The 2014 Mw 6.1 Ludian Earthquake: The Application of RADARSAT-2 SAR Interferometry and GPS for this Conjugated Ruptured Event

Although the Zhaotong–Ludian fault is a seismically active zone located in the boundary between the Sichuan–Yunnan block and the South China block, it has not experienced a large earthquake greater than Mw 7 since at least 1700. On 3 August, 2014, an Mw 6.1 earthquake (the Ludian earthquake) ruptured the Zhaotong active belt in Ludian County, Yunnan province, China. This earthquake was the largest earthquake recorded in the region since 2000, and it provides us with a unique opportunity to study the active tectonics in the region. The analysis of the aftershocks showed that two conjugate faults could have been involved in the event. We first used Global Positioning System (GPS) data and C-band RADARSAT-2 imagery to map the coseismic surface deformation. We then inverted the derived coseismic deformation for the slip distribution based on the constructed conjugate fault model. Finally, the coulomb failure stress due to the Ludian earthquake was estimated to investigate the potential seismic hazards in this region. Our investigations showed that the Ludian earthquake was mainly a bilateral rupture event. The major slip of the main shock was located at depths of 0–5 km, which is close but does not superpose with the aftershocks that are mostly located at depths of 5–20 km. Interestingly, the seismic moment released by the aftershocks (6.9 × 1018 N·m) was greater than that of the main shock (2.6 × 1018 N·m). This evidence suggests that the accumulated elastic strain at depths of 0–20 km could have been fully released by the Ludian earthquake and its subsequent aftershocks. Furthermore, our analysis of the coulomb failure stress changes due to the main shock showed that the aftershocks could be the result of dynamic triggering rather than static triggering.


Introduction
The northeastern Yunnan rift zone is a seismically active region, located along the western boundary of the Sichuan-Yunnan Block and the eastern margin of the Qinghai-Tibet Plateau. The crustal motion in this area is dominated by large-scale fan-shaped rotation, as revealed by the Global Positioning System (GPS) velocity field ( Figure 1) [1]. There are many tectonically active faults in Yunnan, e.g., In this study, we first mapped the surface deformation due to the Ludian earthquake using GPS and interferometric synthetic aperture radar (InSAR) data. We then analyzed the distribution of the aftershocks to determine the ruptured fault and its geometry. The coseismic slip distribution was inverted based on the mapped coseismic deformation, and the constructed fault model was obtained using the steepest-descent method (SDM) [15]. Finally, we investigated how the elastic strain was partitioned during the earthquake and evaluated the potential seismic risk in the region.

GPS Data and Analysis
The coseismic GPS data used in this study were provided by Wei et al. (2018) [17], consisting of data from 12 continuous GPS stations and 21 mobile GPS stations, which were remeasured after the earthquake. All the GPS stations are distributed within 200 km of the epicenter (Figure 2). The nearest station was about 26 km from the epicenter, and the farthest was 190 km.
We can see from Figure 2 that the coseismic deformation shows significant southward motion to the west of the Zhaotong-Ludian fault, whereas northward motion prevails to the east of the fault. This suggests that there were some left-lateral slip components during the earthquake, which is consistent with the focal mechanisms determined using seismic data. However, we note that the crust to the north of the epicenter tended to move eastward during the earthquake, as opposed to the crustal deformation to the south of the epicenter, which is dominated by westward motion. This deformation pattern is unlikely to have resulted from the left-lateral slip on the southwest-oriented Zhaotong-Ludian fault. Therefore, we propose that one more fault (oriented roughly to the W-E Figure 1. Tectonic setting of the study area. The red beach ball is the GCMT solution for the 2014 Ludian earthquake, and the green balls mark the locations of the Mw ≥ 6 historical earthquakes. The black lines are the regional active faults. The blue arrows are the crustal deformation from Zheng et al. [16]. Remote Sens. 2019, 11, x FOR PEER REVIEW 4 of 11 direction) may have formed a conjugate fault with the Zhaotong-Ludian fault, and could have been involved during the earthquake, to produce the deformation we analyzed.

InSAR Data and Analysis
As the GPS stations in this area are extremely sparse and located far from the epicenter, the finer features of the near-field coseismic displacement cannot be well captured. Fortunately, there are two SAR image scenes acquired before and after the earthquake by the RADARSAT-2 satellite (Figure 1), which allows us to image the near-field deformation and constrain the rupture kinematics.
The RADARSAT-2 synthetic aperture radar (SAR) images encompass the epicentral area and

GPS Data and Analysis
The coseismic GPS data used in this study were provided by Wei et al. (2018) [17], consisting of data from 12 continuous GPS stations and 21 mobile GPS stations, which were remeasured after the earthquake. All the GPS stations are distributed within 200 km of the epicenter (Figure 2). The nearest station was about 26 km from the epicenter, and the farthest was 190 km.
We can see from Figure 2 that the coseismic deformation shows significant southward motion to the west of the Zhaotong-Ludian fault, whereas northward motion prevails to the east of the fault. This suggests that there were some left-lateral slip components during the earthquake, which is consistent with the focal mechanisms determined using seismic data. However, we note that the crust to the north of the epicenter tended to move eastward during the earthquake, as opposed to the crustal deformation to the south of the epicenter, which is dominated by westward motion. This deformation pattern is unlikely to have resulted from the left-lateral slip on the southwest-oriented Zhaotong-Ludian fault. Therefore, we propose that one more fault (oriented roughly to the W-E direction) may have formed a conjugate fault with the Zhaotong-Ludian fault, and could have been involved during the earthquake, to produce the deformation we analyzed.

InSAR Data and Analysis
As the GPS stations in this area are extremely sparse and located far from the epicenter, the finer features of the near-field coseismic displacement cannot be well captured. Fortunately, there are two SAR image scenes acquired before and after the earthquake by the RADARSAT-2 satellite (Figure 1), which allows us to image the near-field deformation and constrain the rupture kinematics.
The RADARSAT-2 synthetic aperture radar (SAR) images encompass the epicentral area and were acquired in ascending XF0W2 beam mode: Extra Fine Mode. Extra-Fine Resolution Beam Mode nominally provides a similar swath width and incidence angle coverage as the Wide Fine Beam Mode, at even finer resolutions [18,19]. The image acquisition dates were 4 March, 2014 and 16 April, 2015, respectively. The perpendicular baseline of this interferogram is 49 m, and the temporal baseline is 408 days.
We used GAMMA software to process the images under the framework of the two-pass InSAR approach (e.g., [20,21]). A digital elevation model (DEM) with a resolution of 5 m was used to remove the topographic phase from the interferogram. The coregistration procedure was conducted with the help of the DEM, which improved the coherence for the image pair [22]. The interferogram was multi-looked by a factor of 2 in range and 2 in azimuth (9 m × 9 m). We used the adaptive filter method to filter the interferogram and remove high-frequency noise. In addition, a quadratic ramp was estimated and removed from the interferogram by masking out the deforming area [23][24][25], to consider the orbital error. And the detailed process was shown in Figure S1.
The coseismic deformation derived from InSAR is shown in Figure 2b. In a broad sense, the InSAR coseismic deformation extends approximately 45 km in the E-W direction and about 40 km in the N-S direction. We can see that the near-field displacements are well mapped, except for the central high-relief mountainous areas, due to the earthquake-associated landslides [26][27][28][29][30][31], building damage [32], strong topographic relief, and/or saturation of the fringe rate. From the InSAR displacement field, the ground surface moved toward the sensor, with an approximate maximum of 10 cm in the southeast, and away from the sensor, with the same maximum of 10 cm in the northwest. However, what was the true direction of the coseismic displacement?
When combining the analysis of the GPS observations and the focal mechanism solutions, more specific conjugate faults are inferred in Figure 2b. Because the direction of the near S-N fault or the Xiaohe-Baogunao fault, which was likely the seismogenic fault inferred by many researchers, was almost parallel to the flight direction of the satellite, it was difficult to capture the horizontal displacements of the S-N fault. However, the movement of the inferred W-E fault was still acquired Remote Sens. 2020, 12, 99 5 of 12 from InSAR. As shown in Figure 2b, we can see that the north side of the W-E fault moved eastwards and the other side moved to the west, ignoring the small vertical component (a nearly pure strike-slip earthquake is suggested by many of the focal mechanism solutions). Therefore, the InSAR observations also provide more strong evidence for the existence of the E-W fault, i.e., the Ludian earthquake is highly likely to have been a conjugate rupture event.

Coseismic Deformation Modeling
Modeling coseismic displacement fields can provide new insights into the seismotectonic setting and seismic hazards of a tectonically active region. The analysis of the seismic waveforms [6], aftershock distributions [8][9][10], and apparent source time functions (ASTFs) [5] of the 2014 Ludian earthquake all suggest that the seismogenic structure of the earthquake was quite complex. The aftershocks distribute along the conjugate faults [8][9][10] and show a reverse L-shape. An analysis of the ASTFs also shows that two intersected faults could be responsible for the Ludian earthquake [5]. Zhang et al. [13] also concluded that the Ludian earthquake was a complex conjugated ruptured event. Therefore, in this study, we constructed a fault model consisting of two conjugate faults to shed light on the kinematics of the Ludian earthquake. Details of the constructed fault model are provided in the next section.

Fault Geometry
The critical displacement in the epicenter was lost due to the serious decorrelation of the InSAR data and the lack of near-field GPS stations, which brings challenges to determining the location of the causality fault from the geodetic data. In this study, we attempted to use the distribution of the aftershocks to determine the fault geometry.
The 961 aftershocks, relocated by Wang et al. [10] (Figure 3) using the double-difference location method, were used to help determine the fault geometry. We can see that the aftershock sequences are distributed dominantly in the SSE and EW directions, and clearly form an "L" shape. What is more, the aftershock sequences are vertically distributed at depth. These findings suggest that two steeply dipping faults, possibly forming a conjugate fault, were responsible for the Ludian earthquake ( Figure 3).
Remote Sens. 2019, 11, x FOR PEER REVIEW 5 of 11 almost parallel to the flight direction of the satellite, it was difficult to capture the horizontal displacements of the S-N fault. However, the movement of the inferred W-E fault was still acquired from InSAR. As shown in Figure 2b, we can see that the north side of the W-E fault moved eastwards and the other side moved to the west, ignoring the small vertical component (a nearly pure strikeslip earthquake is suggested by many of the focal mechanism solutions). Therefore, the InSAR observations also provide more strong evidence for the existence of the E-W fault, i.e., the Ludian earthquake is highly likely to have been a conjugate rupture event.

Coseismic Deformation Modeling
Modeling coseismic displacement fields can provide new insights into the seismotectonic setting and seismic hazards of a tectonically active region. The analysis of the seismic waveforms [6], aftershock distributions [8][9][10], and apparent source time functions (ASTFs) [5] of the 2014 Ludian earthquake all suggest that the seismogenic structure of the earthquake was quite complex. The aftershocks distribute along the conjugate faults [8][9][10] and show a reverse L-shape. An analysis of the ASTFs also shows that two intersected faults could be responsible for the Ludian earthquake [5]. Zhang et al. [13] also concluded that the Ludian earthquake was a complex conjugated ruptured event. Therefore, in this study, we constructed a fault model consisting of two conjugate faults to shed light on the kinematics of the Ludian earthquake. Details of the constructed fault model are provided in the next section.

Fault Geometry
The critical displacement in the epicenter was lost due to the serious decorrelation of the InSAR data and the lack of near-field GPS stations, which brings challenges to determining the location of the causality fault from the geodetic data. In this study, we attempted to use the distribution of the aftershocks to determine the fault geometry.
The 961 aftershocks, relocated by Wang et al. [10] (Figure 3) using the double-difference location method, were used to help determine the fault geometry. We can see that the aftershock sequences are distributed dominantly in the SSE and EW directions, and clearly form an "L" shape. What is more, the aftershock sequences are vertically distributed at depth. These findings suggest that two steeply dipping faults, possibly forming a conjugate fault, were responsible for the Ludian earthquake ( Figure 3). We directly determined the seismogenic faults from the relocated aftershock distribution (Figure 3a), which consist of a roughly E-W oriented fault (hereafter F1) and a roughly S-N oriented fault (hereafter F2). The dip angles of the two faults were also calculated by the linear fit of the aftershocks in the vertical direction (Figure 3c,d). The widths and lengths of the two faults were enlarged along the strike and downdip directions (Figure 3). Detailed parameters of the two faults are listed in Table 2.

Coseismic Slip Distributions
We used the SDM [15] to perform the inversion process. A layered Earth model from CRUST1.0 [33] was set up to model the crustal structure. Based on the fault geometry determined in Section 3.1, the fault planes were discretized into 1 km × 1 km patches to resolve the slip distribution. During the inversion, the rake angles of F1 were constrained from 170 • to 185 • , and F2 was from −20 • to 20 • , according to previous studies [9,13]. In order to account for the high spatial correlation of the pixels and to expedite the modeling process, we attempted to down-sample the InSAR data ( Figure S2) using the quadtree method [23,34,35]. In addition, to overcome the problem of the non-uniqueness and instability of the inversion results, a smoothing constraint was applied to the slip distribution. An optimal smoothing factor was determined by analyzing the trade-off curve between data misfit and slip roughness, and a value of 0.11 was ultimately selected in this study ( Figure S3). Figure 4 shows the final predictions and fitting residuals, with the predicted horizontal and vertical displacements shown in Figure 5. And Figure 6 shows the optimal slip model determined in this study.  We directly determined the seismogenic faults from the relocated aftershock distribution ( Figure  3a), which consist of a roughly E-W oriented fault (hereafter F1) and a roughly S-N oriented fault (hereafter F2). The dip angles of the two faults were also calculated by the linear fit of the aftershocks in the vertical direction (Figures 3c and 3d). The widths and lengths of the two faults were enlarged along the strike and downdip directions (Figure 3). Detailed parameters of the two faults are listed in Table 2.

Coseismic Slip Distributions
We used the SDM [15] to perform the inversion process. A layered Earth model from CRUST1.0 [33] was set up to model the crustal structure. Based on the fault geometry determined in Section 3.1, the fault planes were discretized into 1 km × 1 km patches to resolve the slip distribution. During the inversion, the rake angles of F1 were constrained from 170° to 185°, and F2 was from −20° to 20°, according to previous studies [9,13]. In order to account for the high spatial correlation of the pixels and to expedite the modeling process, we attempted to down-sample the InSAR data ( Figure S2) using the quadtree method [23,34,35]. In addition, to overcome the problem of the non-uniqueness and instability of the inversion results, a smoothing constraint was applied to the slip distribution. An optimal smoothing factor was determined by analyzing the trade-off curve between data misfit and slip roughness, and a value of 0.11 was ultimately selected in this study ( Figure S3). Figure 6 shows the optimal slip modle determined in this study, with the predictions and fitting residuals shown in Figure 4. We also shown the predicted horizontal and vertical displacements in Figure 5. We can see from Figure 4 that the predictions from the optimal slip model fit well with the observed InSAR and GPS, and the root-mean-square (RMS) misfits are 1.2 cm and 0.2 cm, respectively. The correlation coefficient between the observations and the prediction is 97.1%. In addition, to further verify the reliability of the inversion, we calculated the 3D deformation to compare with the results of the field survey after the event. As shown in Figure 5, the area with severe deformation is concentrated in regions A (F1) and B, and especially region B (F2). According to the field survey [26,31], most of the landsides triggered by the earthquake were distributed in regions A We can see from Figure 4 that the predictions from the optimal slip model fit well with the observed InSAR and GPS, and the root-mean-square (RMS) misfits are 1.2 cm and 0.2 cm, respectively. The correlation coefficient between the observations and the prediction is 97.1%. In addition, to further verify the reliability of the inversion, we calculated the 3D deformation to compare with the results of the field survey after the event. As shown in Figure 5, the area with severe deformation is concentrated in regions A (F1) and B, and especially region B (F2). According to the field survey [26,31], most of the landsides triggered by the earthquake were distributed in regions A and B. Furthermore, in region B, the field investigation showed that the left plate was uplifted by about 0.3 m relative to the right plane [26], which is very similar to our result of 0.34 m. and B. Furthermore, in region B, the field investigation showed that the left plate was uplifted by about 0.3 m relative to the right plane [26], which is very similar to our result of 0.34 m.  As shown in Figure 6, we can see that the significant slips on F1 and F2 are both concentrated at shallow depths of 0-5 km, but there are some differences. F2 is dominated by left-lateral fault slip, which reaches up to 1.6 m. In contrast, the slip on F1 is characterized by right-lateral strike slip, with a maximum of 0.7 m. The seismic moment released by the fault slip on F2 was 1.9 × 10 18 N•m (73% of the total), which was 2.5-times greater than that released on F1(0.7 × 10 18 N•m, 27% of the total). This suggests that the Ludian earthquake was a predominately left-lateral strike-slip mechanism, which is consistent with the focal mechanism solution determined by the IEF, the USGS, and the GCMT Project. The total optimal slip model yielded a geodetic moment of 2.6 × 10 18 N•m, corresponding to an Mw 6.2 event, assuming a shear modulus of 30 GPa, which is comparable to that estimated from the seismological data (Table 1). The comparable seismic moment magnitudes determined from the geodetic modeling and seismic data demonstrate that the postseismic deformation was likely very limited within the eight months following the earthquake. and B. Furthermore, in region B, the field investigation showed that the left plate was uplifted by about 0.3 m relative to the right plane [26], which is very similar to our result of 0.34 m.  As shown in Figure 6, we can see that the significant slips on F1 and F2 are both concentrated at shallow depths of 0-5 km, but there are some differences. F2 is dominated by left-lateral fault slip, which reaches up to 1.6 m. In contrast, the slip on F1 is characterized by right-lateral strike slip, with a maximum of 0.7 m. The seismic moment released by the fault slip on F2 was 1.9 × 10 18 N•m (73% of the total), which was 2.5-times greater than that released on F1(0.7 × 10 18 N•m, 27% of the total). This suggests that the Ludian earthquake was a predominately left-lateral strike-slip mechanism, which is consistent with the focal mechanism solution determined by the IEF, the USGS, and the GCMT Project. The total optimal slip model yielded a geodetic moment of 2.6 × 10 18 N•m, corresponding to an Mw 6.2 event, assuming a shear modulus of 30 GPa, which is comparable to that estimated from the seismological data (Table 1). The comparable seismic moment magnitudes determined from the geodetic modeling and seismic data demonstrate that the postseismic deformation was likely very limited within the eight months following the earthquake. As shown in Figure 6, we can see that the significant slips on F1 and F2 are both concentrated at shallow depths of 0-5 km, but there are some differences. F2 is dominated by left-lateral fault slip, which reaches up to 1.6 m. In contrast, the slip on F1 is characterized by right-lateral strike slip, with a maximum of 0.7 m. The seismic moment released by the fault slip on F2 was 1.9 × 10 18 N·m (73% of the total), which was 2.5-times greater than that released on F1(0.7 × 10 18 N·m, 27% of the total). This suggests that the Ludian earthquake was a predominately left-lateral strike-slip mechanism, which is consistent with the focal mechanism solution determined by the IEF, the USGS, and the GCMT Project. The total optimal slip model yielded a geodetic moment of 2.6 × 10 18 N·m, corresponding to an Mw 6.2 event, assuming a shear modulus of 30 GPa, which is comparable to that estimated from the seismological data (Table 1). The comparable seismic moment magnitudes determined from the geodetic modeling and seismic data demonstrate that the postseismic deformation was likely very limited within the eight months following the earthquake.

Comparison With Previous Studies
Similar to the conclusions of this study, most of the seismic models have demonstrated that the slips were mainly concentrated on the F2 or approximately F2 fault plane [13,36], and the released seismic moment was 62% on F2 and 38% on F1 [13], although differences exist in the slip magnitude and range. Based on the regional broadband seismograms, Liu et al. [36] showed that the F2 fault released most of the moment magnitude, the location of the maximum slip momentum was on the F2 fault (about 8 km to the south of the epicenter), the main slip occurred at a shallower depth (~3 km), and the fault ruptured to the surface. Zhang et al. [13] also acquired the slip distributions on both F1 and F2 by the use of more data sets, including local seismic data (strong motion and broadband) and teleseismic data. They showed that the range of the slip distribution was more concentrated and that the magnitudes were larger than those of Liu et al. [36] The results obtained in this study are more consistent with those of Zhang et al. [13], but there are some differences. The impacted area identified in this study was smaller than that of Zhang et al. [13], and the depth was shallower on both of the faults (F1: range~7 km, depth 0-5 km; F2: range 17 km, depth, 0-8 km). The magnitudes of both faults in this study were also about three times those of Zhang et al. [13] These differences may have been caused by two factors. The first was the longer interval between the two SAR images, so that the aftershock deformation was included in the inverted deformation map, thus leading to a larger slip magnitude. The other was the near-field constraint by the near-field InSAR deformation map, which can improve the accuracy of the inversion.
In addition, Wei et al. [17] also inverted the 33 far-field GPS sites for slip distribution on two conjugate rectangular planes, and found that the slip patches were mainly concentrated on the northern part of the plane, which was approximately the F2 fault in our study. Although the misfit of GPS was 2.2 mm in the E-W direction and 1.8 mm in the S-N direction in their study, which is similar to our results (0.2 mm in both directions), the slip magnitude of F2 (0.15 m) was much less than our results and those of Zhang et al. [13] and Liu et al. [36], and the range (30 km) and depth (−18 to 0 km) [17] were much larger. Furthermore, the main surface rupture zones were not clearly shown in their slip distribution. This suggests that the inversion may not be reliable in the absence of near-field data sets. The addition of the InSAR-constrained observations significantly optimizes the results of the slip distribution, which makes the results more reasonable and finer. This implies that it was necessary to integrate InSAR observations for an in-depth study of the Ludian earthquake.

Coulomb Static Stress Change
Coseismic slip due to earthquakes can alter the stress states of neighboring faults, as well as the seismicity, via the transfer of stress. To evaluate the effect of the Ludian earthquake on the surrounding tectonic activities and the occurrence of aftershocks, we calculated the static coulomb failure stress (CFS) changes on the optimally oriented strike faults using the inverted best-fitting slip model. The friction coefficient was assumed to be 0.4, and the shear modulus was assumed to be 30 GPa. The regional tectonic principal stress, which is required for the estimation of the optimal failure directions, was simply assumed to be 100 bars, with an orientation of S45 • E [37].
The CFS estimation shows that the CFS was greatly increased at shallow depths of 0-5 km (Figure 7). The Xiaohe and Longshu faults tend to be a high seismic risk due to the enhanced CFS along these two faults. Notably, we can see from Figure 7c,d that most of the aftershocks were located in regions that show a decreased CFS, indicating that these aftershocks were likely to have been triggered by dynamic stress rather than static stress transfer [38]. The seismic moment released by the aftershocks was approximately 6.9 × 10 18 N·m greater than that released by the main shock (2.6 × 10 18 N·m). What is more, the main slip region was located at depths of 0-5 km (Figure 6), which is spatially complementary with the depths of the aftershocks, which were mostly located at depths of 5-20 km (Figure 7c,d). These findings suggest that the elastic strain accumulated in the interseismic cycle in the seismogenic region could have been fully released by the main shock and the subsequent aftershocks. As shown in Figure 2b, many faults conjugate in the epicentral zone of the Ludian earthquake. The absence of coseismic rupture on the deeper fractures (5-20 km) may therefore indicate that the deep structure of the region is extremely complex. At depth (5-20 km), structural barriers could have been formed at the intersection of the sub-faults to prevent the earthquake from fracturing into deeper parts. The elastic strain of the deeper parts could have been partitioned by the aftershocks, which suggests that the seismogenic depths of the Zhaotong-Ludian fault could be 0-20 km.
Remote Sens. 2019, 11, x FOR PEER REVIEW 9 of 11 earthquake from fracturing into deeper parts. The elastic strain of the deeper parts could have been partitioned by the aftershocks, which suggests that the seismogenic depths of the Zhaotong-Ludian fault could be 0-20 km.

Conclusions
In this study, we used GPS and InSAR, plus relocated aftershock data, to study the 2014 Ludian earthquake. InSAR provided an important near-field constraint on the detailed slip distribution inversion. The earthquake showed a complex deformation pattern, and a conjugate fault was deemed to be responsible for its occurrence. The major slip of the main shock was located at depths of 0-5 km, which is spatially complementary with the depth of the aftershocks, which were mostly located at depths of 5-20 km. The accumulated elastic strain at depths of 0-20 km could have been fully released by the Ludian earthquake and its subsequent aftershocks. Furthermore, the aftershocks could have been the result of dynamic triggering rather than static triggering.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1. Figure S1: The detailed process for removing the orbital error. Figure S2: Quadtree decomposition image of coseismic deformation map. Figure S3: Trade-off curve between misfit and model roughness.
Author Contributions: All of the authors participated in the editing and reviewing of the manuscript. Y.N. led the research and processed the InSAR data. S.W. calculated the Coulomb stress change data. Y.N. and S.W. inverted the coseismic slip distribution. Q.Z., W.Z., Z.L., C.Z., and W.Q. analyzed and interpreted the results, and reviewed the manuscript. All authors contributed to the writing of this manuscript.

Conclusions
In this study, we used GPS and InSAR, plus relocated aftershock data, to study the 2014 Ludian earthquake. InSAR provided an important near-field constraint on the detailed slip distribution inversion. The earthquake showed a complex deformation pattern, and a conjugate fault was deemed to be responsible for its occurrence. The major slip of the main shock was located at depths of 0-5 km, which is spatially complementary with the depth of the aftershocks, which were mostly located at depths of 5-20 km. The accumulated elastic strain at depths of 0-20 km could have been fully released by the Ludian earthquake and its subsequent aftershocks. Furthermore, the aftershocks could have been the result of dynamic triggering rather than static triggering.
Author Contributions: All of the authors participated in the editing and reviewing of the manuscript. Y.N. led the research and processed the InSAR data. S.W. calculated the Coulomb stress change data. Y.N. and S.W. inverted the coseismic slip distribution. Q.Z., W.Z., Z.L., C.Z., and W.Q. analyzed and interpreted the results, and reviewed the manuscript. All authors contributed to the writing of this manuscript. All authors have read and agreed to the published version of the manuscript.