Role of the Nyainrong Microcontinent in Seismogenic Mechanism and Stress Partitioning: Insights from the 2021 Nagqu Mw 5.7 Earthquake

: The Nyainrong microcontinent carries key information about the ongoing evolution of the central Tibetan Plateau. The 2021 Mw 5.7 Nagqu earthquake is the largest instrumentally recorded event inside this microcontinent, which provides an ideal opportunity to elucidate the inﬂuence of this ancient microcontinent on the seismogenic mechanisms, stress heterogeneity and strain partitioning across the Tibetan Plateau. Here, we constrain the seismogenic fault geometry and distributed fault slip using Interferometric Synthetic Aperture Radar (InSAR) observations. By using the regional focal mechanism solutions, we invert the stress regimes surrounding the Nyainrong microcontinent. Our analysis demonstrates that the mainshock was caused by a normal fault with a comparable sinistral strike-slip component on a North-West dipping fault plane. The Nyainrong microcontinent is surrounded by a dominant normal faulting stress regime to the northeast and a dominant strike-slip stress regime to the southwest. Moreover, the clockwise rotation of the maximum horizontal stress (SHmax) from the southwest to the northeast is ~20 ◦ . This indicates that the Nyainrong microcontinent is involved in the mainshock occurrence as well as regional stress heterogeneity, and strain partitioning. Our results highlight the signiﬁcance of the ancient microcontinent in the tectonic evolution of the Tibetan Plateau. Author Contributions: Conceptualization, X.L.; Data curation, X.L.; Formal analysis, X.L., L.X. and Y.L.; Funding acquisition, W.X.; Investigation, X.L.; Methodology, X.L.; Resources, X.L., L.X., Y.L. and W.X.; Software, X.L.; Supervision, W.X.; Validation, L.X. and Z.C.; B.H.; Writing—original


Introduction
The Nyainrong microcontinent, which is also called the Amdo basement/terrane, is located at the juncture of the Qiangtang terrane and Lhasa terrane in the central Tibetan Plateau [1,2]. Previous paleoecology and geochemistry studies have demonstrated that the Nyainrong microcontinent is an older tectonic unit, with an age of between~170 tõ 900 million years (Ma) [1][2][3][4]. It appeared much earlier than the initiation age (~50 Ma) of the India-Eurasia collision [5]. The lower intercept age of 170 Ma corresponds to the timing of low-grade metamorphism during the initial Lhasa-Qiangtang collision due to the Bangong Ocean closing [1]. As shown in Figure 1, the ancient basement rock in the Nyainrong microcontinent differs from the surrounding sedimentary rocks [3]. The Nyainrong microcontinent is an east-west-trending eye-shaped terrane, covering an area of~8000 km 2 ( Figure 1). Although it is widely acknowledged that Tibet's thick crust and high elevation originated from the India-Eurasia collision, the role of the older tectonics is less understood [6]. However, previous investigations of the Nyainrong microcontinent were mainly focused on long-term (i.e., geological time of Ma) tectonic evolution based on petrological, geochronological, and geochemical methods [1][2][3][4]. Therefore, a study The Nagqu Mw 5.7 earthquake occurred in the east corner of the Nyainrong microcontinent on 19 March 2021, which is the largest earthquake inside the microcontinent during its instrumental history. This earthquake offers an unprecedented opportunity to explore the kinematics and stress regime of this microcontinent based on the Interferometric Synthetic Aperture Radar (InSAR) images and well-recorded seismic data. Geodetic and seismological data were previously used by various research institutions, which revealed the focal mechanisms of this earthquake, consistently characterized by normal faulting with a moderate strike-slip component (Table 1). Li et al. (2021) investigated the fault geometry and mechanisms of this earthquake using Sentinel-1 data, and the stressloading effects of historical large M > 7 earthquakes on the mainshock [11]. Li et al. (2022) examined the stress and strain characteristics in the seismic region of the 2021 Nagqu Mw 5.7 earthquake based on stress data, focal mechanisms, and the GNSS data [12]. However, these studies did not consider the effect of the Nyainrong microcontinent on the seismic dynamics and surrounding stress and strain. Thus, it remains unclear what role the Nyainrong microcontinent played in the occurrence of the mainshock and the regional Figure 1. (a) Regional seismotectonic context around the 2021 Nagqu Mw 5.7 earthquake. Red focal mechanism is the hypocenter of the 2021 Nagqu event. Gray focal mechanisms are historical events with Mw > 5.5 from Global Centroid Moment Tensor (GCMT) [7]. Black focal mechanisms are Mw > 7 historical earthquakes. Black dots are historical earthquakes with Mw between 4 and 5.5. Purple and black vectors are, respectively, continuous and campaign Global Navigation Satellite System (GNSS) velocities from 1998 to 2014 [8]. Blue polygons are lakes with area larger than 1 km 2 from National Tibetan Plateau Data Center [9]. Black rectangles represent the spatial coverages of Sentinel-1 SAR data from one ascending track 143 (AT143) and one descending track 77 (DT77). Blue rectangle bounds the extent of enlarged (b). (b) The zoom-in of blue box in (a). (c) Geological map covering a similar region of (b) modified from the 1:1.5 million geological maps of the Tibetan Plateau and detailed lithology properties are available [10]. The lithology age marked in (c) is adopted from Xie et al. (2014) [2].
The Nagqu Mw 5.7 earthquake occurred in the east corner of the Nyainrong microcontinent on 19 March 2021, which is the largest earthquake inside the microcontinent during its instrumental history. This earthquake offers an unprecedented opportunity to explore the kinematics and stress regime of this microcontinent based on the Interferometric Synthetic Aperture Radar (InSAR) images and well-recorded seismic data. Geodetic and seismological data were previously used by various research institutions, which revealed the focal mechanisms of this earthquake, consistently characterized by normal faulting with a moderate strike-slip component (Table 1). Li et al. (2021) investigated the fault geometry and mechanisms of this earthquake using Sentinel-1 data, and the stress-loading effects of historical large M > 7 earthquakes on the mainshock [11]. Li et al. (2022) examined the stress and strain characteristics in the seismic region of the 2021 Nagqu Mw 5.7 earthquake based on stress data, focal mechanisms, and the GNSS data [12]. However, these studies did not consider the effect of the Nyainrong microcontinent on the seismic dynamics and surrounding stress and strain. Thus, it remains unclear what role the Nyainrong microcontinent played in the occurrence of the mainshock and the regional tectonic evolution, which is crucial for better understanding the mechanisms of earthquake and faulting [13,14]. Here, we use Sentinel-1 SAR images to produce the coseismic InSAR displacement caused by the 2021 Nagqu Mw 5.7 earthquake. With both ascending and descending InSAR observations, we decompose the 2D displacements including horizontal components along the fault trace and vertical deformation. We further constrain the seismogenic fault geometry and distributed fault slip to explore the faulting kinematics within the Nyainrong microcontinent. Local focal mechanisms are used to perform the stress inversion and investigate the role of the Nyainrong microcontinent in the regional stress heterogeneity and strain partitioning. Our results highlight the significant role of ancient special structures such as the Nyainrong microcontinent in the tectonic evolution and stress/strain state of the Tibetan Plateau.

InSAR Observations
The 2021 Nagqu earthquake occurred in the center of Tibet, where there are arid climate conditions (average monthly precipitation~20 mm) [15] and sparse vegetation coverage. Such conditions are conducive to applying InSAR technology to obtain the completed displacements over the epicenter region [16,17]. The constellation of Sentinel-1 satellites is operated by the European Space Agency with a shortened revisiting period (6/12 days), which strengthens their capacity to capture millimeter-level surface displacement. Here, we use C-band Sentinel-1 SAR data from one ascending track 143 (AT143) and one descending track 77 (DT77) to map coseismic interferograms of the 2021 Nagqu earthquake ( Figure 1 and summarized in Table 2). Relying on the precise Sentinel-1 orbits, high-accuracy coregistration between the primary and secondary Single Look Complex (SLC) was carried out with the Gamma software [18]. Then, two coseismic interferograms were generated with the coregistered SLCs. Furthermore, the one arc-sec digital elevation model from the Shuttle Radar Topography Mission [19] and an improved power spectrum filter method [20] was applied to mitigate the effect of topography and the phase noise in the interferograms, respectively. Finally, the coseismic surface displacement fields were retrieved by unwrapping the filtered interferograms by a minimum cost flow method [21]. In addition, given the topographic relief in the epicenter region, the potential topographydependent tropospheric delay was estimated and removed [22]. Given the excellent coherence and somewhat low noise level in the epicenter region, the obtained coseismic interferograms characterize well-defined surface displacements caused by the Nagqu earthquake ( Figure 2). The displacements are concentrated in the east corner of the Nyainrong microcontinent. The descending coseismic interferogram revealed a symmetric double-lobe pattern with a peak subsidence value of~3 cm away from the satellite line of sight (LOS) direction. The southeastern region is characterized by uplift (LOS range decrease), while the northeastern region is characterized by subsidence (LOS range increase). However, the ascending coseismic interferogram reveals only one subsidence lobe with up to~2 cm LOS displacement. These distinct displacement patterns between the two tracks can be potentially attributed to the less favorable viewing geometry associated with the ascending orbit. Furthermore, we do not discern any remarkable ruptures from the displacement maps, which indicates a blind fault in the SW-NE orientation. Given the excellent coherence and somewhat low noise level in the epicenter region, the obtained coseismic interferograms characterize well-defined surface displacements caused by the Nagqu earthquake ( Figure 2). The displacements are concentrated in the east corner of the Nyainrong microcontinent. The descending coseismic interferogram revealed a symmetric double-lobe pattern with a peak subsidence value of ~3 cm away from the satellite line of sight (LOS) direction. The southeastern region is characterized by uplift (LOS range decrease), while the northeastern region is characterized by subsidence (LOS range increase). However, the ascending coseismic interferogram reveals only one subsidence lobe with up to ~2 cm LOS displacement. These distinct displacement patterns between the two tracks can be potentially attributed to the less favorable viewing geometry associated with the ascending orbit. Furthermore, we do not discern any remarkable ruptures from the displacement maps, which indicates a blind fault in the SW-NE orientation.

Two-dimensional Displacement
As the LOS displacement represents only one-dimensional (1D) manifestation along the satellite viewing direction, it provides only limited constraints on the real pattern of crustal deformation (e.g., dextral or sinistral strike-slip, thrust or dip-slip) [23]. In this context, 2D (two-dimensional) displacement can improve our understanding of the characteristics of crustal deformation caused by the 2021 Mw 5.7 Nagqu earthquake. To this end,

Two-Dimensional Displacement
As the LOS displacement represents only one-dimensional (1D) manifestation along the satellite viewing direction, it provides only limited constraints on the real pattern of crustal deformation (e.g., dextral or sinistral strike-slip, thrust or dip-slip) [23]. In this context, 2D (two-dimensional) displacement can improve our understanding of the characteristics of crustal deformation caused by the 2021 Mw 5.7 Nagqu earthquake. To this end, the 2D coseismic ground deformation fields are decomposed from the ascending and descending coseismic interferograms to better constrain the characteristics of the fault deformation [23]. The decomposed 2D displacements are characterized by southwest horizontal motion along the strike direction and subsidence in the northwestern lobe, and northeast horizontal motion along the strike direction and uplift in the southeast lobe ( Figure 3). This identified pattern suggests that the crust deformation during the 2021 Nagqu earthquake is characterized by both normal and sinistral fault slips, corresponding to a transtensional fault slip during the earthquake.
scending coseismic interferograms to better constrain the characteristics of the fault deformation [23]. The decomposed 2D displacements are characterized by southwest horizontal motion along the strike direction and subsidence in the northwestern lobe, and northeast horizontal motion along the strike direction and uplift in the southeast lobe (Figure 3). This identified pattern suggests that the crust deformation during the 2021 Nagqu earthquake is characterized by both normal and sinistral fault slips, corresponding to a transtensional fault slip during the earthquake.

Seismogenic Fault Geometry and Distributed Slip
To reconstruct the source parameters of the seismogenic fault (longitude and latitude of fault location, length, width, depth, strike angle, dip angle, strike-slip and dip-slip), we applied the Geodetic Bayesian Inversion Software (GBIS), which had been previously proposed by Bagnardi and Hooper [24]. Before implementing the source inversion, we calculated the experimental semivariogram to quantify the covariance with tropospheric delay and topographic residuals in the interferograms [24]. Moreover, to achieve a tractable computational burden, we applied an adaptive gradient-based quadtree sampling method to downsample the coseismic interferograms to ~400 grids [25]. By assuming that the seismogenic fault was a rectangular plane with a uniform slip, the elastic half-space dislocation model was utilized. In this way, we determined the Green function, linking the down-sampled InSAR observations with the uniform fault slip [26].
Two NW-SW and SW-NE-orientated fault plane solutions for the Nagqu earthquake were previously issued by different agencies (Table 1). Given the prior information of double-lob coseismic displacement features (Figure 2), the decomposed 2D displacement fields (Figure 3), and the consistent normal or transtensional faulting from different agencies (Table 1), we preferred the SW-NE orientated fault plane dipping northwest. Thus, we set the sampling boundary for a strike angle (180, 300) and a dip angle (0, 90), and loose sampling boundaries were provided for other source parameters to broadly cover the solution space. In the Bayesian inverse approach, the optimal model parameters were determined by finding the maximum-a-posteriori probability solution from the posterior

Seismogenic Fault Geometry and Distributed Slip
To reconstruct the source parameters of the seismogenic fault (longitude and latitude of fault location, length, width, depth, strike angle, dip angle, strike-slip and dip-slip), we applied the Geodetic Bayesian Inversion Software (GBIS), which had been previously proposed by Bagnardi and Hooper [24]. Before implementing the source inversion, we calculated the experimental semivariogram to quantify the covariance with tropospheric delay and topographic residuals in the interferograms [24]. Moreover, to achieve a tractable computational burden, we applied an adaptive gradient-based quadtree sampling method to downsample the coseismic interferograms to~400 grids [25]. By assuming that the seismogenic fault was a rectangular plane with a uniform slip, the elastic half-space dislocation model was utilized. In this way, we determined the Green function, linking the down-sampled InSAR observations with the uniform fault slip [26].
Two NW-SW and SW-NE-orientated fault plane solutions for the Nagqu earthquake were previously issued by different agencies (Table 1). Given the prior information of double-lob coseismic displacement features (Figure 2), the decomposed 2D displacement fields (Figure 3), and the consistent normal or transtensional faulting from different agencies (Table 1), we preferred the SW-NE orientated fault plane dipping northwest. Thus, we set the sampling boundary for a strike angle (180, 300) and a dip angle (0, 90), and loose sampling boundaries were provided for other source parameters to broadly cover the solution space. In the Bayesian inverse approach, the optimal model parameters were determined by finding the maximum-a-posteriori probability solution from the posterior probability density functions, which were obtained by sampling with the Markov chain Monte Carlo method [27]. More detailed information about the inversion was provided in a previous study [24].
After 5 × 10 6 iterations, we obtained well-converged Markov chains, indicating it did explore the parameter space sufficiently. The maximum-a-posteriori probability uniform slip model solution reveals the optimal fault plane of 9.8 km long and 9.3 km wide with a strike angle of 240.4 • and a dip angle of 59.2 • (Table 3). Trade-offs between the fault parameters are somewhat inconspicuous (i.e., strike and dip angles), as indicated by the histograms of posterior probability distributions (Figure 4). These parameters of fault geometry are consistent with previous studies that used either geodetic or seismological data (Table 1).   To resolve a distributed coseismic slip model, while also improving the fitness of the data and models, we fixed the fault geometry, derived from GBIS non-linear inversion. Moreover, we extended the fault length and the width to 30 km and 18 km (i.e., to 15 km depth), respectively. In this context, we utilized an automated fault discretization method to invert the finite coseismic slip model [28]. This approach iteratively discretizes the fault plane to account for the spatial variations of model resolution during the distributed slip inversion. The higher order Tikhonov Regularization strategy was adopted to stabilize the  To resolve a distributed coseismic slip model, while also improving the fitness of the data and models, we fixed the fault geometry, derived from GBIS non-linear inversion. Moreover, we extended the fault length and the width to 30 km and 18 km (i.e., to 15 km depth), respectively. In this context, we utilized an automated fault discretization method to invert the finite coseismic slip model [28]. This approach iteratively discretizes the fault plane to account for the spatial variations of model resolution during the distributed slip inversion. The higher order Tikhonov Regularization strategy was adopted to stabilize the slip estimation, and the regularization factor was determined using the jRi strategy ( Figure 5) to balance the regularization and perturbation errors [28]. Note that the regularization and perturbation errors reflect the difference between noise-free observations and regularized model results and the influence of observation noise on the inversion results, respectively. The coseismic slip inversion results on a northwest dipping fault configurati cate that the 2021 Nagqu earthquake ruptured on a fault structure with a length of and a width of ~12 km ( Figure 6). Coseismic slip mainly concentrates at 2−12 km i and the maximum slip (~0.3 m) is located at a depth of ~8 km (Figure 6a). The co fault slip reveals the comparable components of dip-slip and sinistral slip (Figu with a mean rake angle of −56° within the slip zone (>0.1 m). This illustrates that Nagqu earthquake occurred within local transtensional tectonics. The absence of able coseismic slip in shallow depth (0−2 km) suggests that the Nagqu earthqua not have ruptured the surface ( Figure 6). The predicted displacements from the ting model explain both the ascending and descending coseismic observations w a Root Mean Square of ~0.7 cm (ascending) and ~0.5 cm (descending), respectively 2). The remaining residuals can be potentially explained by the residual topograp atmospheric artifacts. Assuming the average shear moduli of 33 GPa in the epice gion [29], we calculated a geodetic moment as ~4.5 × 10 17 Nm, corresponding to Our estimated coseismic slip distribution and moment magnitude are comparab those reported in previous studies using the geodetic or seismological dataset (Ta The coseismic slip inversion results on a northwest dipping fault configuration indicate that the 2021 Nagqu earthquake ruptured on a fault structure with a length of~15 km and a width of~12 km ( Figure 6). Coseismic slip mainly concentrates at 2−12 km in depth, and the maximum slip (~0.3 m) is located at a depth of~8 km (Figure 6a). The coseismic fault slip reveals the comparable components of dip-slip and sinistral slip (Figure 6b,c) with a mean rake angle of −56 • within the slip zone (>0.1 m). This illustrates that the 2021 Nagqu earthquake occurred within local transtensional tectonics. The absence of remarkable coseismic slip in shallow depth (0−2 km) suggests that the Nagqu earthquake may not have ruptured the surface ( Figure 6). The predicted displacements from the best-fitting model explain both the ascending and descending coseismic observations well, with a Root Mean Square of~0.7 cm (ascending) and~0.5 cm (descending), respectively ( Figure 2). The remaining residuals can be potentially explained by the residual topography and atmospheric artifacts. Assuming the average shear moduli of 33 GPa in the epicenter region [29], we calculated a geodetic moment as~4.5 × 10 17 Nm, corresponding to Mw 5.7. Our estimated coseismic slip distribution and moment magnitude are comparable with those reported in previous studies using the geodetic or seismological dataset (Table 1).

Stress Inversion from Local Focal Mechanisms
We estimated the local stress fields by the focal mechanisms during 1990−2019 from a previous study [11]. We focused on three event clusters (M 4.3), namely Zone 1-3, surrounding the Nyainrong microcontinent ( Figure 7). Zone 1 is adjacent to the southwest region of the Nyainrong microcontinent, while Zone 2 and Zone 3 are adjacent to the northeastern region of the Nyainrong microcontinent ( Figure 7). In total, 22 focal solutions were classified as six, eight, and eight records in Zone 1, Zone 2 and Zone 3, respectively (Table 4). We utilized an iterative joint inversion method [30] to calculate principal stresses , , , and the stress ratio / , 0 1 representing the relative magnitudes of the principal stresses [31]. With the bootstrap resampling approach [32], this method allows for determining the 95% confidence intervals of an optimal stress tensor. To obtain the optimal stress parameters ( Table 4), we estimated 2000 bootstrap samples with random noise of 10°. The average misfit angle α is the difference between the observed and predicted fault slip directions, which can be used to reflect the degree of stress heterogeneity and evaluate the performance of stress inversion.

Stress Inversion from Local Focal Mechanisms
We estimated the local stress fields by the focal mechanisms during 1990−2019 from a previous study [11]. We focused on three event clusters (M ≥ 4.3), namely Zone 1-3, surrounding the Nyainrong microcontinent ( Figure 7). Zone 1 is adjacent to the southwest region of the Nyainrong microcontinent, while Zone 2 and Zone 3 are adjacent to the northeastern region of the Nyainrong microcontinent ( Figure 7). In total, 22 focal solutions were classified as six, eight, and eight records in Zone 1, Zone 2 and Zone 3, respectively (Table 4). We utilized an iterative joint inversion method [30] to calculate principal stresses σ 1 , σ 2 , σ 3 , and the stress ratio R = (σ 1 − σ 2 )/(σ 1 − σ 3 ), 0 < R < 1 representing the relative magnitudes of the principal stresses [31]. With the bootstrap resampling approach [32], this method allows for determining the 95% confidence intervals of an optimal stress tensor. To obtain the optimal stress parameters ( Table 4), we estimated 2000 bootstrap samples with random noise of 10 • . The average misfit angle α is the difference between the observed and predicted fault slip directions, which can be used to reflect the degree of stress heterogeneity and evaluate the performance of stress inversion.   Zone 1 is characterized by a near EW σ 1 orientation with a 36 • plunge angle, a low R (0.34) value and near horizontal σ 3 (Figure 7b and Table 4). This implies that Zone 1 is dominated by a strike-slip (SS) stress regime according to Zoback's classification scheme [33]. However, the plunge angles (~80 • ) of σ 1 axes in Zone 2 and Zone 3 are substantially larger than those in Zone 1. In addition, the near horizontal σ 2 and σ 3 in Zone 2 and Zone 3 also indicate that the northeastern zones are dominated by normal fault (NF) stress regime (Figure 7 and Table 4). However, the typical girdle distributions of the σ 1 /σ 2 samples and relatively low stress ratio R (~0.25), especially in Zone 3, indicate a stress permutation between σ 1 and σ 2 (Figure 7) [34]. These switches between σ 1 and σ 2 reflect the close correlation between strike-slip and normal faulting behaviors [34], resonating with different faulting behaviors in the study region (Figure 7). Considering the used focal mechanisms are within acceptable uncertainties, the larger misfit angle of 28 • in Zone 1 than that of~15 • in Zone 2 and Zone 3, indicates the background tectonic stress fields in Zone 1 are higher heterogeneous [35]. The maximum horizontal stress (SHmax), estimated based on the method proposed by Lund and Townend [36], shows the directions of SHmax in the three subzones are directed on near north-south (Figure 7 and Table 4). However, the striking clockwise rotation of SHmax is obvious from Zone 1 to Zone 3. This stress clockwise rotation from southwest to northeast is consistent with the clockwise rotation of GNSS velocity [8].

Origins of the 2021 Nagqu Earthquake
The collision of the India-Eurasia plates caused widespread Cenozoic deformation throughout the Tibetan Plateau and its surrounding regions [37,38]. The east-west extension and the north-south shortening has dominated crust deformation in the Tibetan Plateau from the Neogene to the present day, along with the ongoing convergence that occurs at the Tibetan Plateau margin [39,40]. Thus, the forces driving the India-Eurasia plate collision are suggested as the fundamental driving force for the occurrence of the 2021 Nagqu event. Seismo-tomographic results corroborate the suggestion that the eastward extension resulted from the eastward movement of the lower crust and/or upper mantle [41,42]. In addition, the fact that about 85% of movement release of normal faulting over the last half century has occurred in the region with high altitudes (>5 km), suggests variations in the gravitational potential energy of the lithosphere are likely responsible for material extension [43].
As the 2021 Nagqu Mw 5.7 earthquake is the largest ever recorded event within the Nyainrong microcontinent, this indicates a relatively low frequency of strong earthquakes inside the microcontinent. The earthquake occurred within the ancient (~170−900 Ma) Nyainrong microcontinent [1][2][3][4], containing rocks with a distinct lithology (e.g., orthogneiss) from the surrounding sedimentary rocks (Figure 1). Thus, the Nyainrong microcontinent may impede lateral material movement, forcing the material to detour around it. Consequently, the accumulated tectonic strain at the boundary modulates the local geodynamic process, which helps explain why strong earthquakes are rare within the Nyainrong microcontinent. This is similar to the special structure, namely the Emeishan large igneous province, in the southeast Tibetan Plateau. Few strong earthquakes occurred in this special structure with similar detoured crustal flow [44,45], which played a vital role in the occurrence of the 2021 Mw 6.1 Yangbi earthquake [46]. Similarly, the microplates Shillong Plateau and Assam Basin play an essential role in controlling the regional seismicity patterns (i.e., sparse seismic activity and small magnitude), as previously indicated by the numerical block-and-fault dynamics model [47]. This provides a reasonable explanation for the rare moderate-strong earthquakes recorded inside the Nyainrong microcontinent.
In light of the above analysis, we suggest that the north-south shortening, the eastwest extension, and the ancient Nyainrong microcontinent are jointly involved in controlling the occurrence of the 2021 Nagqu Mw 5.7 earthquake. The north-south shortening and east-west extension formed stress on the Nyainrong microcontinent (Figure 8), which is constituted of rigid basement rock that has substantially lower tensile strength than compressive strength [48] and is vulnerable to normal faulting behavior. This explains the normal faulting with a strike-slip component in the interior of the Nyainrong micro-continent, especially in the corner region where the strike direction of boundary faults changes considerably and is conducive to stress accumulation and eventually becomes the nucleation site of the 2021 Nagqu earthquake. Interestingly, this phenomenon is consistent with the normal faulting (i.e., Cona Lake) in the west counterpart of the Nyainrong microcontinent ( Figure 8).

Implication of the Nyainrong Microcontinent on Strain Partitioning and Stress Heterogeneous
Previous studies have reported that strain partitioning across faults is a widespread phenomenon in the Tibetan Plateau [42,[49][50][51][52][53]. According to wave speed models derived from dense seismographs, significant differences in structure and rheology are discerned across large faults in the eastern Tibetan Plateau [42], which indicates strain partitioning across major faults. Block kinematic modeling based on the GNSS velocity field reveals strain partitioning on major fault structures in the northeast Tibetan Plateau [49]. Joint GNSS-InSAR data models reveal strain partitioning along the Altyn Tagh Fault and the Jinsha suture zone in the northwestern Tibetan Plateau [51]. In the southeast Tibetan Plateau, quantitative kinematical data of main faults along the Sichuan-Yunnan block from photogrammetric, geomorphological, and chronological methods, reveal dip-slip components in many fault segments while they are strike-slip dominated [52]. Earthquake focal mechanisms in the Tibetan Plateau indicate that deformation is dominated by thrust faulting in the margin of the Tibetan Plateau, normal faulting in the southern Tibetan Plateau and strike-slip faulting in the northern Tibetan Plateau (Figure 1) [45,50,51].
The inverted stress fields of three subzones surrounding the Nyainrong microcontinent ( Figure 7) allow for exploring the effect of local structure (i.e., the Nyainrong microcontinent) on stress/strain partitioning. Zone 1 is under a strike-slip regime, which contrasts with the widespread normal faulting and dip-slip events in the further southwest region (Figures 1 and 7). While both Zone 2 and Zone 3 are under the stress regime of normal faulting, which contrasts with the widespread strike-slip faulting and events in the further northeast region (Figures 1 and 7). In comparison, our slip model indicates a transtensional slip in the interior of the Nyainrong microcontinent ( Figure 6). The NE-SW variation in faulting behavior across the Nyainrong microcontinent is potentially associated with the obstruction of the Nyainrong microcontinent, which transforms the strain

Implication of the Nyainrong Microcontinent on Strain Partitioning and Stress Heterogeneous
Previous studies have reported that strain partitioning across faults is a widespread phenomenon in the Tibetan Plateau [42,[49][50][51][52][53]. According to wave speed models derived from dense seismographs, significant differences in structure and rheology are discerned across large faults in the eastern Tibetan Plateau [42], which indicates strain partitioning across major faults. Block kinematic modeling based on the GNSS velocity field reveals strain partitioning on major fault structures in the northeast Tibetan Plateau [49]. Joint GNSS-InSAR data models reveal strain partitioning along the Altyn Tagh Fault and the Jinsha suture zone in the northwestern Tibetan Plateau [51]. In the southeast Tibetan Plateau, quantitative kinematical data of main faults along the Sichuan-Yunnan block from photogrammetric, geomorphological, and chronological methods, reveal dip-slip components in many fault segments while they are strike-slip dominated [52]. Earthquake focal mechanisms in the Tibetan Plateau indicate that deformation is dominated by thrust faulting in the margin of the Tibetan Plateau, normal faulting in the southern Tibetan Plateau and strike-slip faulting in the northern Tibetan Plateau (Figure 1) [45,50,51].
The inverted stress fields of three subzones surrounding the Nyainrong microcontinent ( Figure 7) allow for exploring the effect of local structure (i.e., the Nyainrong microcontinent) on stress/strain partitioning. Zone 1 is under a strike-slip regime, which contrasts with the widespread normal faulting and dip-slip events in the further southwest region (Figures 1 and 7). While both Zone 2 and Zone 3 are under the stress regime of normal faulting, which contrasts with the widespread strike-slip faulting and events in the further northeast region (Figures 1 and 7). In comparison, our slip model indicates a transtensional slip in the interior of the Nyainrong microcontinent ( Figure 6). The NE-SW variation in faulting behavior across the Nyainrong microcontinent is potentially associated with the obstruction of the Nyainrong microcontinent, which transforms the strain behavior around the margin of the Nyainrong microcontinent. This implies that the Nyainrong microcontinent may play an important role in the transformation of stress/strain behavior (i.e., different faulting behaviors). In addition, a striking clockwise rotation (~20 • ) of SHmax from Zone 1 to Zone 3 ( Figure 7) shows a good agreement with the clockwise rotation of GNSS velocity [8]. Small magnitude events (M < 6.2) are not statistically considered to induce SHmax rotation [54,55]. Thus, stress rotation, caused by a coseismic slip is unlikely to be the scenario observed in this study due to the small event magnitude (M < 6.2) in these three subzones. The spatial relationship between the three subzones and the Nyainrong microcontinent indicates that special structures with distinctly different lithology compared to the surrounding region play an essential role in controlling the stress rotation. This is consistent with the structure controls (e.g., fault structure geometry, contrasting rheology/lithology) in stress behavior proposed in other regions [56][57][58][59][60][61][62].
Thus, being inspired by the 2021 Nagqu Mw 5.7, we conclude that the Nyainrong microcontinent plays a key role in controlling the occurrence of the earthquake and the surrounding strain partitioning (i.e., different faulting behaviors), as well as affecting the stress heterogeneity. Although these ancient structures are not susceptible to internal nucleation of strong earthquakes, the surrounding and corner regions are more prone to stress accumulation and represent highly seismic areas. The 1:1.5 million geological maps of the Tibetan Plateau and its surrounding areas demonstrate multiple similar structures such as the Nyainrong microcontinent [10]. Thus, it is of great significance to consider the effect of these special structures on strain partitioning in future research for deepening our understanding of the evolution, seismogenic mechanism and the dynamic process of the Tibetan Plateau.

Conclusions
In this study, we use surface displacement fields retrieved from InSAR data to explore the seismogenic fault geometry and distributed coseismic slip for the 2021 Mw 5.7 Nagqu earthquake that occurred in the east corner zone of the Nyainrong microcontinent. The 2D displacement maps decomposed with both ascending and descending InSAR data, reveal the maximum surface subsidence of~3 cm at the hanging wall as well as up to 1.5 cm displacement in the strike direction. We find that the InSAR observations can be best explained by the fault slip on a seismogenic fault with a strike of 240 • and a dip angle of 59 • . Coseismic slip is featured by normal faulting and a comparable sinistral strike-slip component with a peak value of 0.3 m at 8 km depth. The ancient Nyainrong microcontinent, as well as the north-south shortening and east-west extension during the India-Eurasia collision both contribute to the occurrence of the 2021 Nagqu Mw 5.7 earthquake. The stress inversion results of three subzones surrounding the Nyainrong microcontinent reveal a distinct strike-slip stress regime in the southwest region and a normal faulting stress regime in the northeast region. This regional stress and strain heterogeneity highlights the significant modulation from the Nyainrong microcontinent on the occurrence of the 2021 Nagqu earthquake. Thus, the Nyainrong microcontinent plays an essential role in the regional tectonic evolution. Despite these promising results, further studies are needed to quantify the effect of special structures such as the Nyainrong microcontinent on the earthquake nucleation process, strain partitioning and stress heterogeneity through numerical simulation using the finite element method and laboratory experiments in rock physics.