Quantifying plate interface coupling in the Mexican subduction zone from InSAR and GNSS using Bayesian inversion methods

The Mexican subduction zone, characterized by intense tectonic activity, constitutes a natural laboratory for investigating the mechanisms controlling seismic-cycle dynamics. This margin has experienced both large, devastating earthquakes (e.g., Michoacán 1985 ; Tehuantepec 2017) and frequent episodes of slow slip. Quantifying interseismic coupling along the subduction interface is therefore essential to better understand the interaction between seismic and aseismic processes and to refine seismic hazard assessment models.In this study, we establish an interseismic coupling map over nearly 1000 km of the Mexican subduction margin using six years of geodetic observations (2015–2022). Our analysis relies on the joint integration of GNSS velocities from 72 carefully selected stations and ten Sentinel-1 tracks (descending andascending) covering the subduction zone from Jalisco to Oaxaca. Velocity maps derived from FLATSIM (ForM@Ter LArge-scale multi-Temporal Sentinel-1 InterferoMetry) processing were corrected for coseismic offsets, cleaned of non-tectonic signals, and referenced to GNSS interseismic velocities. To reduce noise and computational cost while preserving essential information, the InSAR data were spatially downsampled. The resulting interseismic velocities were then used as input for a joint coupling inversion.The inversion is performed within a Bayesian framework (AlTar/CATMIP) and relies on a forward model of dislocations in a homogeneous elastic medium, with a 3D subduction interface discretized into triangular elements. Data uncertainties are incorporated through the covariance matrix, enhancing the robustness of the results. This probabilistic approach, applied for the first time to this study area, allows exploration of the model space and estimation of both the most probable coupling distribution and its posterior uncertainties.The results reveal strong and well-constrained coupling in the Jalisco and Michoacán regions, indicating high seismogenic potential. In contrast, coupling in the Guerrero and Oaxaca regions is more heterogeneous and locally appears negative over the observation period, due to the presence of recurrent slow-slip events and post-seismic deformation, whose transient contributions may exceed the plate-convergence rate.