layers are well recovered both within the geo-
thermal field and its surrounding area, owing to the inclusion of both
the NCSN and EGS catalog data. The Vp model along the two cross-
section profiles is almost fully-recovered with slight smearing near the
edge. The Vp/Vs model (Fig. 4) is not resolved as well as the Vp model,
with only the structure within the geothermal field recovered because
the S arrivals are mostly from the EGS stations, which are located
within the geothermal field.
The tomographic inversion is convergent after 7 iterations. For the
master events, the root-mean-square (RMS) of the arrival-time residuals
is reduced from 0.78 to 0.29 s for P arrivals and from 0.47 to 0.37 s for S
arrivals after the inversion. For the explosion data with only P arrivals,
the RMS drops from 0.34 to 0.19 s, indicating that the shallow crustal
structure near the geothermal field, where the explosion data are
available, has been greatly improved.
Fig. 5 shows the map views and cross-sections of our final P-wave
velocity model. The white contours enclose the well-resolved regions
where the diagonal element of the resolution matrix is greater than 0.3.
At −1 and 0 km depths, the geothermal field is dominated by relatively
high velocity anomalies than the surrounding area. In contrast, the two
deeper layers (2 and 4 km depths) show low velocity structures within
the geothermal field. The final Vp/Vs model is shown in Fig. 6. The first
layer has limited resolution due to the lack of seismicity and displays
high Vp/Vs values up to 1.82. The GGF in the next two layers shows low
Vp/Vs ratios with relatively high values near the boundary. The model is
only resolved in the central area of the steam-production zone at 4 km

227
G. Lin, B. Wu
Fig. 4. Checkerboard resolution test for the Vp/Vs model. The synthetic times are computed from the constant Vp/Vs with ± 5% anomalies across two grid nodes. (a1–a4) Map views of the
synthetic model. (b1–b4) Map views of the inverted model. (c1–c2) Cross sections of the synthetic model along the profiles shown in (a1). (d1–d2) Cross sections of the inverted model
along the same profiles. Symbols are the same as those in Fig. 3.
depth.
3.2. Earthquake relocation
The final velocity model is used to relocate all the ∼246,000 NCSN
recorded earthquakes in the study area. The simul2000 algorithm es-
timates the absolute location uncertainties in the horizontal and ver-
tical directions. For all the 3-D relocations, the median of the location
uncertainties is 98 m for horizontal and 137 m for vertical.
Relative location accuracy of the 3-D relocated earthquakes is im-
proved by applying similar-event cluster analysis and differential-time
location methods based on waveform cross-correlation data, as pre-
sented in Lin et al. (2007). The cluster analysis approach is to identify
groups of earthquakes that are correlated with one other. Given our
criteria for the waveform cross-correlation processing, which are the
same as those used in a previous study (Lin, 2015), over 64% of all the
3-D relocated earthquakes fall within similar-event clusters. The
147,663 earthquakes in 1588 similar-event clusters were then relocated
using the waveform cross-correlation times. The relative earthquake
228
Geothermics 71 (2018) 225–233
location uncertainties estimated by a bootstrap approach (Efron and
Gong, 1983; Efron and Tibshirani, 1991) have a median of 7 m for
horizontal and 10 m for vertical.
Fig. 7 shows the location comparison between the initial NCSN
catalog (a–c) and the waveform cross-correlation relocation of this
study (d–f). Note that the NCSN catalog is based on 1-D velocity models,
therefore zero depth refers to the average station elevation used in the
location process (about 800 m above sea level) instead of mean sea
level in this study. The difference in absolute locations between the two
catalogs displays the importance of 3-D velocity structure in earthquake
relocation studies. The use of waveform cross-correlation data in dif-
ferential time relocation process enables the dramatic sharpening in
seismicity distribution.
3.3. Earthquake focal mechanism
Based on the waveform cross-correlation relocations, we compute
focal mechanisms from P-wave first motion polarity observations by
applying the HASH program (Hardebeck and Shearer, 2002), assuming
G. Lin, B. Wu Geothermics 71 (2018) 225–233

Fig. 5. Map views and cross sections of the final P-wave velocity model. (a–d) Map views of the P-wave velocity model at each layer depth where the inversion nodes are located. The
white contours enclose the well-resolved area with the diagonal element of the resolution matrix greater than 0.3. Black lines denote the surface traces of mapped faults. Red straight lines
in (d) are the profiles for the cross sectional views. Inset in (a) shows the Vp versus Vs (derived from dividing Vp by Vp/Vs) values at the well-resolved inversion nodes. (e–f) Cross sections
of the Vp model along the two profiles in (d), including the waveform cross-correlation relocated seismicity (gray dots) within ± 0.5 km distance of the profile lines. Dotted curves at top
of cross sections illustrate the local topography. Abbreviations are MF for the Mercuryville Fault and CF for the Collayomi Fault. (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of the article.)

double-couple fault plane solutions. Over 2.2 million first-motion po-
larity records are obtained from the vertical component of each seismic
station and were used in the computation. We assign four types of
qualities (A–D) to the solved 98,453 focal mechanisms based on the
criteria in the HASH algorithm and consider the quality A and B me-
chanisms as good-quality solutions, which are ∼15% of the total me-
chanisms. We then compute the scalar faulting type ([−1, −0.33) for
normal, [−0.33, 0.33] for strike-slip, and (0.33, 1] for reverse), for
each solution from the rakes of the two nodal planes (Shearer et al.,
2006; Lin and Shearer, 2009; Lin and Okubo, 2016). The mechanisms in
our study area are dominated by normal faulting (51%) and strike-slip
(38%) and much fewer reverse (11%) focal types. Fig. 8 shows map
view and depth distribution of good quality focal mechanisms plotted at
their cross-correlation relocations.

Fig. 6. Map views and cross sections of the final Vp/Vs model. (a–d) Map views of the Vp/Vs model at each layer depth where the inversion nodes are located. The white contours enclose
the well-resolved area with the diagonal element of the resolution matrix greater than 0.3. Black lines denote the surface traces of mapped faults. Red straight lines in (d) are the profiles
for the cross sectional views. (e–f) Cross sections of the Vp/Vs model along the two profiles in (d), including the waveform cross-correlation relocated seismicity (gray dots) within ±
0.5 km distance of the profile lines. Dotted curves at top of cross sections illustrate the local topography. Abbreviations are MF for the Mercuryville Fault and CF for the Collayomi Fault.
(For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

229
G. Lin, B. Wu Geothermics 71 (2018) 225–233

Fig. 7. Earthquake location comparison between the initial NCSN catalog location (a–c) and the waveform cross-correlation relocation in this study (d–f). Events in (a) and (d) are colored
by depth. Red triangles in (d) are the locations of the operational wells obtained from the California Oil, Gas, and Geothermal Resources. The depth distributions show seismicity
within ± 0.5 km distance of the two profile lines (straight pink lines in a and d). Orange dotted curves at top of the cross-sections illustrate the local topography. MF is short for the
Mercuryville Fault and CF for the Collayomi Fault. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

4. Discussion
In this study, we perform a series of seismic analyses in order to
investigate the effect of the geothermal operational activities on the
geological and seismological features in the Geysers area. We first de-
velop a fine-scale 3-D seismic velocity model using both P and S wave
arrivals obtained from the NCEDC. The major contribution of this work
is a high-precision earthquake catalog with over 147,000 events re-
located based on the 3-D velocity model and waveform cross-correla-
tion data and 67% of them have focal mechanisms derived from P wave
first motion polarity observations. In addition to the enormous volume
of seismic data (both inside and outside of the GGF) recorded by the
NCSN stations, the EGS data and the controlled sources used in the
tomographic inversion greatly helped to improve the resolution of the
Vp/Vs model and the shallow Vp structure, therefore the overall ro-
bustness of our results.
The seismic velocity structure in The Geysers has been investigated
by numerous studies. The 1-D initial velocity model used in this study is
similar to the layered model by Eberhart-Phillips and Oppenheimer
(1984) that shows Vp varying from 4.4 km/s at the surface to 5.9 km/s
at 8 km and deeper. Some teleseismic and local studies reveal low P-
wave velocity anomalies under the steam-production zone from the
crust to the upper mantle (e.g., Iyer et al., 1981; Oppenheimer and
Herkenhoff, 1981; Zucca et al., 1994). However, some studies found
that P-wave velocity in The Geysers production zone is faster than the
surrounding area (e.g., Eberhart-Phillips, 1986; Majer and Mcevilly,
1979). The Vp model in our study is well resolved from the surface to
4 km depth due to the dominating shallow earthquakes used in the
tomographic inversion. According to the laboratory measurements for
The Geysers core material, Vp and Vs vary in opposite ways with degree
of saturation (Kirkpatrick et al., 1997). However, the well-resolved Vp
and Vs values in our study show a positive correlation (inset in Fig. 5a),
indicating that they do not significantly depend on the saturation. In-
stead, our Vp model greatly agrees with the geological structure in the
area (McLaughlin, 1981), suggesting that lithology plays a more im-
portant role in determining Vp value in our study area. Our velocity
model is most consistent with the one by Romero et al. (1995) for the
Northwest Geysers region, whose surface geology reflects the general
geology of our study region. We interpret the high velocity anomalies at
shallow depth (1 km above and at sea level) as a reflection of the

230
G. Lin, B. Wu
Fig. 8. Map view (a) and cross-sections (b and c) of good quality (i.e., quality A and B) focal mechanisms based on the waveform cross-correlation relocation, colored by their focal type
with red for normal, green for strike-slip, and blue for reverse. The depth profiles for (b) and (c) are shown by the yellow straight lines in Fig. 2, same as those for the velocity model and
earthquake relocation presentations. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
Franciscan graywacke and greenstone and the low-velocity anomalies
at 2 km depth as a correlation with Franciscan melange.
In contrast to the Vp model, the Vp/Vs model is more sensitive to
temperature and fluid content than Vp and Vs alone. The common fea-
ture of the existing Vp/Vs models (e.g., Julian et al., 1996; Foulger et al.,
1997; Kirkpatrick et al., 1997; Gritto et al., 2013; Gritto and Jarpe,
2014) is the dominating low ratio (∼1.58), corresponding to the
exploited section of the geothermal reservoir, which is usually ex-
plained by low pore pressure and dry conditions at depth. The low Vp/
Vs anomalies observed in our model between 0 and 4 km depth below
sea level (bsl) is consistent with these previous models. The high Vp/Vs
ratios at shallow depth may suggest a recharge region or a condensation
zone (Romero et al., 1995). The relocated seismicity is well-correlated
with the inverted low Vp/Vs anomalies (Fig. 6e and f). Bigger event
clusters seem to occur at the lower boundary of the anomalous body
(∼4 km depth), similar to the seismicity distribution observed by
O’Connell and Johnson (1991). Gritto et al. (2013) also observed the
correlation of seismicity with higher Vp/Vs estimates. However, our
resolved area is dominated by low Vp/Vs anomalies and does not show
this pattern.
With more and more seismic stations and records available in the
past few decades, the correlation between injection and seismicity has
been commonly observed in The Geysers (e.g., Majer and Mcevilly,
1979; Denlinger and Bufe, 1980; Bufe and Shearer, 1981; Allis, 1982;
Eberhart-Phillips and Oppenheimer, 1984; O’Connell and Johnson,
1991; Stark, 1992; Romero et al., 1995; Alwyn, 1996), although the
seismicity rate and shape vary over the years. The recent work by
Trugman et al. (2016) observed influence of fluid injection on the
seismicity rate, b value, earthquake magnitude, and event depth within
The Geysers. In order to investigate the spatial relationship between
seismicity and injection, we plot the locations of the operational wells
obtained from the California Oil, Gas, and Geothermal Resources
(DOGGR) together with the relocated events in similar event clusters
(Fig. 7d). Although not all the wells have their corresponding
231
Geothermics 71 (2018) 225–233
earthquake clusters, in general seismicity seems to occur where the
wells are located. In addition, the seismicity at depth is more correlated
with the resolved Vp/Vs anomalies (Fig. 6e and f) than the absolute Vp
variations (Fig. 5e and f), suggesting that the seismicity is induced. We
also acquired the well extraction and injection data from the DOGGR
and plot their annual rates as a function of time for our study time
period (Fig. 9a). Between 1984 and 1995, the net production (i.e., ex-
traction-injection) rate showed similar pattern as the extraction because
the injection remained almost constant. Since 1995, the injection rate
has been fluctuating with a general increasing trend and the extraction
rate has been nearly flat. Following the method of the maximum cur-
vature by Wiemer and Baer (2000), we calculated the magnitude of
completeness for the relocated events in our study as 1.4. In Fig. 9a, we
plot the annual seismicity rate for events above magnitude 1.4, which is
correlated with the injection rate extremely well. The seismicity rate
increases in 1997 and 2003 correspond to the pipeline constructions
during those time periods, which was also observed in Trugman et al.
(2016). The event spatio-temporal correlation with the well operations
and their shallow depths (≤4 km bsl) further support that the majority
of the seismicity is induced seismicity.
Our study area is dominated by normal (51%) and strike-slip (38%)
focal types, consistent with the extensional and strike-slip tectonic
setting in the area surrounding The Geysers. For events above 1 km
depth, 56% of the well-resolved focal solutions are strike-slip and 29%
are normal, indicating the shallow events are more affected by tectonic
stress and deeper ones by operational activity. We also apply the focal
mechanism stress algorithm SATSI by Hardebeck and Michael (2006) to
invert for the stress field in the study area. The N28°E orientation of the
inverted maximum horizontal compressional stress SH max is consistent
with previous focal-mechanism-based stress directions in the area (e.g.,
Oppenheimer, 1986; Boyle and Zoback, 2014; Heidbach et al., 2016). In
order to investigate the temporal variations of the event focal types, if
there are any, we plot the scalar focal solutions for all the relocated
events above magnitude 1.4 with quality A and B focal mechanisms in
G. Lin, B. Wu
Fig. 9. (a) Background seismicity rate compared with the operational fluid volumes at the
Geyers Geothermal Field. (b) Scalar solutions for quality A and B focal mechanisms
(M ≥ 1.4). Annual average is shown by the black circles. Vertical arrows mark 2008,
when the average focal scalar started to increase.
Fig. 9b. There is a gap in the focal mechanism data in 2007 because the
waveforms are not available from the NCSN for events between 2007/
04/23 and 2007/07/20. The average focal solution before 2008 is
about −0.37, indicating the dominating normal and strike-slip faulting
regime. However, it rose to about −0.23 since 2008, which was mainly
due to the increase in the number of reverse faulting after 2008. In
order to examine this temporal change, we plot both the number of
focal solutions and percentage of each focal type as a function of time in
Fig. 10 and found that both have increased since 2008. In addition, the
percentage of normal faulting shows the corresponding reduction. This
focal type variation matches with the dates of the Northwest Geysers
EGS project (2008/09/30–2012/10/31) very well, which aimed at en-
hancing production from a known high temperature reservoir located
under a conventional geothermal steam reservoir. The increased re-
verse faulting after 2008 in our study area is consistent with the thermal
contraction created by this project.
5. Conclusions
In this paper, we present a comprehensive investigation of the
structural variations near the Geysers Geothermal Field, including a
new 3-D Vp and Vp/Vs tomographic model, precise earthquake reloca-
tions, and focal mechanisms based on local seismic data. Our Vp model
mainly reflects the local geological features and the Vp/Vs model is
more sensitive to the fluid saturation in the geothermal reservoir. The
local seismicity shows both spatial and temporal correlations with the
geothermal operations in the study area. Although the focal solutions
232
Geothermics 71 (2018) 225–233
Fig. 10. (a) Number of earthquakes above magnitude 1.4 per year with quality A and B
focal mechanisms. (b) Percentage of each focal type per year. Pink arrows mark the time
period of the Northwest Geysers EGS project (2008/09/30–2012/10/31). (For inter-
pretation of the references to color in this figure legend, the reader is referred to the web
version of the article.)
are dominated by normal and strike-slip regimes throughout our entire
study time period, there is an increase in reverse faulting since 2008,
which may be due to the thermal contraction associated with the
Northwest Geysers EGS project. Our study provides a groundwork for
the future remote triggering study and other seismological studies in
The Geysers.
Acknowledgements
Seismic data for this study were accessed through the Northern
California Earthquake Data Center (NCEDC), doi:10.7932/NCEDC. We
thank the anonymous reviewer who provided constructive comments.
Figures were made using the public domain GMT software (Wessel and
Smith, 1991) and matlab. This research work was supported by the
National Science Foundation grant EAR-1447105.
References
Allis, R., 1982. Mechanism of induced seismicity at The Geysers geothermal reservoir,
California. Geophys. Res. Lett. 9, 629–632.
Alwyn, R.C., 1996. The geysers geothermal area, California: tomographic images of the
G. Lin, B. Wu
depleted steam reservoir and non-double-couple earthquakes. Durham theses),
Durham University. Durham E-Theses Online: http://etheses.dur.ac.uk/5349/.
Boitnott, G.N., Kirkpatrick, A., 1997. Interpretation of field seismic tomography at The
Geysers geothermal field, California. In: Proceedings, Twenty-Second Workshop on
Geothermal Reservoir Engineering, Stanford University. Stanford, California, January
27–29, 1997, SGP-TR-I 55.
Boyle, K., Zoback, M., 2014. The stress state of the northwest geysers, California geo-
thermal field, and implications for fault-controlled fluid flow. Bull. Seismol. Soc. Am.
104, 1–10.
Bufe, C.G., Shearer, P., 1981. Geothermal induced seismicity-Bane or blessing.
Proceedings Second DOE-ENEL Workshop for Cooperative Research in Geothermal
Energy 86–97.
Denlinger, R., Bufe, C., 1980. Seismicity induced by steam production at The Geysers
steam field in northern California. Eos Trans. AGU 61, 1051.
Eberhart-Phillips, D., 1986. Three-dimensional velocity structure in the northern
California Coast Ranges from inversion of local earthquake arrival times. Bull.
Seismol. Soc. Am. 76, 1025–1052.
Eberhart-Phillips, D., 1990. Three-dimensional P and S velocity structure in the Coalinga
region, California. J. Geophys. Res. 95, 15343–15363.
Eberhart-Phillips, D., Oppenheimer, D.H., 1984. Induced seismicity in The Geysers geo-
thermal area, California. J. Geophys. Res. 89, 1191–1207.
Efron, B., Gong, G., 1983. A leisurely look at the bootstrap, the jackknife and cross-va-
lidation. Am. Stat. 37, 36–48.
Efron, B., Tibshirani, R., 1991. Statistical data analysis in the computer age. Science 253,
390–395.
Foulger, G.R., Grant, C.C., Ross, A., Julian, B.R., 1997. Industrially induced changes in
Earth structure at The Geysers geothermal area, California. Geophys. Res. Lett. 24,
135–137.
Gritto, R., Jarpe, S.P., 2014. Temporal variations of Vp/Vs-ratio at The Geysers geo-
thermal field, USA. Geothermics 52, 112–119. http://dx.doi.org/10.1016/j.
geothermics.2014.01.012.
Gritto, R., Yoo, S.H., Jarpe, S.P., 2013. Three-dimensional seismic tomography at the
Geysers geothermal field, CA, USA. In: Proceedings, Thirty-Eighth Workshop on
Geothermal Reservoir Engineering, Stanford University. Stanford, California,
February 11-13, 2013, SGP-TR-198.
Gunasekera, R., Foulger, G., Julian, B., 2003. Reservoir depletion at The Geysers geo-
thermal area, California, shown by four-dimensional seismic tomography. J.
Geophys. Res. 108, 2134. http://dx.doi.org/10.1029/2001JB000638.
Hardebeck, J., Shearer, P., 2002. A new method for determining first-motion focal me-
chanisms. Bull. Seismol. Soc. Am. 92, 2264–2276.
Hardebeck, J.L., Michael, A.J., 2006. Damped regional-scale stress inversions: metho-
dology and examples for southern California and the Coalinga aftershock sequence. J.
Geophys. Res. 111, B11310. http://dx.doi.org/10.1029/2005JB004144.
Heidbach, O., Rajabi, M., Reiter, K., Ziegler, M., 2016. The WSM Team: World Stress Map
Database Release 2016, GFZ Data Services.
Iyer, H., Oppenheimer, D., Hitchcock, T., Roloff, J., Coakley, J., 1981. Large teleseismic p-
wave delays in the geysers-clear lake geothermal area. Research in The Geysers-Clear
Lake Geothermal Area, Northern California, vol. 1141. pp. 97–116.
Julian, B.R., Ross, A., Foulger, G.R., Evans, J.R., 1996. Three-dimensional seismic image
of a geothermal reservoir: the Geysers, California. Geophys. Res. Lett. 23, 685–688.
http://dx.doi.org/10.1029/95GL03321.
Kirkpatrick, A., Peterson, J.J., Majer, E., 1997. Three-dimensional compressional- and
shear-wave seismic velocity models for the southeast Geysers geothermal field,
California. Proceedings 27th Workshop on Geothermal Reservoir Engineering, vol. 22
399–410.
Lin, G., 2013. Three-dimensional seismic velocity structure and precise earthquake re-
locations in the Salton Trough, southern California. Bull. Seismol. Soc. Am. 103,
2694–2708. http://dx.doi.org/10.1785/0120120286.
Lin, G., 2015. Seismic velocity structure and earthquake relocation for the magmatic
system beneath Long Valley Caldera, eastern California. J. Volcanol. Geotherm. Res.
296, 19–30.
Lin, G., Okubo, P.G., 2016. A large refined catalog of earthquake relocations and focal
mechanisms for the Island of Hawai’i and its seismotectonic implications. J. Geophys.
Res. http://dx.doi.org/10.1002/2016JB013042.
Lin, G., Shearer, P.M., 2009. Evidence for water-filled cracks in earthquake source re-
gions. Geophys. Res. Lett. 36.
233
Geothermics 71 (2018) 225–233
Lin, G., Shearer, P.M., Hauksson, E., 2007. Applying a three-dimensional velocity model,
waveform cross correlation, and cluster analysis to locate Southern California seis-
micity from 1981 to 2005. J. Geophys. Res. 112. http://dx.doi.org/10.1029/
2007JB004986.
Lin, G., Shearer, P.M., Matoza, R.S., Okubo, P.G., Amelung, F., 2014. Three-dimensional
seismic velocity structure of Mauna Loa and Kilauea volcanoes in Hawaii from local
seismic tomography. J. Geophys. Res. 119, 4377–4392. http://dx.doi.org/10.1002/
2013JB010820.
Lin, G., Thurber, C.H., Zhang, H., Hauksson, E., Shearer, P.M., Waldhauser, F., Brocher,
T.M., Hardebeck, J., 2010. A California statewide three-dimensional seismic velocity
model from both absolute and differential times. Bull. Seismol. Soc. Am. 100,
225–240. http://dx.doi.org/10.1785/0120090028.
Majer, E.L., Mcevilly, T.V., 1979. Seismological investigations at the Geysers geothermal
field. Geophysics 44, 246–269.
McLaughlin, R.J., 1981. Tectonic setting of pre-Tertiary rocks and its relation to geo-
thermal resources in The Geysers-Clear Lake area. Research in The Geysers-Clear
Lake Geothermal Area, Northern California, vol. 1141. pp. 3–23.
Meador, P.J., Hill, D.P., Luetgert, J.H., 1985. Data report for the July–August 1983
seismic-refraction experiment in the Mono Craters-Long Valley region, California.
U.S. Geol. Surv. Open-File Rept. 85-708 70 pp.
NCEDC, 2014. Northern California Earthquake Data Center. UC Berkeley Seismological
Laboratory. Dataset. http://dx.doi.org/10.7932/NCEDC.
O’Connell, D.R.H., Johnson, L.R., 1991. Progressive inversion for hypocenters and P wave
and S wave velocity structure: application to the Geysers, California, Geothermal
Field. J. Geophys. Res. 96, 6223–6236. http://dx.doi.org/10.1029/91JB00154.
Oppenheimer, D.H., 1986. Extensional tectonics at The Geysers Geothermal Area,
California. J. Geophys. Res.: Solid Earth 91, 11463–11476. http://dx.doi.org/10.
1029/JB091iB11p11463.
Oppenheimer, D.H., Herkenhoff, K.E., 1981. Velocity–density properties of the litho-
sphere from three-dimensional modeling at the Geysers-Clear Lake region, California.
J. Geophys. Res. 86, 6057–6065.
Romero, A.E., McEvilly, T.V., Majer, E.L., Vasco, D., 1995. Characterization of the geo-
thermal system beneath the Northwest Geysers steam field, California, from seismi-
city and velocity patterns. Geothermics 24, 471–487. http://dx.doi.org/10.1016/
0375-6505(95)00003-9.
Shearer, P.M., Prieto, G.A., Hauksson, E., 2006. Comprehensive analysis of earthquake
source spectra in Southern California. J. Geophys. Res. 111, B06303. http://dx.doi.
org/10.1029/2005JB003979.
Stark, M.A., 1992. Microearthquakes – a tool to track injected water in The Geysers re-
servoir. Monograph on the Geyser geothermal field, Special report 111–117.
Thompson, R.C., 1989. Structural stratigraphy and intrusive rocks at the geysers geo-
thermal field. Geotherm. Resour. Counc. Trans. 13, 481–485.
Thompson, R.C., 1992. Structural stratigraphy and intrusive rocks at The Geysers geo-
thermal field. Monogr. Geysers Geotherm. Field 17, 59–63.
Thurber, C., Eberhart-Phillips, D., 1999. Local earthquake tomography with flexible
gridding. Comput. Geosci. 25, 809–818.
Thurber, C., Zhang, H., Brocher, T., Langenheim, V.E., 2009. Regional three-dimensional
seismic velocity model of the crust and uppermost mantle of northern California. J.
Geophys. Res. 114, B01304. http://dx.doi.org/10.1029/2008JB005766.
Thurber, C.H., 1983. Earthquake locations and three-dimensional crustal structure in the
Coyote Lake area, central California. J. Geophys. Res. 88, 8226–8236.
Thurber, C.H., 1993. Local earthquake tomography: velocities and Vp/Vs-theory. In: Iyer,
H.M., Hirahara, K. (Eds.), Seismic Tomography: Theory and Practice. Chapman and
Hall, London, pp. 563–583.
Trugman, D.T., Shearer, P.M., Borsa, A.A., Fialko, Y., 2016. A comparison of long-term
changes in seismicity at The Geysers, Salton Sea, and Coso geothermal fields. J.
Geophys. Res. 121. http://dx.doi.org/10.1002/2015JB012510.
Wessel, P., Smith, W.H.F., 1991. Free software helps map and display data. Eos Trans.
AGU 72, 441.
Wiemer, S., Baer, M., 2000. Mapping and removing quarry blast events from seismicity
catalogs. Bull. Seismol. Soc. Am. 90, 525–530.
Zhang, Q., Lin, G., Zhan, Z., Chen, X., Qin, Y., Wdowinski, S., 2017. Absence of remote
earthquake triggering within the Coso and Salton Sea geothermal production fields.
Geophys. Res. Lett. http://dx.doi.org/10.1002/2016GL071964.
Zucca, J., Hutchings, L., Kasameyer, P., 1994. Seismic velocity and attenuation structure
of the Geysers geothermal field, California. Geothermics 23, 111–126.