The Mw 6.1 2016 Parina earthquake led to extension of the south Peruvian Andes along a normal fault with evidence of Holocene slip. We use interferometric synthetic aperture radar, seismology, and field mapping to determine a source model for this event and show that extension at Parina is oriented NE‐SW, which is parallel to the shortening direction in the adjacent sub‐Andean lowlands. In addition, we use earthquake source models and GPS data to demonstrate that shortening within the sub‐Andes is parallel to topographic gradients. Both observations imply that forces resulting from spatial variations in gravitational potential energy are important in controlling the geometry of the deformation in the Andes. We calculate the horizontal forces per unit length acting between the Andes and South America due to these potential energy contrasts to be 4–8 ×1012 N/m along strike of the mountain range. Normal faulting at Parina implies that the Andes in south Peru have reached the maximum elevation that can be supported by the forces transmitted across the adjacent foreland, which requires that the foreland faults have an effective coefficient of friction ≲0.2. Additionally, the onset of extension in parts of the central Andes following orogen‐wide compression in the late Miocene suggests that there has been a change in the force balance within the mountains. We propose that shortening on weak detachment faults within the Andean foreland since ∼5–9 Ma reduced the shear tractions acting along the base of the upper crust in the eastern Andes, leading to extension in the highest parts of the range.

S U M M A R YThe extent to which aseismic deformation relaxes co-seismic stress changes on a fault zone is fundamental to assessing the future seismic hazard following any earthquake, and in understanding the mechanical behaviour of faults. Here we use models of stress-driven afterslip and viscoelastic relaxation, in conjunction with post-seismic InSAR measurements, to show that there has been minimal release of co-seismic stress changes through post-seismic deformation following the 2003 M w 6.6 Bam earthquake. Our analysis indicates the faults at Bam remain predominantly locked, suggesting that the co-plus interseismically accumulated elastic strain stored downdip of the 2003 rupture patch may be released in a future M w 6 earthquake. Our observations and models also provide an opportunity to probe the growth of topography at Bam. We find that, for our modelled afterslip distribution to be consistent with forming the sharp step in the local topography over repeated earthquake cycles, and also to be consistent with the geodetic observations, requires either (1) far-field tectonic loading equivalent to a 2-10 MPa deviatoric stress acting across the fault system, which suggests it supports stresses 60-100 times less than classical views of static fault strength, or (2) that the fault surface has some form of mechanical anisotropy, potentially related to corrugations on the fault plane, that controls the sense of slip.

We present a compilation of 2131 high-fidelity mechanisms and centroid depths of moderate-magnitude earthquakes derived using synthetic body-waveform modeling (the Global Waveform-Modelled Earthquake Catalog v1.0—gWFM), which can be visualized and downloaded online (see Data and Resources). In this article, we describe the methods used to construct the gWFM and present a comparison between the earthquake depths and focal mechanisms in the gWFM with those derived by the International Seismological Centre, Global Centroid Moment Tensor (Global CMT) project, and the U.S. Geological Survey (USGS) W-phase, as well as 60 events studied using geodesy. We find that 20%–30% of the earthquakes in routine global catalogs have depths that differ by more than 10 km from those in the gWFM. Shallow-crustal earthquakes of Mw 5–6 are typically the worst located in depth by routine catalogs. Over 90% of the earthquakes in the gWFM are within ±15° in strike, ±5° in dip, and ±15° in rake of the Global CMT and USGS W-phase best double-couple moment tensor solutions. However, the mechanisms of shallow Mw 5–6 earthquakes in the routine catalogs can be inaccurate, due to the well-known insensitivity of long-period surface waves to the vertical dip-slip components of the moment tensor. The gWFM is an archive of well-constrained earthquake source parameters, though it will continue to update as new earthquake mechanisms and depths are published, thereby remaining an up-to-date research tool.