Qingjun Meng, Aizhi Guo, Xiangteng Wang, Shuofan Wang (2019). Source parameter and rupture process of the MW6.3 early strong aftershock immediately following the 2016 MW7.8 Kaikoura earthquake (New Zealand). Earthq Sci 32(1): 1-11. DOI: 10.29382/eqs-2019-0001-1
Citation:
Qingjun Meng, Aizhi Guo, Xiangteng Wang, Shuofan Wang (2019). Source parameter and rupture process of the MW6.3 early strong aftershock immediately following the 2016 MW7.8 Kaikoura earthquake (New Zealand). Earthq Sci 32(1): 1-11. DOI: 10.29382/eqs-2019-0001-1
Qingjun Meng, Aizhi Guo, Xiangteng Wang, Shuofan Wang (2019). Source parameter and rupture process of the MW6.3 early strong aftershock immediately following the 2016 MW7.8 Kaikoura earthquake (New Zealand). Earthq Sci 32(1): 1-11. DOI: 10.29382/eqs-2019-0001-1
Citation:
Qingjun Meng, Aizhi Guo, Xiangteng Wang, Shuofan Wang (2019). Source parameter and rupture process of the MW6.3 early strong aftershock immediately following the 2016 MW7.8 Kaikoura earthquake (New Zealand). Earthq Sci 32(1): 1-11. DOI: 10.29382/eqs-2019-0001-1
The 2016 MW7.8 Kaikoura (New Zealand) earthquake was the most complex event ever instrumentally recorded and geologically investigated, as it ruptured on more than 12 fault segments of various geometries. To study the mainshock rupture characteristics, geodetic methods like InSAR and GPS play an essential role in providing satisfactory spatial resolution. However, early strong aftershocks may cause extra ground deformation which bias the mainshock rupture inversion result. In this paper, we will focus on studying the MW 6.3 aftershock, which is the only M6+ thrust slip aftershock that occurred only 30 minutes after the Kaikoura mainshock. We will relocate the hypocenter of this event using the hypo2000 method, make the finite fault model (FFM) inversion for the detailed rupture processes and calculate the synthetic surface displacement to compare with the observed GPS data and figure out its influence on the mainshock study. Although we are not able to resolve the real ruptured fault of this event because of limited observation data, we infer that it is a west-ward dipping event of oblique slip mechanism, consistent with the subfault geometries of the Kaikoura mainshock. According to the inverted FFM, this event can generate 10–20 cm ground surface displacement and affect the ground displacement observation at nearby GPS stations.
On November 13, 2016, a MW7.8 (Global CMT) earthquake struck the northeastern part of the South Island, New Zealand. Within 14 hours afterwards, four MW≥6.0 aftershocks occurred with different focal mechanisms including both thrust and strike-slip faulting (Kaiser et al., 2017). This earthquake sequence ruptured on more than 12 fault segments (Stirling et al., 2017), causing complicated surface breaks, extensive surface displacement and abundant landslides. The earthquake occurred in a region of complicated tectonics, featuring the inland Marlborough fault system which contains strike-slip faults such as Hope fault and reverse faults such as the Jordan thrust. The earthquake could also be related to the subduction of the Pacific ocean plate at Hikurangi trench (Wallace et al., 2007). Seismological, geological and geodetic studies demonstrated that the complex rupture processes provide an important case study of earthquake physics and fault system mechanics (Hamling et al., 2017; Kaneko et al., 2017; Stirling et al., 2017).
Based on dual high frequency (0.4–2.5 Hz and 0.2–1.0 Hz) teleseismic P wave data recorded at two arrays, Zhang et al. (2017) used the back-projection method on seismic array data and found that the earthquake rupture propagated for ~170 km all the way towards the northeastern corner of the South Island New Zealand and lasted for more than 100 s in time. They also performed low frequency finite fault inversion which shows multiple sub-events. Mostly with forward modeling, Duputel and Rivera (2017) analyzed long period waveform data and argued that the three sub-events source models could not explain some features observed on many broadband seismograms. Therefore, they proposed that the mainshock was mainly composed of four sub-events, involving both strike slip and thrust slip mechanisms, thus demonstrating unusual rupture complexity. Cesca et al. (2017) performed a broad seismological analysis using both regional and teleseismic seismograms, as well as GPS and InSAR data to determine the rupture process of the mainshock and aftershocks. Their result suggested slip distribution on at least three active fault domains. Clark et al. (2017) observed a highly variable coastal deformation and attributed it to very complicated ruptures along multiple fault segments. Stirling et al. (2017) made extensive field invesigation and confirmed surface ruptures along more than 12 faults, and the maximum displacement exceeded 10 m on the ground surface. Holden et al. (2017) inverted strong motion and high rate GPS data to resolve the rupture processes and demonstrated that most of the slip occurred on crustal faults and a very small amount of the ruptured energy was released from the subduction interface. However, Wang et al. (2018) inverted the rupture process utilizing local strong motion data, teleseismic waveforms, GPS and InSAR data. They proposed that the slab interface was substantially involved in the rupture processes, and in this way the far field surface displacement and teleseismic body waves could be better fitted. According to back-projection of the high-frequency (0.1–1 Hz) P-wave recordings, Du et al. (2018) found that the high-frequency sources distributed nearly along the lower edge of the rupture area. Lo et al. (2018) demonstrated that aftershocks tended to occur near the edges of the major slip zones of the mainshock. Among these studies, the GPS stations and InSAR images were especially useful for providing near field displacement information to study mainshock rupture process. Furthermore, field investigation on ground surface breaks is also essential in determining causative faults responsible for generation of the earthquakes.
However, sometimes those geodetic data can be “contaminated”, when post-seismic deformation and early strong aftershocks contribute additional ground surface displacement after mainshocks. Although the overall mainshock inversion will not be substantially affected by aftershocks, because the moment energy released from aftershocks is small compared with that from mainshocks. However, mainshock rupture models can be affected locally if the observed displacement is affected by nearby strong aftershocks. At present, campaign GPS data and InSAR images have limited usage for studying early strong aftershocks. For campaign GPS stations, displacement data is usually resolved per day. For InSAR data, researchers may try to use early InSAR images, such as images within 24–36 hours after the mainshock, to minimize post-seismic deformation effects (Hamling et al., 2017). However, it is still difficult to avoid the effect from early strong aftershocks in InSAR or static GPS data, because most of them occur within several hours or even tens of minutes after mainshock initiation, according to Omori’s law (Omori, 1894). Similarly, some surface breaks are not necessarily caused by mainshock, as shallow aftershocks may cause ground rupture. For example, Li et al. (2011) proposed that an early M6 aftershock of the 2010 Yushu earthquake caused surface break to the west of the mainshock, leading to overestimated rupture length with the earthquake scaling laws.
For the Kaikoura earthquake sequence of November 13, 2016, two M6+ and twelve M5+ aftershocks (GeoNet) occurred within the first hour after the mainshock as shown in Figure 1 and Table 1. Among them, the MW 6.3 (GeoNet No.8 event) aftershock occurred about 30 minutes after the mainshock. This event was recorded by the USGS and GeoNet, while there exists a discrepancy between the two organizations about magnitude, hypocenter and focal mechanism of this event as shown in Figure 1. USGS located the hypocenter 5 km to south of the GeoNet hypocenter, and the W-Phase magnitude from USGS MWW6.5 is also larger than the GeoNet moment magnitude MW6.3. Generally, this event locates near the end of the mainshock surface rupture on the Upper Kowhai fault or the Jordan thrust system. As it is the earliest M6 aftershock, we would like to study this event to estimate its influence on the locally observed surface displacement after the mainshock and to determine whether it may affect the mainshock inversion, as well as to investigate its potential in causing surface ruptures if it was very shallow.
Figure
1.
Early strong M≥5 aftershocks within the first hour following the Kairoura mainshock (red star), with hypocenter locations from GeoNet. Triangles (grey/black) are strong motion stations from GeoNet, blue cyan squares are high-rate GPS stations, grey squares are static GPS stations. Strong motion stations denoted by black triangles are used in this study for hypocenter relocation and finite fault modeling. For the Mw 6.3 aftershock in this study, the hypocenter results from USGS (grey dot), GeoNet (orange dot) and the hypo2000 relocation (yellow dot) are shown in the black dashed rectangle. The mapped ruptures by black lines are historical major active crustal ruptures in New Zealand, and ruptures by red lines are the observed Kaikoura mainshock surface ruptures (http://data.gns.cri.nz/af/)
In addition, accurate detection and characterization of early aftershock activity are helpful for interpreting aftershock sequence evolution and triggering mechanisms (Lengline et al., 2012; Chang et al., 2007). In this study, we will utilize the local strong motion data to study the MW6.3 aftershock. First, we relocate the hypocenter location using the hypo2000 method, utilizing manually picked and S arrival times from eight local strong motion stations. Second, we apply the simulated annealing (SA) method in our FFM inversion to determine the fault geometry and fault rupture processes. Lastly, we calculate the synthetic surface displacement based on the inverted FFM to evaluate its contribution to the geodetic observation data.
2.
Hypocenter relocation
Accurate hypocenter location is a prerequisite for studying earthquake rupture process of the MW6.3 aftershock. We downloaded seismic waveform data from GeoNet of New Zealand, which included both broadband and strong motion recordings. All the stations used in this study have three component data, making it possible to identify both P and S waves. First, we manually picked the P and S arrivals from three component velocity waveforms at eight local strong motion stations as shown in Figures 1 and 2. These strong motion stations are of good azimuth coverage with epicentral distance ranging within 20–80 km, as shown in Figures 1 and 2. The local P and S wave velocity models are taken from Eberhart-Phillips and Bannister (2010) and Lin et al. (2007). Then, using the hypo2000 method (Klein, 2002), we relocated the hypocenter of this event as ~10 km to the west of the GeoNet location. The relocated hypocenter depth is about 10 km, the same as the USGS depth but shallower than the GeoNet depth of 19 km. After relocation, the P and S arrival time residuals are all less than 0.5 s, averaging 0.28 s (Figure 2). And the uncertainty of the hypo2000 relocation is less than 1 km both horizontally and vertically. Compared with the locations provided by the USGS and GeoNet, the relocated hypo2000 hypocenter is closer to the southern end of the surface rupture on the Upper Kowhai fault. It’s possible that the MW6.3 aftershock is triggered by the mainshock rupture on this fault segment.
Figure
2.
(a) Observed P (dashed lines) and S (solid lines) arrivals of the No.9 aftershock at nearby strong motion stations; (b) Misfit between observed and calculated P (open circles) and S arrivals (solid dots) at nearby strong motion stations listed in (a); (c) P (dashed line) and S (solid line) wave velocity models used for hypocenter relocation and finite fault inversion processes
A M6.3 earthquake could be well studied with teleseismic body waves (Chen et al., 2015) because of the relatively simple propagation effects of P and SH waves. However, this aftershock occurred immediately after the mainshock, teleseismic body waves are strongly contaminated by the mainshock surface coda waves. Therefore we only use local waveform data to study this event. Furthermore, to avoid the influence of local crustal velocity structure uncertainty, we only utilize waveforms at nine local strong motion stations within distance of 80 km shown as black triangles in Figure 1. We calculate the synthetic waveforms using a 1-D layered P and S velocity models from Eberhart-Phillips and Bannister (2010) and Lin et al. (2007), shown in Figure 2c. Green’s function are computed using the frequency-wavenumber integral algorithm (Zhu and Rivera, 2002). To suppress the effects caused by complex 3-D velocity structure, we band-pass filter both the observed and synthetic velocity waveforms to low frequency band according to the epicentral distance of each station. Waveforms at stations KHZ and KIKS are band-pass filtered by 0.05–0.3 Hz, and both synthetic and observed data are aligned by origin time. Waveforms at farther stations, such as MOLS, KEKS, WTMC, HSES, CECS, WDFS and WVFS, are band-pass filtered by 0.05–0.2 Hz, and both synthetic and observed data are aligned either by P or S arrivals, shown in Figure 3c.
Figure
3.
Finite fault inversion result on the fault model of strike 229° and dip 34°. (a) The inverted source time function; (b) The inverted finite fault slip model, with fault rupture velocity setting at 2.5 km/s; (c) The matching result between the observed (black) and synthetic velocity waveforms (red) at local strong motion stations. Arrows denote whether the waveforms are aligned by origin time (O), P arrival time (P) or S arrival time (S)
There are many algorithms in performing inversion of finite fault model (FFM), and they have been successfully applied in modeling strong earthquakes (Kikuchi et al., 1993; Ji et al., 2002; Zhang et al., 2009; Wang et al., 2008). Here we adopt the method used by Yue and Lay (2013), which utilizes a multi time-window linear inversion technique (Hartzell and Heaton, 1983). For model set-up, we divide our FFM into 7×7 subfaults with each subfault of 3 km×3 km, and the total fault dimensions are 21 km×21 km. According to the earthquake scaling law (Leonard, 2010), the fault length of a MW 6.3 earthquake can reach ~12 km. Considering the possibility of unilateral rupture, we set the total length of the FFM as ~20 km. The source time function of each subfault is composed by 6 symmetric triangles, with rise time of 0.5 s and offset of 0.5 s. The total duration of the source time function is approximately 8 s. In each subfault, two slip-vector orientations are set to allow rake-varying slip, applying a nonnegative least squares inversion (Lawson and Hanson, 1995) in which the rake of each subfault is allowed to vary between ±45° around the central rake angle. The rupture time of each subfault is determined by a uniform rupture-front expansion velocity, which we set at 2.5 km/s assuming a sub-shear rupture process. Because most of the data we use comes from strong motion stations farther than 40 km from the source, and observations from nearby stations are limited, the rupture expansion velocity value is not very precisely constrained.
3.2
Searching for fault geometry
Earthquakes point source mechanism can provide a reference fault geometry for the FFM. However, there exists a discrepancy of the point source mechanisms between the USGS and GeoNet, especially for the values of dipping angle, shown in Table 2. The GeoNet mechanism of strike/dip/rake, 201°/31°/113° has a steeper dipping angle than the USGS mechanism of strike/dip/rake, 218°/16°/109°. Thus we make a search of the strike angle, dip angle and hypocenter depth using simulated annealing (SA) method (Kirkpatrick et al., 1983) by fitting the observed three component velocity waveforms with synthetic velocity waveforms on nine local strong motion stations, mentioned above. Though, the focal mechanism results from the USGS and GeoNet are different, they are helpful for narrowing the searching range of parameters. While searching for parameters in the SA procedure, the hypocenter latitude is fixed at 42.66°S and longitude is fixed at 173.534°E, using the hypo2000 relocation result. Because there exists a trade-off between origin time and depth in the hypo2000 method, we will also search for the hypocenter depth. Because we could not determine the real fault geometry from two conjugate fault planes in point source focal mechanism, we need to make the FFM inversion based on two possible fault planes.
Table
2.
Parameters for the Mw 6.3 aftershock on Nov. 13, 2016
For fault model 1 (FM1), the searching range for strike angle is 180° to 260°, for dip angle is from 10° to 50° and for depth is from 6 to 20 km. The central rake angle is set as 120°, close to the rake angle values from the USGS and GeoNet and allowing the rake angle in each subfault to vary between 75° and 165°. After 1000 calculations steps in SA trial 1, we select the optimal parameters: strike 229°, dip 34° and depth 11.8 km when misfit ratio reaches minimum of ~0.35. To test the stability of SA searching result, we apply an independent SA trial 2 and the best searching parameters are very similar to SA trial 1, as shown in Figure S1, which proves that the SA searching parameters are reliable and stable.
Figure
S1.
Parameter searching result for FM1 plane. Four subfigures in (a) SA trial 1 and (b) SA trial 2 display the misfit, hypocenter depth, strike, dip at each accepted calculation step, grey areas are final parameters that become stable.
For fault model 2 (FM2), the searching range for strike angle is –40° to 40° degrees, for dip angle is from 40° to 90° and for depth is from 6 to 20 km. The central rake angle is set as 70°, allowing the rake angle in each subfault to vary between 25° and 115°. After 1000 calculations steps in SA trial 1, the best searching parameters are: strike –3°, dip 71° and depth 10.2 km, when misfit ratio reaches minimum at ~0.34. We also apply an independent SA trial 2, and the best searching parameters are very similar to SA trial 1, as shown in Figure S2.
Figure
S2.
Parameter searching result for FM2 plane. Four subfigures in (a) SA trial 1 and (b) SA trial 2 display the misfit, hypocenter depth, strike, dip at each accepted calculation step, grey areas are final parameters that become stable.
For FM1 (strike 229°, dip 34° and depth 11.8 km), Figure 3 shows the source time function, slip model and waveform fitting. The half duration of source time function tc is about 3.8 s, the maximum slip is 1.8 m and the magnitude of MW 6.39 is between the magnitude values from USGS MWW 6.5 and GeoNet MW 6.3. By calculating the moment weighted average rake angle on all subfaults, the overall rake angle is 137°, representing an oblique slip earthquake. The centroid location is –43.27° in latitude, 173.53°in longitude and 11.8 km in depth, similar to the hypocenter location, indicating that FM1 is a circular rupture.
For FM2 (strike –3°, dip 71° and depth 10.2 km), Figure 4 shows the source time function, slip model and waveform fitting situation. The half duration of source time function tc is about 3.6 s, which is similar to the tc value for FM1. The maximum slip is 1.9 m and the magnitude of MW 6.34 is close to the GeoNet magnitude MW 6.3. The moment weighted average rake angle is 70°, mainly reverse slip component. The centroid location is –43.27° in latitude, 173.54° in longitude and 12.2 km in depth, representing a slightly downward rupture direction, inferring from a relatively shallower hypocenter depth of 10.2 km.
Figure
4.
Finite fault inversion result on the fault model of strike –3° and dip 71°. (a) The inverted source time function; (b) The inverted finite fault slip model, with fault rupture velocity setting at 2.5 km/s; (c) The matching result between the observed (black) and synthetic velocity waveforms (red) at local strong motion stations, the waveforms are aligned in the same way as that in Figure 3c
From the waveform fitting situation, we could not determine whether FM1 or FM2 is the real fault rupture geometry, because they fit the observed strong motion data nearly equally well. It probably results from lacking of adjacent seismic stations within the epicentral distance of a fault length and that the fault did not ruptured unilaterally in the strike direction but ruptured with a circular rupture front, which could be regarded as a point source distantly.
4.
Surface displacement
Based on the FFMs we inverted using strong motion data previously, we could calculate the surface displacement generated from them and estimate its influence on surrounding GPS observation. We use the Coulomb 3 software (Toda et al., 2011) to compute the horizontal and vertical ground surface displacements generated by the MW 6.3 aftershock. In this process, the earthquake faults are modeled by setting displacements on 7×7 rectangular areas, according to the FFMs, in an elastic half-space (Okada, 1992). For FM1 strike/dip of 229°/34°, the horizontal and vertical surface displacements are shown in Figure 5. The maximum displacement can reach 10 cm both horizontally and vertically. For FM2 strike/dip of –3°/71°, the horizontal and vertical surface displacements are shown in Figure 6, with maximum horizontal and vertical displacements of ~10 cm and ~25 cm. In Figures 5 and 6, we also plot the horizontal and vertical displacements observed by nearby GPS stations since the Kaikoura mainshock. The horizontal displacement on nearest GPS station 1165 is approximately 2 m and the vertical displacement is ~0.5 m. Thus, the local influence of the MW 6.3 event on surface displacement, recorded by 1165, is about 5% horizontally and 20%–50% vertically. Such an influence within an area of ~20 km×20 km will not affect the overall mainshock rupture inversion, but should be considered locally before using the GPS and InSAR data for mainshock study
Figure
5.
Synthetic horizontal (a) and vertical (b) surface ground motion generated by the slip model on fault geometry of strike 229° and dip 34° shown in Figure 3. The purple and orange arrows represent the observed GPS horizontal and vertical offsets after the Kaikoura mainshock. The green dot array represents the location of the finite fault grids projected onto the ground surface. The mapped ruptures by black lines are historical major active crustal ruptures in New Zealand, and ruptures by red lines are the observed Kaikoura mainshock surface ruptures (http://data.gns.cri.nz/af/). The relocated hypocenter of the No. 9 event is shown by a yellow solid circle
Figure
6.
Synthetic horizontal (a) and vertical (b) surface ground motion generated by the slip model on fault geometry of strike –3° and dip 71° shown in Figure 4. The purple and orange arrows represent the observed GPS horizontal and vertical offsets after the Kaikoura mainshock. The green dot array represents the location of finite fault grids projected onto the ground surface. The mapped ruptures by black lines are historical major active crustal ruptures in New Zealand, and ruptures by red lines are the observed Kaikoura mainshock surface ruptures (http://data.gns.cri.nz/af/). The relocated hypocenter of the No. 9 event is shown by a yellow solid circle
Currently, we find that it is difficult to discriminate the real fault geometry for the MW 6.3 aftershock using only the seismic waveforms. Because the two fault geometries of FM1 and FM2 can fit the observed strong motion data nearly equally well. For geodetic data such as static GPS and InSAR data, observed displacement is dominated by the mainshock and the displacement generated by an early strong aftershock cannot be easily separated from it. However, several high-rate GPS stations displayed above the fault planes may be helpful for constraining the fault geometry of this event. We calculate the difference of displacement caused by the two fault geometries of FM1 and FM2 horizontally, shown in Figure 7a, and vertically, shown in Figure 7b. The difference is as large as 10 cm horizontally and 20 cm vertically above the fault zone area. However, this big displacement difference is only limited within an area of 20 km×20 km, consistent with the fault size. The displacement difference is very small at locations only several kilometers from the fault planes, for example at stations like high-rate GPS station KAIK and campaign GPS station 1166, shown in Figures 7a and 7b. In this way, a dense high-rate GPS array, with distance between stations less than 10 km, should be helpful to determine the geometry of a ~M6.3 shallow earthquake that occurred on unmapped faults.
Figure
7.
Synthetic difference of horizontal (a) and vertical (b) surface ground motion between that generated by the slip model on fault geometry of strike 229° and dip 34° shown in Figure 3 and that generated by the slip model on fault geometry of strike –3° and dip 71° shown in Figure 4. The green dot arrays represent the locations of finite faults grids, striking 229° and and –3°, projected onto the ground surface. The mapped ruptures by black lines are historical major active crustal ruptures in New Zealand, and ruptures by red lines are the observed Kaikoura mainshock surface ruptures (http://data.gns.cri.nz/af/). The relocated hypocenter of the No. 9 event is shown by a yellow solid circle. (c) and (d) are map view and 3-D side view of the Mw 6.3 aftershock fault geometry of FM1 comparing with the nearby mainshock fault segments (Wang et al., 2018). The fault with colored slip patches represents the MW 6.3 FM1 fault plane
However, the high-rate GPS stations were sparse in New Zealand and no high-rate GPS stations were installed near the MW 6.3 aftershock rupture zone. At last, we try to determine the real fault geometry based on the mainshock dipping trend. From previous studies on the mainshock rupture process, nearly all the fault segments dip westward (Hamling et al., 2017; Wang et al., 2018). If the MW6.3 aftershock conform with this trend and the real fault geometry could be better represented by FM1 strike/dip: 229°/34° rather than FM2 strike/dip of –3°/71°. In addition, the strike angle 229°of FM1 is more consistent with the strike angle of the Upper Kowhai fault, shown in Figure 1.We plot the FM1 fault together with the nearby mainshock fault segments 5, 6, 7 and 8a from Wang’s mainshock slip model (Wang et al., 2018), shown in Figure 7c and 7d. We find that the rupture of the MW 6.3 aftershock initiated from the gap zone between fault segments 5 and 6. The aftershock fault plane located closely beneath and to the east of these two fault segments.
6.
Conclusions
Based on the searching result from the simulated annealing method and geometries of the mainshock fault segments, we prefer that the MW6.3 aftershock occurred on a fault striking 229° and dipping 34°. The rupture initiated on depth of 10 km and propagated along the fault plane with a circular rupture front. The maximum slip is about 1.8 m near the centroid location and the maximum surface displacements generated are more than 10 cm both in horizontal and vertical directions. The surface displacements are approximately 5% and 20% of the observed local horizontal and vertical GPS displacements respectively. Such an influence may affect an area of 20 km×20 km. Thus, this event may not affect the overall feature of the Kaikoura mainshock inversion. However, for other moderately sized mainshocks, surface displacement caused by adjacent early strong aftershocks should not be neglected. In addition, this aftershock is not very shallow (about 10 km), it should not have caused surface breaks, as confirmed with the FFM inversion results. Therefore, the observed mainshock rupture near the Upper Kowhai fault is probably reliable and is not an artifact due to this strong aftershock.
Through studying this early strong aftershock, we find that static GPS stations and InSAR images have a difficulty in differentiating the early strong aftershock effect, continuous GPS data and seismic waveform data are very helpful in distinguishing mainshocks and aftershocks. High-rate GPS stations near the fault rupture zone are also useful for constraining the real fault geometry. In the future, implementing a dense joint network of GPSs and accelerometers around active fault zones is beneficial (Tu et al., 2013). Because it could provide real-time detection of ground motion and estimation of ground surface displacement, this joint network could effectively discriminate the surface displacement caused either by co-seismic deformation, post-seismic deformation or strong aftershocks. But dual-frequency GPS receivers are quite expensive, integration of low cost single-frequency GPS and MEMS would be a more feasible approach in installing dense observation network.
Acknowledgments
We thank the New Zealand GeoNet (http://geonet.org.nz) project and its sponsors EQC, GNS Science and LINZ for providing earthquake event and mechanism catalog, strong motion and high-rate GPS data used in this study. We thank JAXA and GSI for providing ALOS-2 data and images. Earthquake catalog also comes from USGS website (https://earthquake.usgs.gov/) and Global CMT (http://www.globalcmt.org/). This work made use of GMT and SAC software. This research is financially supported by National Natural Science Foundation of China (Nos. 41590854 and 41461164003).
Chang CH, Wu YM, Zhao L and Wu FT (2007) Aftershocks of the 1999 Chi-Chi, Taiwan, earthquake: The first hour. Bull Seismol Soc Am 97(4): 1 245–1 258 doi: 10.1785/0120060184
Cesca S, Zhang Y, Mouslopoulou V, Wang R, Saul J, Savage M, Heimann S, Kufner S-K, Oncken O and Dahm T (2017) Complex rupture process of the Mw7.8, 2016, Kaikoura earthquake, New Zealand, and its aftershock sequence Earth Planet Sci Lett 478: 110–120 doi: 10.1016/j.jpgl.2017.08.024
Chen W, Ni S, Kanamori H, Wei S, Jia Z and Zhu L (2015) CAPjoint, a computer software package for joint inversion of moderate earthquake source parameters with local and teleseismic waveforms. Seismol Res Lett 86(2A): 432–441 doi: 10.1785/0220140167
Clark K J, Nissen E K, Howarth J D, Hamling I J, Mountjoy J J , Ries W F, Jones K , Goldstien S, Cochran U A, Villamor P , Hreinsdóttir S , Litchfield N J, Mueller C, Berryman K R and Strong D T (2017) Highly variable coastal deformation in the 2016 Mw7.8 Kaikōura earthquake reflects rupture complexity along a transpressional plate boundary Earth Planet Sci Lett 474: 334–344 doi: 10.1016/j.jpgl.2017.06.048
Du H, Zhang X, Xu L, Feng W, Yi L and Li P (2018) Source complexity of the 2016 Mw7.8 Kaikoura (New Zealand) earthquake revealed from teleseismic and InSAR data Earth Planet Phys 2: 1–17 doi: 10.26464/epp2018029
Duputel Z and Rivera L (2017) Long-period analysis of the 2016 Kaikoura earthquake Phys. Earth Planet. Inter. 265: 62–66 doi: 10.1016/j.pepi.2017.02.004
Eberhart-Phillips D and Bannister S (2010) 3-D imaging of Marlborough, New Zealand, subducted plate and strike-slip fault systems. Geophys J Int 182(1): 73–96 doi: 10.1111/j.1365-246X.2010.04621.x
Hamling IJ, Hreinsdóttir S, Clark K, Elliott J, Liang C, Fielding E, Litchfield N, Villamor P, Wallace L, Wright T J, D'Anastasio E, Bannister S, Burbidge D, Denys P, Gentle P, Howarth J, Mueller C, Palmer N, Pearson C, Power W, Barnes P, Barrell AJD, Van Dissen R, Langridge R, Little T, Nicol A, Pettinga J, Rowland J and Stirling M (2017) Complex multifault rupture during the 2016 Mw7.8 Kaikōura earthquake, New Zealand. Science 356 (6334): aam7194
Hartzell SH and Heaton TH (1983) Inversion of strong ground motion and teleseismic waveform data for the fault rupture history of the 1979 Imperial Valley, California, earthquake. Bull Seismol Soc Am 73(6A): 1 553–1 583
Holden C, Kaneko Y, D’Anastasio E, Benites R, Fry B and Hamling IJ (2017) The 2016 Kaikōura earthquake revealed by kinematic source inversion and seismic wavefield simulations: Slow rupture propagation on a geometrically complex crustal fault network. Geophys Res Lett 44(11): 320–328 doi: 10.1002/2017GL075301
Ji C, Wald DJ and Helmberger DV (2002) Source description of the 1999 Hector Mine, California, earthquake (part I): Wavelet domain inversion theory and resolution analysis. Bull Seismol Soc Am 92(4): 1 192–1 207 doi: 10.1785/0120000916
Kaiser AE, Balfour N, Fry B, Holden C, Litchfield N, Gerstenberger M, D’Anastasio E, Horspool N, McVerry G, Ristau J, Gledhill K, Bannister S, Christopherson A, Clark K, Power W and Rhoades D (2017) The 2016 Kaikōura (New Zealand) earthquake: Preliminary seismological report. Seismol Res Lett 88(3): 727–739 doi: 10.1785/0220170018
Kaneko Y, Fukuyama E and Hamling I J (2017) Slip-weakening distance and energy budget inferred from near-fault ground deformation during the 2016 Mw7.8 Kaikoura earthquake. Geophys Res Lett 44(10): 4 765–4 773 doi: 10.1002/2017GL073681
Kikuchi M, Kanamori H and Satake K (1993) Source complexity of the 1988 Armenian earthquake: Evidence for a slow after-slip event. J Geophys Res 98(B9): 15 797–15 808 doi: 10.1029/93JB01568
Kirkpatrick S, Gelatt CD and Vecchi MP (1983) Optimization by simulated annealing. Science 220(4598): 671
Klein FW (2002) User’s guide to HYPOINVERSE2000, a Fortran program to solve for earthquake locations and magnitudes. US Geol Surv Open-File Rep, 02-172, 123., 01-113, Menlo Park, California
Lawson CL and Hanson R J (1995) Solving Least Squares Problems, Classics in Applied Mathematics, Society for Industrial Mathematics. Englewood Cliffs N J, vol. 15, 337
Lengline O, Enescu B, Peng Z and Shiomi K (2012) Decay and migration of the early aftershock activity following the Tohoku Mw9.0 2011 earthquake Geophys Res Lett 39(18): L18309 doi: 10.1029/2012GL052797
Leonard M (2010) Earthquake fault scaling: Self-consistent relating of rupture length, width, average displacement, and moment release. Bull Seismol Soc Am 100(5A): 1 971–1 988 doi: 10.1785/0120090189
Li Z, Elliott JR, Feng W, Jackson JA, Parsons BE and Walters RJ (2011) The 2010 MW 6.8 Yushu (Qinghai, China) earthquake: Constraints provided by InSAR and body wave seismology J Geophys Res 116(B10): B10302
Lin F, Ritzwoller MH, Townend J, Savage M and Bannister S (2007) Ambient noise Rayleigh wave tomography of New Zealand. Geophys J Int 170(2): 649–666 doi: 10.1111/j.1365-246X.2007.03414.x
Lo Y, Zhao L, Xu X, Chen J and Hung S (2018) The 13 November 2016 Kaikoura, New Zealand earthquake: Rupture process and seismotectonic implications. Earth Planet Phys 2(2): 139–149
Okada Y (1992) Internal deformation due to shear and tensile faults in a half-spacr. Bull Seismol Soc Am 82(2): 1 018–1 040
Omori F (1894) On the aftershocks of earthquakes J Coll Sci Imper Univ Tokyo 7: 111–200
Stirling M W, Litchfield NJ, Villamor P, Van Dissen R J, Nicol A, Pettinga J, Barnes P, Langridge R M, Little T, Barrell DJA, Mountjoy J, Ries WF, Rowland J, Fenton C, Hamling I, Asher C, Barrier A, Benson A, Bischoff A, Borella J, Carne R, Cochran UA, Cockroft M, Cox SC, Duke G, Fenton F, Gasston C, Grimshaw C, Hale D, Hall D, Hao K, Hatem A, Hemphill-Haley M, Heron D W, Howarth J, Juniper Z, Kane T, Kearse J, Khajavi N, Lamarche G, Lawson S, Lukovic B, Madugo C, Manousakis J McCol l S, Noble D, Pedley K, Sauer K, Stahl T, Strong D T, Townsend D B, Toy V, Villeneuve M, Wandres A, Williams J, Woelz S and Zinke R (2017) The Mw 7.8 2016 Kaikoura earthquake: Surface fault rupture and seismic hazard context Bull NZ Soc Earthqu Eng 50: 73–84
Toda S, Stein RS, Sevilgen V, and Lin J (2011) Coulomb 3.3 Graphic-Rich Deformation and Stress-Change Software for Earthquake, Tectonic, and Volcano Research and Teaching-User Guide. U. S. Surv Open-File Rep 2011–1060
Tu R, Wang, R, Ge M, Walter T R, Ramatschi M, Milkereit C, Bindi D and Dahm T (2013) Cost-effective monitoring of ground motion related to earthquakes, landslides or volcanic activity by joint use of a single-frequency GPS and a MEMS accelerometer. Geophys. Res Lett 40(15): 3 825–3 829 doi: 10.1002/grl.50653
Wallace L M, Beavan R J, McCaffrey R, Berryman K R and Denys P (2007) Balancing the plate motion budget in the South Island, New Zealand using GPS, geological and seismological data. Geophys J Int 168(1): 332–352 doi: 10.1111/j.1365-246X.2006.03183.x
Wang WM, Zhao L F, Li J and Yao ZX (2008) Rupture process of the Ms8.0 Wenchuan earthquake of Sichuan, China. Chin J Geophys 51(5): 1 403–1 410 (in Chinese with English abstract) doi: 10.1785/0120120119
Wang T, Wei S, Shi X, Qiu Q, Li L, Peng D, Weldon R J and Barbot S (2018) The 2016 Kaikōura earthquake: Simultaneous rupture of the subduction interface and overlying faults Earth Planet Sci Lett 482: 44–51
Yue H and Lay T (2013) Source rupture models for the Mw 9.0 2011 Tohoku earthquake from joint inversions of high-rate geodetic and seismic data. Bull Seismol Soc Am 103(28): 1 242–1 255
Zhang H, Koper K D, Pankow K and Ge Z (2017) Imaging the 2016 MW7.8 Kaikoura, New Zealand, earthquake with teleseismic P waves: A cascading rupture across multiple faults. Geophys Res Lett 44(10): 4 790–4 798 doi: 10.1002/2017GL073461
Zhang Y, Xu LS, Chen YT, Feng WP and Du HL (2009) Source process of MS6.4 earthquake in Ning’er, Yunnan in 2007. Science China: Earth Sciences 52(2): 180–188 doi: 10.1007/s11430-009-0016-0
Zhu L and Rivera L (2002) A note on the dynamic and static displacements from a point source in multilayered media. Geophys J Int 148(3): 619–627
DownLoad:
