The left-lateral East Anatolian Fault (EAF), one of the most active intra-continental transform faults in the Eastern Mediterranean, has a history of destructive earthquakes. The EAF's intricate geometry, with bends, step-overs, and sub-parallel faults, is particularly complex in southern Turkey where it connects with the Dead Sea Fault (DSF). This complexity might seem to hinder the development of large earthquakes. However, on February 6, 2023, a M7.8 earthquake struck southeastern Turkey and northern Syria, followed by a M7.5 event 9 hours later along the Sürgü Çardak (S-C) fault. This doublet caused over 50,700 casualties, making it the deadliest in the region since 525 AD. The M7.8 event's epicenter was on the Nurdagi-Pazarik Fault (NPF) splay fault, separate from the main EAF strand, suggesting a complex rupture history. Most studies suggest the initial rupture occurred on the NPF before propagating onto the main EAF strand. However, the subsequent rupture evolution and speed remain debated, with some suggesting instantaneous bilateral rupture, while others show a period of lateral propagation before bilateral rupture. This discrepancy highlights the need for a more comprehensive analysis of the rupture dynamics and the role of pre-existing stress.

Prior research on the 2023 Kahramanmaraş earthquake doublet has focused on various aspects of the event, including the rupture characteristics of both the M7.8 and M7.5 earthquakes. Several studies have used different data sources and methods to estimate the rupture speed, resulting in varying conclusions regarding the presence of supershear rupture. Some studies have focused on the complex fault geometry and its influence on rupture propagation, while others have investigated the role of stress transfer between the two events. While studies provide insights into different aspects, inconsistencies exist regarding the transition from the initial rupture to the main fault, and the rupture speeds, especially for the M7.5 event. There is a need to integrate multiple datasets and advanced inversion techniques to better understand these complex processes.

This study employed a multi-faceted approach to analyze the 2023 Kahramanmaraş earthquake doublet. Fault geometry was constrained by combining surface rupture measurements from optical image correlation of Sentinel-2 data and 2D horizontal deformation from C-band Sentinel-1 SAR images using precise co-registration and sub-pixel correlation. 3D surface deformation was determined by inverting optical and radar measurements from six independent look directions. The fault slip vector was measured from fault-parallel discontinuities of 3D surface displacements. Subsurface dip angles were estimated for each fault segment using Bayesian inversion, incorporating interferometric synthetic aperture radar (InSAR) measurements from ALOS-2 and coseismic GNSS offsets. A parallel sequential Markov Chain Monte Carlo (MCMC) sampler was used for this Bayesian inference. Kinematic finite source models for both events were then determined using a standard methodology, jointly inverting high-rate GNSS data, strong motion waveforms, and GNSS static displacement. The background stress field was estimated from pre-event focal mechanisms (2007–2020) to investigate the mechanisms enabling energetic ruptures in a complex fault system. The inversion included GNSS static offsets at 26 stations (M7.8) and 14 stations (M7.5), and 1-Hz waveforms at 17 stations for both events, along with strong motion records from 16 (M7.8) and 31 (M7.5) stations. ALOS-2 L-band ScanSAR data were processed to obtain InSAR measurements.

The study's key findings include: 1) The M7.8 event showed an average rupture velocity of 3.0–4.0 km/s, with a 10-second delay before bilateral propagation. 2) The M7.5 event exhibited bilateral supershear rupture velocities of 5.0–6.0 km/s on segments B2, B3, and B4, dropping to 2.8 km/s on segments B1, B5, and B6. 3) The M7.8 event released a moment of 7.76 × 10²⁰ Nm (Mw 7.86) over 75 s, with a maximum slip of ~11 m. 4) The M7.5 event released a moment of 5.57 × 10²⁰ Nm (Mw 7.76) over ~35 s, demonstrating a more energetic rupture. 5) For the M7.8 event, most fault segments were nearly optimally oriented relative to the local stress tensor, even across a ~40° bend, potentially due to dynamically increased shear stresses at the bend and along-strike rotation of the pre-stress field. 6) The M7.5 event, however, involved rupture on segments highly oblique to the regional stress field. This suggests significant local stress heterogeneity, potentially linked to a 30° rotation of the principal strain-rate between the northeast region of the M7.8 rupture and the S-C fault region. This local stress field may have been caused by a crustal weakness or variations in crustal material properties. 7) The study highlights the importance of considering local stress heterogeneities, even for large earthquakes. The observed supershear ruptures, despite geometric complexities and apparent misorientations to regional stress for the M7.5 event, were likely facilitated by these stress heterogeneities and free surface effects.

The findings challenge the conventional understanding that geometric complexities inhibit rupture propagation in large earthquakes. The sustained high-velocity rupture along the EAF–DSF intersection and across the ~40° bend in the M7.8 event, and the supershear rupture of the M7.5 event on faults significantly misoriented with respect to the regional stress field, strongly indicate the importance of pre-stress heterogeneities. The study underscores that dynamic stress changes near the free surface on dipping faults can significantly affect rupture propagation and potentially explain the supershear behavior. The significant local stress heterogeneity observed for the M7.5 event emphasizes the need for better resolution of local stress fields in seismic hazard assessment. Integrating strain-rate data with focal mechanism data can provide a more complete picture of stress conditions in regions with sparse seismicity.

This study demonstrates the critical role of pre-stress heterogeneities in steering super-shear ruptures during the 2023 Kahramanmaraş earthquake doublet. The near-optimal orientation of the M7.8 event's sub-faults, despite geometric complexity, and the supershear rupture of the M7.5 event on locally well-oriented faults, even though misaligned with the regional stress field, highlight the complex interplay between fault geometry, pre-stress, and dynamic stress changes. Future research should focus on refining models of dynamic rupture propagation incorporating high-resolution stress estimations that integrate geodetic and seismological data, including the investigation into how stress heterogeneity might arise. This would enhance our understanding of earthquake rupture and improve seismic hazard assessments.

The study's limitations include the uncertainties associated with subsurface fault geometries, although Bayesian inversion was used to quantify these uncertainties. The assumptions made in inverting the stress field from focal mechanisms and the relative weighting of different data types in the finite source inversion also introduce uncertainty. While efforts were made to minimize these limitations, their impact on interpretation should be acknowledged. Future work could further refine the models and address some of these uncertainties.