Tectonophysics 712–713 (2017) 1–9
Contents lists available at ScienceDirect
Tectonophysics journal homepage: www.elsevier.com/locate/tecto
An integrated analysis of source parameters, seismogenic structure, and seismic hazards related to the 2014 MS 6.3 Kangding earthquake, China Zujun Xie a, Yong Zheng b,⁎, Chengli Liu b, Bin Shan b, Muhammad Shahid Riaz a,c,d, Xiong Xiong a a
State Key Laboratory of Geodesy and Earth's Dynamics, Institute of Geodesy and Geophysics, CAS, Wuhan 430077, China State Key Laboratory of Geological Processes and Mineral Resources, Institute of Geophysics and Geomatics, China University of Geosciences, Wuhan, Hubei 430074, China University of Chinese Academy of Science, Beijing, China d Center for Earthquake Studies, NCP, Islamabad, Pakistan b c
a r t i c l e
i n f o
Article history: Received 7 July 2016 Received in revised form 26 April 2017 Accepted 28 April 2017 Available online 30 April 2017 Keywords: Kangding earthquake Source parameter Afterslip Seismic hazard
a b s t r a c t On 22 November, 2014, an MS 6.3 earthquake occurred in Kangding County, China. Focal mechanism solution shows that the two nodal planes were 235°/82°/−173° and 144°/83°/−8° and the focal depth was 9 km. Seismic slip of the Kangding earthquake was bilateral with about 0.5 m maximum slip. The rupture zone was confined to depths ranging from 5 to 15 km and laterally extended along the slip and strike directions by about 10 and 12 km, respectively. Most of the seismic moment was released in the first 5 s of the rupture, resulting in an earthquake magnitude of MW 6.01. In contrast, a slip model obtained by interferometric synthetic aperture radar (InSAR) data indicates that the rupture zone was longer than that determined from the seismic data and the earthquake magnitude should be about MW 6.2. Although accounting for the contribution of the MS 5.8 aftershock and the other small aftershocks that occurred during the InSAR observations period, the total moment estimated based on the seismic slip model was significantly smaller than that obtained from the InSAR data. Based on our analysis, we found that the inconsistency between the results determined from the seismic data and the InSAR data may be caused by the decrease in the shear modulus at shallow depths, the noise in the InSAR data, and the occurrence of some afterslips in the northwest region of the fault zone. The seismic slip of this earthquake was too small to release the accumulated energy within the entire Xianshuihe fault. We also found that the Coulomb stress in the northwest zone of the Kangding–Daofu seismic gap increased as a result of the historical, 2008 MS 8.0 Wenchuan and the 2014 MS 6.3 Kangding earthquakes, suggesting that this area is expected to be a high seismic hazard region for the future. © 2017 Elsevier B.V. All rights reserved.
1. Introduction An MS 6.3 earthquake struck Kangding County, Sichuan province, China at 08:55:00 UTC (16:54:35.7, Beijing local time) on 22 November 2014 (hereafter referred to as the Kangding earthquake). The earthquake epicenter was located at 101.69°E and 30.27°N (Yi et al., 2015) with a focal depth of 18 km, according to the China Earthquake Network Center (CENC, http://www.cea.gov.cn/publish/ dizhenj/464/479/index_8.html) and 9 km, according to the United States Geological Survey (USGS, http://comcat.cr.usgs.gov/ earthquakes/eventpage/usb000syy0#summary). This event was deemed one of the most destructive earthquakes that occurred along the eastern boundary of the Sichuan–Yunnan rhombic fault block, southwestern China. Three days later, a strong M S 5.8 aftershock occurred about 10 km away from the Kangding earthquake
⁎ Corresponding author. E-mail address:
[email protected] (Y. Zheng).
http://dx.doi.org/10.1016/j.tecto.2017.04.030 0040-1951/© 2017 Elsevier B.V. All rights reserved.
focal area, followed by approximately 1000 aftershocks in the following weeks. The NW–trending linear distribution of aftershocks recorded by a regional seismic network (Fig. 1) indicates that the seismogenic fault of the Kangding earthquake is the Selaha fault (Yi et al., 2015), which is the central branch of the Xianshuihe fault. The Xianshuihe fault splits into three branches by two cities, Bamei and Kangding. The other two faults are the north Yalahe and south Zheduotang faults (Fig. 1). The Xianshuihe, Anninghe, Zemuhe, and Xiaojiang faults constitute the eastern boundary of the Sichuan–Yunnan rhombic block (Wen et al., 2008) (Fig. 1). The Xianshuihe– Xiaojiang fault (hereafter referred to as the XXF) system has experienced large-scale, left-lateral strike-slip motions since the Holocene Epoch. Observations of geomorphic offset of Quaternary landforms indicate slip rates of about 15 ± 5 mm/a in the northwestern segment and about 9.6 ± 1.7 mm/a in the southeastern segment of the Xianshuihe fault. Global Positioning System (GPS) observations reveal a current slip rate of about 10 mm/a along the entire Xianshuihe fault (Allen et al., 1991; Shan et al., 2013a; Shen et al., 2005; Zhang, 2013). Such a high slip rate results in large strain accumulation
2
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
Fig. 1. Seismogenic environment of the Kangding earthquake and historical earthquakes that occurred on the Xianshuihe fault. The red star represents the epicenter of the Kangding earthquake (Yi et al., 2015). The solid black lines represent the active faults, and the solid colored line represents the slip rate area of the Xianshuihe fault. The focal mechanisms of historical earthquakes were taken from Shan et al. (2013a). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
along the fault zone, which is thus capable of generating strong earthquakes. The seismogenic environment of the Xianshuihe-Xiaojiang fault system makes it seismically active. N35 events with magnitude M N 6.0 transpired along this fault system since 1700, among them there were six events that were concentrated in a segment between Bamei and Kangding (Jiang et al., 2015a; Shan et al., 2013a; Wen et al., 2008). The 1973 M7.5 Luhuo event was the strongest earthquake to have occurred on the XXF thus far. However, strong earthquakes are expected to occur if a slip deficit is present. Moreover, as a result of the 2008 MS 8.0 Wenchuan and 2013 MS 7.0 Lushan earthquakes, which occurred along the Longmenshan fault zone which is close to the Xianshuihe fault, the Coulomb stress (ΔCFS) increased in the segment of the Xianshuihe fault between the cities of Daofu and Kangding (Shan et al., 2009, 2013a, 2013b; Shao et al., 2010; Toda et al., 2008). Historical seismic records (Wen et al., 2008), b-value distributions (Yi et al., 2008), geodetic imaging (Jiang et al., 2015b) and the estimated ΔCFS (Shan et al., 2009, 2013a, 2013b) indicate that the possibility for the occurrence of moderate-to-strong earthquakes along the 135 km long segment of the Xianshuihe fault has increased. The Kangding earthquake was the first moderate-to-strong earthquake occurred along the Xianshuihe fault in the last 40 years. Some studies have been carried out to ascertain source parameters (Fang et al., 2015) and a seismic slip model (Jiang et al., 2015a) for the Kangding earthquake. However, the source parameters determined from the seismic data significantly differ from those obtained from the geodetic data. For example, the earthquake magnitude determined from the seismic data by Yi et al. (2015) and Fang et al. (2015) is about MW 6.0, whereas study based on interferometric synthetic aperture radar (InSAR) data gives a value of about MW 6.2 (Jiang et al., 2015a). Significant difference in released energy between the geodetic and seismic models is still present after taking into account the contribution of the MS 5.8 aftershock and other small aftershocks that occurred during the observation period of InSAR data. It is important to study the causes of the difference
between the two models, as this will affect investigation of the afterslip distribution and evaluation of potential hazards in the source region. Moreover, there is a lack of a thorough analysis of ΔCFS and seismic hazard because this event has relatively low magnitude, thereby requiring further investigation on the source parameters, post-seismic deformation and seismic hazard in the source region. In this work, the focal mechanism solutions (FMSs) and the centroid depths of the mainshock and the largest aftershock were obtained using the cut and paste (CAP) method (Zhao and Helmberger, 1994; Zhu and Helmberger, 1996). A slip model for the Kangding earthquake was further derived by joint inversion using regional and teleseismic waveform data. We compared the present slip model with a previous model derived from geodetic data and analyze various factors that could cause the differences between the two models. Moreover, the present study generated a unified slip model by joint inversion of geodetic and seismic data. Based on the inversion results, we calculated the static Coulomb stress around the Xianshuihe fault and further discussed seismic hazards in the Xianshuihe fault system. 2. Focal mechanism of the Kangding earthquake FMS and centroid depth are critically important for understanding the seismogenic structure and properties of a rupture fault. Yi et al. (2015) derived the FMS and centroid depth of the Kangding earthquake using the generalized cut and paste (gCAP) method (Zhu and Ben-Zoin, 2013). The FMSs obtained by Yi et al. (2015) were 234°/81°/− 172° (representing strike/dip/rake) and 143°/82°/− 9°, with a centroid depth of about 9 km. However, the magnitude and the centroid focal depth differed significantly from those obtained from the geodetic data (Jiang et al., 2015a). Therefore, the accuracy and reliability of these results need to be re-evaluated. Because of the complexity of the geological structure in the source region, particularly in the zone between the Songpan–Ganzi Block and the Sichuan Basin, the simplified one-dimensional (1D) crustal
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
structure in the inversion of Yi et al. (2015) may introduce uncertainty into their results. Based on Xie et al. (2013), we selected three different crustal velocity models for the northwestern Bayankala block, the northeastern Sichuan Basin and the southwestern Sichuan–Yunnan block. The results show an uncertainty of FMS within 3° with only 1 km fluctuation of centroid depth, which indicate that the lateral heterogeneity of the crustal structure does not significantly affect the model results due to the application of the near-field data which had good azimuthal coverage and the use of the gCAP method. Therefore, the inverted focal mechanism and depth values are considered reliable. Considering the strong heterogeneity in the source region, we consolidated the crustal models presented by Xu et al. (2007), Zheng et al. (2013), Lei et al. (2014) and Liu et al. (2014) to build an average crustal model (Fig. 2). On the basis of this averaged model, Green's functions were calculated using the frequency–wavenumber spectrum (F–K) method (Zhu and Rivera, 2002) for each regional seismic station (Zheng et al., 2009, 2010). The best waveform match using the CAP method was achieved for the event at a focal depth of 9 km, with a magnitude of MW
3
6.0 (Fig. 2). Cross-correlation coefficients between the observed and synthetic seismograms calculated from the best-fit model were very high. Most coefficients for the Pnl phase were N0.9 and for other phases were N0.8 (Fig. 3), which suggest that the source parameters were well-resolved. The FMSs of the two nodal planes were 235°/82°/−173° and 144°/83°/−8°, which are close to the results obtained by Yi et al. (2015). Similarly, the FMSs for the MS 5.8 aftershock were 242°/81°/− 171° and 151°/81°/−9° with a centroid depth of 9 km, which is consistent with the results of Yi et al. (2015) (242°/84°/−173° and 151°/83°/−6°). The robustness of FMS based on different crustal models and the consistency between our results and those of Yi et al. (2015) thus demonstrate the reliability of the FMSs of the Kangding earthquake and its aftershock. Since the mainshock and the largest aftershock have different locations and FMSs, they could occur in different seismotectonic zones. The mainshock was likely to have occurred on the Selaha fault, whereas the aftershock probably took place along the Zheduotang fault (Yi et al., 2015).
Fig. 2. a) One-dimensional velocity model used for waveform data inversion. The dashed red and green lines represent S- and P-wave velocities, respectively. b) Distribution of the teleseismic stations and c) regional broadband seismic stations. The white star in (b) and the red star in (c) represent the epicenter of the Kangding earthquake (Yi et al., 2015). The black inverted triangles in (b) and the blue ones in (c) display the locations of seismic stations. d) Fit error between observed and synthetic seismograms, and focal mechanisms are shown to change with focal depth. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
4
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
Fig. 3. Cut-and-paste (CAP) inversion results including the focal mechanism and a comparison of the observed (black lines) and synthetic waveforms (red lines). The top line illustrates one fault plane of the earthquake, moment magnitude, and fit error. The beachball drawn in the lower hemisphere projection shows the focal mechanism of the event. The first column describes the distance, name, and azimuth of each station. The numbers in the other five columns represent the time shifts (upper) and cross-correlation coefficients in percentages (lower). Positive time shifts mean that the synthetic seismograms were delayed to the observed data and vice versa. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3. Seismic slip of the Kangding earthquake and discrepancy between seismic and geodetic models 3.1. Seismic slip of the Kangding earthquake Although FMS can provide general information about an earthquake, it does not precisely reveal the detailed seismogenic structure, thereby challenging the FMS-based seismic hazard evaluations. However, a source slip model can provide detailed information about the slip pattern and a major slip patch of an earthquake which can be further applied to ΔCFS calculations to evaluate impending seismic hazards (Shan et al., 2013a; Stein, 1999). Although the magnitude of the Kangding earthquake was relatively small, its rupture duration presented a distinctive characteristic. Teleseismic data can be employed to obtain a slip model, but the spatial resolution of the model constructed from teleseismic data alone is poor due to wave propagation effects. Therefore, the use of near-field seismic records is essential in improving the resolution and precision of a slip model (Liu et al., 2015). Although near field strong motion data were not collected, the seismic data from dense regional broadband seismic stations with a good azimuthal coverage were available (Fig. 2c). Most of the regional seismic records were not clipped due to the relatively small magnitude of the earthquake, so the unclipped seismograms were employed in the inversion. Regional records were, therefore, more precise and provided good constraints to generate a
slip model, complementary to the teleseismic body waves. We, thus, jointly inverted the regional and teleseismic waveform data to infer a slip model for the Kangding earthquake (Fig. 2). Teleseismic P waveforms from 22 broadband seismic stations and teleseismic SH waveforms from 14 stations were used (Fig. 2b). The distribution of the stations produced good azimuthal coverage around the earthquake epicenter. High signal-to-noise (SNR) ratio waveforms were obtained from the Incorporated Research Institutions for Seismology (IRIS) data center. Initially, raw records (Fig. 2c) were deconvolved with instrument responses to obtain displacement. Seismograms were then bandpass filtered at a frequency band of 0.003–1 Hz. Similarly, instrument responses were removed from the regional seismic data and then integrated to displacement and were further filtered at a frequency band of 0.02–0.5 Hz. In consideration of the tectonic setting, the FMS, and the aftershock distribution, we designed a fault model consisting of a single plane with a spatial scale of 26 km along the strike (142.36°) and 20 km along the down-dip (82°), which were equivalent to those used by Jiang et al. (2015a). The fault plane was discretized into 130 subfaults with spatial dimensions of 2.0 km × 2.0 km. In the inversion, the search ranges for the model parameters were (a) slip value for each sub-fault: 0–2 m, (b) rake angle: −33°–27° in steps of 2°, (c) average rupture velocity: 1.0–3.5 km/s in steps of 0.1 km/s and rise time: 0.4–2.4 s in steps of 0.4 s. On the basis of the earthquake relocation result reported by Yi et al. (2015), the depth of hypocenter was initially set to 14 km and
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
allowed to vary from 0 to 30 km until the best-fit model was achieved by a trial-and-error method. Checkerboard test results (Fig. S1) indicate the applicability of the joint inversion in generating a relatively higher resolution slip model for the Kangding earthquake. We adopted a finite fault inversion method in the wavelet domain and a simulated annealing method to simultaneously invert the slip amplitude, rake angle, rise time and average rupture velocity (Ji et al., 2002; Liu et al., 2013, 2015). The joint inversion-derived optimal model for the Kangding earthquake is shown in Fig. 4. The waveform fits of all the datasets are presented in Figs. S2 and S3. Our results indicated that rupture initiated at a depth of 11 km and laterally extended along the dip and strike directions by 10 and 12 km, respectively. The major slips concentrated near the hypocenter at depths ranging from about 5–15 km with a maximum slip of 0.5 m. Based on the distribution of the slip model, the inferred centroid depth is about 9 km, which is consistent with the result obtained by the CAP method. Most of the seismic moment was released during the first 5 s after the onset of the seismic slip. 3.2. Discrepancy between seismic and geodetic slip models Our slip model significantly differs from the one derived from the InSAR data (Jiang et al., 2015a). The magnitude in our result is about MW 6.01, which is different from the value of MW 6.20 obtained by Jiang et al. (2015a). The depth of the major slip zone is shallower in the geodetic model as compared to the seismic model, as shown in
5
Fig. 4a. Moreover, the InSAR model indicates significant seismic slip on the northwest side of the rupture zone which is not shown in the seismic model. In order to understand the causes for these differences, we investigated the source parameters of the MS 5.8 aftershock and compared the total seismic slips of the mainshock and the strong aftershock with those of the InSAR data-derived. Because the MS 5.8 earthquake was relatively small, it was difficult to obtain a detailed slip model. We therefore determined the average slip for the entire rupture area by using an empirical relationship (Wells and Coppersmith, 1994). Although the MS 5.8 aftershock occurred on a fault different from that of the mainshock, they were close to each other and the strikes of their rupture faults were parallel. The ground displacement caused by the MS 5.8 earthquake was therefore merged with that of the Kangding earthquake and was detected by InSAR imagery. The CAP method-derived results exhibited a magnitude of MW 5.57 for the MS 5.8 event, with an average rupture slip of about 8 km along the strike and 4.5 km along the dip (Bonilla et al., 1984). According to the slip model of the mainshock and the average slip model of the MS 5.8 earthquake, the total moment was still significantly smaller than that obtained from the InSAR data. Although other aftershocks that occurred within the succeeding two weeks following the mainshock could also contribute to the displacements measured by InSAR, their significantly small magnitudes did not account for this difference. The contribution of aftershocks in the displacements can be neglected since the maximum magnitude of the aftershocks was Ms
Fig. 4. (a) Seismic slip model, where the direction of the fault plane is indicated by a long, black arrow on the top of the figure, which is same as reported by Jiang et al. (2015a). The red star indicates the hypocenter of the mainshock, colour bar scales indicate the slip amplitude, white arrows represent the slip vector, and contours indicate rupture initiation time in seconds. The gray circles show relocated aftershocks (Yi et al., 2015) with the size of the circle proportional to the magnitude. (b) Slip model determined on the basis of the same InSAR data as used by Jiang et al. (2015a) using a stratified crustal model. (c) Horizontal projection of co-seismic slip model for the Kangding earthquake and distribution of aftershocks. The gray (red) filled circles represent aftershocks that occurred before (after) the MS 5.8 aftershock. (d) Distribution of the seismic slip and aftershocks. The gray filled circles represent the aftershocks. The red stars show the epicenters of the Kangding earthquake and the MS 5.8 aftershock. The black dashed circle outlines the slip zone of the Kangding earthquake with 10 cm seismic slip, and the black dashed rectangle outlines the possible average rupture area of the MS 5.8 aftershock. (e) Moment-rate function of this earthquake. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
6
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
3.3. A deficit in moment magnitude equivalent to MW 5.9 exists between the seismic and InSAR models. 3.3. Possible reasons for the discrepancy between seismic and geodetic models To quantitatively analyze the causes of the large difference between the seismic and geodetic models, we analyzed several factors that may generate misfits in the inversion of the slip models. Firstly, we calculated the synthetic Advanced Land Observing Satellite (ALOS) interferogram using our seismic slip model (Fig. S4). The synthetic interferogram agreed with the interferogram observed by InSAR very well, with a misfit of about 1 cm. The main difference was observed on the northwest side relative to the epicenter, which showed a possible consistency with the aftershock distribution in that region. A difference of 1 cm is slightly higher than the noise level in the InSAR data and could, therefore, be neglected. The difference between the seismic and geodetic slip models may have been partially caused by the noise in the InSAR data. Secondly, to clarify the larger magnitude determined by the geodetic data, we compared the synthetic seismograms calculated from a geodetic slip model (Fig. 4b) derived using the same InSAR data as used by Jiang et al. (2015a) with the observed seismograms. As shown in Fig. S5, although the amplitudes of the synthetic teleseismic seismograms were slightly larger than those of the observed seismograms, most local seismograms agreed reasonably well. This indicated that the InSAR model could predict the observed seismograms quite well, although the synthetic data had slightly larger amplitudes. Given a 0.2 magnitude difference exists between the seismic and geodetic models, the amplitudes of the synthetic waveforms were expected to be doubly relative to the observed seismograms. However, the synthetic seismograms exhibited amplitudes that were only about 1.2–1.3 times of those of the observed data. Therefore, the difference in magnitude was deemed not so significant. Three possible factors may have resulted in the large difference in magnitude. Firstly, the crustal structure model may amplify the magnitude of the slip model. In the model presented by Jiang et al. (2015a), a constant shear modulus value of 30 GPa was used in the estimation of the magnitude. This value may be too high for the Kangding earthquake as it was quite shallow. Because rigidity increases with depth, we modified the model rigidity based on the stratified crust model used for the co-seismic slip modeling, which exhibited a decrease in the modified moment magnitude from MW 6.20 to MW 6.14. Therefore, shallow crustal velocity structure may significantly impact the magnitude determination of the shallow earthquakes. We performed a joint inversion by seismic and InSAR data to obtain an integrated InSAR and seismic model, as shown in Fig. S6. In this model, the magnitude was about MW 6.10 with a centroid depth at about 10 km, and the rupture was concentrated along the dip direction at depths ranging from 5 to 12 km. The rupture slip was in good agreement with the distribution of aftershocks. The second reason may be related to the uncertainty of the InSAR data. In view of the probable unreliability of the small slip below 20 cm in the geodetic slip model, we compared the synthetic seismic waves based on a new geodetic slip model created by simply removing the slip below 10 cm and on the seismic slip model (Fig. S7). The comparison revealed that synthetic waves obtained from the modified InSAR model were not clearly larger than those derived by the seismic slip model, which suggests that the discrepancy of magnitude between the two models could be due to the uncertainty of the InSAR data. The third reason may be related to a possible afterslip. As shown in Fig. S5, the amplitudes of the synthetic waveforms calculated from the geodetic slip model were slightly larger than those of the observed seismograms, particularly notable on the northwest side of the epicenter. Although the amplitudes of the synthetic seismograms, calculated from the modified model after removing the small scale slips, closely
resembled those of the observed seismograms, as shown in Fig. S7, these small slips should represent some scale of the rupture in the fault given their systematic occurrence in the slip model. Considering these factors, some afterslip may have been generated during the InSAR observation period, though its equivalent magnitude would surely be less than MW 5.9. A comparison of the distribution of the two slip models (Fig. 4) exhibits main afterslip concentrations in the northwest area of the rupture fault and in the surrounding areas of the major slip zone of the mainshock. However, no obvious afterslips were observed in the southeast region of the co-seismic slip zone. The distribution of the aftershocks could be explained by the afterslip (Fig. 4a). The aftershock zone in the seismogenic fault was significantly larger than the rupture zone of the mainshock, which was outlined with the dashed line in Fig. 4d. Although no significant co-seismic slip appeared in these areas, many aftershocks were observed specifically on the northwest side and at the shallower depths, where the afterslip was most obvious. Almost no early aftershocks occurred on the southeast of the epicenter of the mainshock, and the afterslip in that area was also negligible. Although a number of aftershocks transpired following the MS 5.8 aftershock, most of them occurred on a different fault and were located within the possible rupture zone of the MS 5.8 event (Fig. 4c), thereby indicating a small afterslip size in the area surrounding the MS 5.8 earthquake. 4. ΔCFS and seismic hazard assessment Following the methodology of Shan et al. (2013a), we calculated the ΔCFS evolution around the Kangding earthquake area, including the stress accumulation caused by historical earthquakes and the stress perturbation on the Xianshuihe fault induced by the Kangding earthquake. We then analyzed the possible relationship between the mainshock and its aftershocks. The velocity model used in this work is shown in Fig. 2 and the physical parameters modified from Shan et al. (2013a) were summarized in Table 1. The effective friction coefficient, which minimally influences the numerical result, was set at a moderate value of 0.4 (Shan et al., 2013a; Stein, 1999). To estimate the viscoelastic ΔCFS caused by historical earthquakes on the rupture plane of the Kangding earthquake, the 144°/83°/−8° nodal plane was selected as the receiver fault. Based on the parameters and the slip model determined in section 2 and 3.1, we calculated the co-seismic static ΔCFS caused by the Kangding earthquake. Shan et al. (2013a) analyzed the ΔCFS evolution caused by historical earthquakes in the XXF zone. Until 2010, the maximum co-seismic ΔCFS caused by the Wenchuan earthquake was larger than the threshold value (0.01 MPa) at the epicenter of the Kangding earthquake (Shan et al., 2009). Moreover, the stress accumulation in this segment of the Xianshuihe fault is still increasing due to post-seismic stress transfer. We, therefore, calculated the time-dependent ΔCFS evolution process on the rupture fault of the Kangding earthquake after the occurrence of the Wenchuan and Lushan earthquakes. Sensitivity of the results to the effective frictional coefficient was also investigated. Fig. 5 shows ΔCFS in 2008 immediately after the Wenchuan earthquake and the Table 1 Crust and mantle parameters used in this study (from Shan et al., 2013a). Crustal thickness (km)
Density Vs (km·s−1) (g·cm−3)
Shear modulusa (×1010 Pa)
Viscosity coefficient (Pa·s)
0–2 2–12 12–20 20–31 31–43 Upper mantle
2.83 3.52 3.28 3.58 3.70 4.35
2.1 3.6 3.1 3.7 4.1 6.3
∞ ∞ ∞ 1.0 × 1019 1.0 × 1019 1.0 × 1020
2.65 2.75 2.88 2.91 3.10 3.35
a The shear modulus was derived by multiplying the density by the square of Vs (Aki and Richards, 2002).
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
7
Fig. 5. Coulomb stress changes (a) after the 2008 Wenchuan earthquake and (b) immediately before the Kangding earthquake. (c) Time-dependent stress evolution at the hypocenter of the 2014 Kangding earthquake with different effective friction coefficient values.
combined ΔCFS in 2014 caused by the Wenchuan and Lushan earthquakes on the nodal plane of the Kangding earthquake. Stress accumulation on the Kangding–Daofu segment, which is close to the epicenter of the Kangding earthquake, was enhanced by the 2008 Wenchuan earthquake and is continuously increasing (Shan et al., 2009, 2013b). ΔCFS caused by post-seismic relaxation is almost equivalent to co-seismic ΔCFS (Fig. 5a and b), indicating that the time-dependent stress transfer caused by visco-elastic relaxation is important. A ΔCFS increase of N0.02 MPa was observed in the source region of the Kangding earthquake, which is significantly larger than the threshold value for earthquake triggering (Stein, 1999). Compared with the Wenchuan earthquake, the influence of the Lushan earthquake was very weak. To examine the influence of the effective frictional coefficient on the result, we examined different coefficient values. As shown in Fig. 5c, some remarkable changes were observed in the calculation of ΔCFS with different effective frictional coefficient values. Higher frictional coefficients generated greater stress increases in the hypocenter of the Kangding earthquake. However, the ΔCFS in the hypocenter was consistently larger than 0.01 MPa regardless of the coefficient value adopted. Thus it is reasonable to conclude that the Kangding earthquake was likely to be triggered by the Wenchuan earthquake. Furthermore, we investigated ΔCFS induced by the Kangding earthquake on the Xianshuihe fault and in the surrounding areas. The Kangding earthquake occurred in a seismic gap along the Kangding–
Daofu segment (Fig. 6a), as proposed by Wen et al. (2008). In this segment, the ΔCFS increment caused by the Wenchuan earthquake (shown with a red curve in Fig. 6a) exhibited the highest value of the entire seismic gap on the Xianshuihe fault. After the Kangding earthquake, the southeastern part of the gap partly unloaded, whereas the entire northwestern segment was stressed by over 0.01 MPa with a maximum value of ~ 0.02 MPa (shown with a blue curve in Fig. 6a). The combined effect from the Wenchuan and Kangding earthquakes may increase the likelihood of forthcoming seismic hazards in this seismic gap. After the Kangding mainshock, numerous aftershocks occurred in the area surrounding the source region. The aftershocks that occurred between the mainshock and MS 5.8 aftershock were mostly concentrated in the ΔCFS-enhanced region, implying that the aftershocks might be triggered by the mainshock as shown in Figs. 4c and 6b. Moreover, the MS 5.8 aftershock was also located in the ΔCFS-enhanced zone and might be triggered by the mainshock. However, few aftershocks occurred in the southeast region of the mainshock, though the ΔCFS of this region was also enhanced by the Kangding earthquake. After the MS 5.8 earthquake, many aftershocks were struck in the surrounding areas. These later aftershocks might thus be triggered by the MS 5.8 aftershock rather than by the mainshock. The rupture fault split into two branches near the southeast end of the epicenter of the mainshock (Fig. 6b). Early aftershocks occurred on the same fault as the mainshock,
Fig. 6. (a) Coulomb stress along the Xianshuihe fault caused by the Kangding earthquake. The red (blue) line represents the CFS change induced by the Wenchuan (Kangding) earthquake. The red star represents the location of the 2014 Kangding earthquake on the Xianshuihe fault. The colored histograms denote the estimated rupture scales of historical earthquakes and are labeled on top with year of occurrence. (b) Coulomb stress caused by the co-seismic rupture of the Kangding earthquake. Green stars represent the epicenters of the Kangding earthquake and the MS 5.8 aftershocks, gray dots represent the aftershocks and the colour scale indicates the magnitude of Coulomb stress. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
8
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9
whereas the MS 5.8 aftershock and later aftershocks occurred mostly on the other branched fault. Few aftershocks and no obvious afterslips were observed on the southeast side of the mainshock seismogenic fault. An increase of the ΔCFS was observed in both branches of the fault. If the two branches were separated from each other, the seismogenic potential on the southeast side of the mainshock's rupture fault should increase. This hypothesis requires further examination, which goes beyond the scope of current study.
on the southeast side, the Zheduotang fault, which was also struck by the MS 5.8 earthquake of this sequence (Yi et al., 2015), experienced an MS 7.5 earthquake in 1955, thereby resulting in a relatively low probability of earthquake occurrence in this region. On the contrary, the northwest side of the Kangding–Daofu segment and the northwest section of the Selaha fault were exhibited higher changes of experiencing moderate-to-strong earthquakes. Acknowledgements
5. Discussion and conclusion We obtained FMS and centroid depth of the Kangding earthquake. By considering the tectonic environment of the source region and the distribution of aftershocks, we determined the Kangding earthquake FMS to be 144°/83°/− 8° for strike/dip/rake at a centroid focal depth of ~ 9 km, and a magnitude of MW 6.0. The causative fault of the Kangding earthquake was the NW–SE Selaha fault. The slip model for the Kangding earthquake was investigated by conducting a joint inversion using the teleseismic waveform data and regional seismograms. Seismic slips were mainly concentrated at depths ranging between about 5 km and 15 km, having a maximum slip amplitude of 0.5 m at a depth of 9 km. The rupture area spanned ~ 10 km along the dip and ~ 12 km along the strike, and the rupture was bilateral from the hypocenter. The moment magnitude was MW 6.01 and most of the seismic moment was released during the first 5 s. Obvious differences between the result of our co-seismic slip model and that of the slip model determined from the InSAR data (Jiang et al., 2015a) were noted. The seismic moment and rupture zone of the latter model were larger than those of our model. Through quantitative analysis of possible factors that could affect the result determined from the geodetic data, we found that the magnitude of in the InSAR model was artificially amplified by the crustal structures, particularly by the shallow shear wave velocity and shear modulus. The InSAR data noise may have also resulted in the amplification of the geodetic model. A revised crustal model revealed that the moment magnitude of the geodetic model was about MW 6.14. An integrated analysis of the InSAR and seismic data revealed that the moment magnitude of the Kangding earthquake was about MW 6.10. The magnitude difference between the geodetic and the seismic models was reduced to about 0.1. A 0.1 difference in magnitude between the geodetic and seismic models likely resulted from the aftershocks and afterslip during InSAR measurement. Afterslips were mainly observed on the side northwest of the epicenter of the mainshock and in the shallower part of the coseismic zone of the rupture fault. No obvious afterslips and aftershocks transpired on the side southeast of the rupture zone, thereby implying the presence of a possible seismic barrier in the co–seismic rupture zone of the rupture fault on the southeast side. Relocation results determined the Kangding earthquake to have occurred on the Selaha fault, which is the central branch of the Xianshuihe fault between two cities, Kangding and Daofu. The MS 5.8 aftershock occurred on the Zheduotang fault, which is the southern branch fault. The length of the Selaha fault is about 40 km (Yang et al., 2015), nearly twice the rupture length of the Kangding earthquake. A seismic gap is present in the Kangding–Daofu segment (Wen et al., 2008). The length scale of the partially coupled asperity is likely to be up to 140 km, which is capable of generating an earthquake of about MW 6.6 (Jiang et al., 2015b; Yi et al., 2015). Therefore the seismic slip of the Kangding earthquake was not large enough to completely release the accumulated strain on the Xianshuihe fault segment, nor on the Selaha fault. Therefore, the likelihood of a moderate-to-strong earthquake occurring in this region is still high. ΔCFS caused by historical earthquakes and the Wenchuan earthquake significantly increased the stress accumulation in the Kangding–Daofu segment. Moreover, the Kangding earthquake-derived ΔCFS resulted in an increase in the stress accumulation on the northwest and southeast sections of the Kangding–Daofu segment. However,
The waveform data presented in this study were provided by the Data Management Centre of China National Seismic Network at Institute of Geophysics (doi:10.11998/SeisDmc/SN), China National Seismic Network, China Earthquake Administration. Teleseismic data were provided by the Incorporated Research Institutions for Seismology Data Management Center (IRIS-DMC). The authors thank G. X. Yi and G. Y. Jiang for sharing their relocation results and InSAR data and Prof. R. Wang and Prof. M. H. Ritzwoller for helping us to improve the quality of this work. All figures were drawn by GMT. This work is supported by the Natural Science Foundation of China (Grant nos. 41574057, 41422401, 41321063 and 41274104) and by a scientific grant from the Chinese Earthquake Administration (Grant no. 2016CESE0204). Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.tecto.2017.04.030. References Aki, K., Richards, P.G., 2002. Quantitative Seismology. second ed. University Science Books, California, USA (73 pp.). Allen, C.R., Lou, Z.L., Qian, H., et al., 1991. Field study of a highly active fault zone: the Xianshuihe fault of Southwestern China. Geol. Soc. Am. Bull. 103 (9), 1178–1199. Bonilla, M.G., Mark, R.K., Lienkaemper, J.J., 1984. Statistical relations among earthquake magnitude, surface rupture length, and surface fault displacement. Bull. Seismol. Soc. Am. 74 (6), 2379–2411. Fang, L., Wu, J., Liu, J., et al., 2015. Preliminary report on the 22 November 2014 Mw 6.1/ Ms 6.3 Kangding Earthquake, Western Sichuan, China[J]. Seismol. Res. Lett. 86 (6). Ji, C., Wald, D.J., Helmberger, D.V., 2002. Source description of the 1999 Hector Mine, California, earthquake, partI: wavelet domain inversion theory and resolution analysis. Bull. Seismol. Soc. Am. 92, 1192–1207. Jiang, G.Y., Xu, X.W., Chen, G., et al., 2015b. Geodetic imaging of potential seismogenic asperities on the Xianshuihe–Anninghe–Zemuhe fault system, southwest China, with a new 3D viscoelastic intersiesmic coupling model. J. Geophys. Res. Solid Earth http:// dx.doi.org/10.1002/2014JB011492. Lei, J., Li, Y., Xie, F.R., et al., 2014. Pn anisotropic tomography and dynamics under eastern Tibetan plateau. J. Geophys. Res. Solid Earth 119. http://dx.doi.org/10.1002/ 2013JB010847,2174-2198. Liu, Q.Y., van der Hilst, Robert D., Li, Y., et al., 2014. Eastward expansion of the Tibetan Plateau by crustal flow and strain partitioning across faults. Nat. Geosci. http://dx.doi. org/10.1038/NGEO2130,361-365. Liu, C.L., Zheng, Y., Ge, C., et al., 2013. Rupture process of the Ms7.0 Lushan earthquake, 2013. Sci. China Earth Sci. 56 (7), 1187–1192. Liu, C.L., Zheng, Y., Xiong, X., et al., 2015. Kinematic rupture process of the 2014 Chile Mw8.1 earthquake constrained by strong-motion, GPS static offsets and teleseismic data. Geophys. J. Int. 202, 1137–1145. Shan, B., Xiong, X., Zheng, Y., et al., 2009. Stress changes on major faults caused by Mw7.9 Wenchuan earthquake May 12, 2008. Sci. China Ser. D Earth Sci. 52, 593–601. Shan, B., Xiong, X., Wang, R.J., et al., 2013a. Coulomb stress evolution along Xianshuihe– Xiaojiang fault system since 1713 and its interaction with Wenchuan earthquake, May 12, 2008. Earth Planet. Sci. Lett. 377–378, 199–210. Shan, B., Xiong, X., Zheng, Y., et al., 2013b. Stress changes on major faults caused by 2013 Lushan earthquake and its relationship with 2008 Wenchuan earthquake. Sci. China Earth Sci. 56, 1169–1176. Shao, Z.G., Zhou, L.Q., Jiang, C.S., et al., 2010. The impact of Wenchuan Ms8.0 earthquake on the seismic activity of surrounding faults. Chin. J. Geophys. 53 (8), 1784–1795 (in Chinese). Shen, Z.K., Lu, J., Wang, M., et al., 2005. Contemporary crustal deformation around the southeast borderland of the Tibetan Plateau. J. Geophys. Res. 110, B11409. http:// dx.doi.org/10.1029/2004JB003421. Stein, R.S., 1999. The role of stress transfer in earthquake occurrence. Nature 402, 605–609. Toda, S., Lin, J., Meghraoui, M., et al., 2008. 12 May 2008 M = 7.9 Wenchuan, China, earthquake calculated to increase failure stress and seismicity rate on three major fault systems. Geophys. Res. Lett. 35 (17):L17305. http://dx.doi.org/10.10029/ 2008GL034903.
Z. Xie et al. / Tectonophysics 712–713 (2017) 1–9 Wells, D.L., Coppersmith, J.J., 1994. New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement. Bull. Seismol. Soc. Am. 84 (4), 974–1002. Wen, X.Z., Ma, S.L., Xu, X.W., et al., 2008. Historical pattern and behavior of earthquake ruptures along the boundary of the Sichuan–Yunnan faulted–block, southwestern China. Phys. Earth Planet. Inter. 168, 16–36. Jiang, G.Y., Wen, Y.M., Liu, Y.J., et al., 2015a. Joint analysis of the 2014 Kangding, southwest China, earthquake sequence with seismicity relocation and InSAR inversion. Geophys. Res. Lett. 42, 3273–3281. Xie, Z.J., Jin, B.K., Zheng, Y., et al., 2013. Source parameters inversion of the 2013 Lushan earthquake by combining teleseismic waveforms and local seismograms. Sci. China Earth Sci. 56 (7), 1177–1186. Xu, L.L., Rondenay, S., van der Hilst, R.D., 2007. Structure of the crust beneath the southeastern Tibetan Plateau from teleseismic receiver functions. Phys. Earth Planet. Inter. 165, 176–193. Yang, W., Chang, J., Liu, J., et al., 2015. The Kangding earthquake swarm of November, 2014. Earth Sci. 28 (3), 197–207. Yi, G.X., Wen, X.Z., Su, Y.J., 2008. Study on the potential strong-earthquake risk for the eastern boundary of the Sichuan–Yunnan active faulted block, China. Chin. J. Geophys. 51 (6), 1719–1725 (in Chinese). Yi, G.X., Long, F., Wen, X.Z., et al., 2015. Seismogenic structure of the M6.3 Kangding earthquake sequence on 22 Nov. 2014, Southwestern China. Chin. J. Geophys. 58 (4), 1205–1219 (in Chinese).
9
Zhang, P.Z., 2013. A review of on active tectonics and deep crustal processes of the Western Sichuan region eastern margin of the Tibetan Plateau. Tectonophysics 584, 7–22. Zhao, L.S., Helmberger, D.V., 1994. Source estimation from broadband regional seismograms. BSSA 84 (1), 91–104. Zheng, X.F., Ouyang, B., Zhang, D.N., et al., 2009. Technical system construction of Data Backup Centre for China Seismograph Network and the data support to researches on the Wenchuan earthquake. Chin. J. Geophys. 52, 1412–1417 (in Chinese). Zheng, X.F., Yao, Z.X., Liang, J.H., Zheng, J., 2010. The role played and opportunities provided by IGP DMC of China National Seismic Network in Wenchuan earthquake disaster relief and researches. Bull. Seismol. Soc. Am. 100 (5B):2866–2872. http://dx.doi.org/ 10.1785/0120090257. Zheng, Y., Ge, C., Xie, Z.J., et al., 2013. Crustal and upper mantle structure and the deep seismogenic environment in the source areas of the Lushan earthquake and the Wenchuan earthquake. Sci. China Earth Sci. 56 (7), 1158–1168. Zhu, L.P., Rivera, 2002. A note on the dynamic and static displacements from a point source in multilayered media. Geophys. J. Int. 148, 619–627. Zhu, L.P., Ben-Zoin, Y., 2013. Parametrization of general seismic potency and moment ternsors for source inversion of seismic waveform data. Geophys. J. Int. 194 (2): 839–843. http://dx.doi.org/10.1093/gji/ggt137. Zhu, L.P., Helmberger, D.V., 1996. Advancement in source estimation techniques using broadband regional seismograms. BSSA 86 (5), 1634–1641.