Search

SUMMARYEarthquakes often rupture across more than one fault segment. If such rupture segmentation occurs on a significant scale, a simple point-source or one-fault model may not represent the rupture process well. As a consequence earthquake characteristics inferred, based on one-source assumptions, may become systematically wrong. This might have effects on follow-up analyses, for example regional stress field inversions and seismic hazard assessments. While rupture segmentation is evident for most Mw > 7 earthquakes, also smaller ones with 5.5 < Mw < 7 can be segmented. We investigate the sensitivity of globally available data sets to rupture segmentation and their resolution to reliably estimate the mechanisms in presence of segmentation. We focus on the sensitivity of InSAR (Interferometric Synthetic Aperture Radar) data in the static near-field and seismic waveforms in the far-field of the rupture and carry out non-linear and Bayesian optimizations of single-source and two-sources kinematic models (double-couple point sources and finite, rectangular sources) using InSAR and teleseismic waveforms separately. Our case studies comprises of four Mw 6–7 earthquakes: the 2009 L’Aquila and 2016 Amatrice (Italy) and the 2005 and 2008 Zhongba (Tibet) earthquakes. We contrast the data misfits of different source complexity by using the Akaike informational criterion (AIC). We find that the AIC method is well suited for data-driven inferences on significant rupture segmentation for the given data sets. This is based on our observation that an AIC-stated significant improvement of data fit for two-segment models over one-segment models correlates with significantly different mechanisms of the two source segments and their average compared to the single-segment mechanism. We attribute these modelled differences to a sufficient sensitivity of the data to resolve rupture segmentation. Our results show that near-field data are generally more sensitive to rupture segmentation of shallow earthquakes than far-field data but that also teleseismic data can resolve rupture segmentation in the studied magnitude range. We further conclude that a significant difference in the modelled source mechanisms for different segmentations shows that an appropriate choice of model segmentation matters for a robust estimation of source mechanisms. It reduces systematic biases and trade-off and thereby improves the knowledge on the rupture. Our study presents a strategy and method to detect significant rupture segmentation such that an appropriate model complexity can be used in the source mechanism inference. A similar, systematic investigation of earthquakes in the range of Mw 5.5–7 could provide important hazard-relevant statistics on rupture segmentation. In these cases single-source models introduce a systematic bias. Consideration of rupture segmentation therefore matters for a robust estimation of source mechanisms of the studied earthquakes.

The Bayesian earthquake analysis tool (BEAT) is an open-source Python software to conduct source-parameter estimation studies for crustal deformation events, such as earthquakes and magma intrusions, by employing a Bayesian framework with a flexible problem definition. The software features functionality to calculate Green’s functions for a homogeneous or a layered elastic half-space. Furthermore, algorithm(s) that explore the solution space may be selected from a suite of implemented samplers. If desired, BEAT’s modular architecture allows for easy implementation of additional features, for example, alternative sampling algorithms. We demonstrate the functionality and performance of the package using five earthquake source estimation examples: a full moment-tensor estimation; a double-couple moment-tensor estimation; an estimation for a rectangular finite source; a static finite-fault estimation with variable slip; and a full kinematic finite-fault estimation with variable hypocenter location, rupture velocity, and rupture duration. This software integrates many aspects of source studies and provides an extensive framework for joint use of geodetic and seismic data for nonlinear source- and noise-covariance estimation within layered elastic half-spaces. Furthermore, the software also provides an open platform for further methodological development and for reproducible source studies in the geophysical community.

Inferring the geometry and evolution of an earthquake sequence is crucial to understand how fault systems are segmented and interact. However, structural geological models are often poorly constrained in remote areas and fault inference is an ill-posed problem with a reliability that depends on many factors. Here, we investigate the geometry of the Mw 6.3 2008 and 2009 Qaidam earthquakes, in northeast Tibet, by combining InSAR time series and teleseismic data. We conduct a multi-array back-projection analysis from broadband teleseismic data and process three overlapping Envisat tracks covering the two earthquakes to extract the spatio-temporal evolution of seismic ruptures. We then integrate both geodetic and seismological data into a self-consistent kinematic model of the earthquake sequence. Our results constrain the depth and along-strike segmentation of the thrust-faulting sequence. The 2008 earthquake ruptured a ∼32∘ north-dipping fault that roots under the Olongbulak pop-up structure at ∼12 km depth and fault slip evolved post-seismically in a downdip direction. The 2009 earthquake ruptured three south-dipping high-angle thrusts and propagated from ∼9 km depth to the surface and bilaterally along the south-dipping segmented 55–75∘ high-angle faults of the Olonbulak pop-up structure that displace basin deformed sedimentary sequences above Paleozoic bedrock. Our analysis reveals that the inclusion of the post-seismic afterslip into modelling is beneficial in the determination of fault geometry, while teleseismic back-projection appears to be a robust tool for identifying rupture segmentation for moderate-sized earthquakes. These findings support the hypothesis that the Qilian Shan is expanding southward along a low-angle décollement that partitions the oblique convergence along multiple flower and pop-up structures.

How an earthquake rupture propagates strongly influences the potentially destructive ground shaking. Complex ruptures often involve slip along multiple faults, which masks information on the frictional behaviour of fault zones. Geometrically smooth ocean transform fault plate boundaries offer a favourable environment to study fault dynamics, because strain is accommodated along a single, wide fault zone that offsets the homogeneous geology. Here we present an analysis of the 2016 Mw 7.1 earthquake on the Romanche fracture zone in the equatorial Atlantic, using data from both nearby seafloor seismometers and global seismic networks. We show that this rupture had two phases: (1) upward and eastward propagation towards a weaker region where the transform fault intersects the mid-ocean ridge, and then (2) an unusual back-propagation westwards at a supershear speed towards the centre of the fault. We suggest that deep rupture into weak fault segments facilitated greater seismic slip on shallow locked zones. This highlights that even earthquakes along a single distinct fault zone can be highly dynamic. Observations of back-propagating ruptures are sparse, and the possibility of reverse propagation is largely absent in rupture simulations and unaccounted for in hazard assessments. In one earthquake, an oceanic transform fault ruptured in one direction and then backwards at a speed exceeding that of shear-wave propagation, according to an analysis of data recorded by nearby seafloor and global seismometers.

An 8 year archive of Envisat synthetic aperture radar (SAR) data over a 300 × 500 km2 wide area in northwestern Tibet is analyzed to construct a line‐of‐sight map of the current surface velocity field. The resulting velocity map reveals (1) a velocity gradient across the Altyn Tagh fault, (2) a sharp velocity change along a structure following the base of the alluvial fans in southern Tarim, and (3) a broad velocity gradient, following the Jinsha suture. The interferometric synthetic aperture radar velocity field is combined with published GPS data to constrain the geometry and slip rates of a fault model consisting of a vertical fault plane under the Altyn Tagh fault and a shallow flat décollement ending in a steeper ramp on the Tarim side. The solutions converge toward 0.7 mm/yr of pure thrusting on the décollement‐ramp system and 10.5 mm/yr of left‐lateral strike‐slip movement on the Altyn Tagh fault, below a 17 km locking depth. A simple elastic dislocation model across the Jinsha suture shows that data are consistent with 4–8 mm/yr of left‐lateral shear across this structure. Interferometric synthetic aperture radar processing steps include implementing a stepwise unwrapping method starting with high‐quality interferograms to assist in unwrapping noisier interferograms, iteratively estimating long‐wavelength spatial ramps, and referencing all interferograms to bedrock pixels surrounding sedimentary basins. A specific focus on atmospheric delay estimation using the ERA‐Interim model decreases the uncertainty on the velocity across the Tibet border by a factor of 2.

The 24 January 2020 Mw 6.77 Elazığ-Sivrice earthquake (Turkey), responsible for 42 casualties and ~ 1600 injured people, is the largest earthquake affecting the East Anatolian Fault (EAF) since 1971. The earthquake partially ruptured a seismic gap. The mainshock was preceded by two foreshocks with Mw ≥ 4.9 and small seismicity clusters occurring in the previous months close to the nucleation point of the main rupture. The significant aftershock sequence comprises twelve earthquakes with Mw ≥ 4.5 within 60 days. We jointly model quasi co-seismic static surface displacements from Interferometric Synthetic Aperture Radar (InSAR) and high-frequency co-seismic data from seismological networks at local, regional and teleseismic distances to retrieve source parameters of the mainshock. We reconstruct the rupture process using a Bayesian bootstrap based probabilistic joint inversion scheme to obtain source parameters and their uncertainties. Full moment tensor for 18 fore-/after-shocks with Mw ≥ 4.3 are obtained based on the modeling of regional broadband data. The posterior mean model for the 2020 Elazığ-Sivrice mainshock shows that the earthquake, with a magnitude Mw 6.77, ruptured at shallow depth (5 ± 2 km) with a left-lateral strike-slip focal mechanism, with a dip angle of 74° ± 2° and a causative fault plane strike of 242° ± 1°, which is compatible with the orientation of the EAF at the centroid location. The rupture nucleated in the vicinity of small foreshock clusters and slowly propagated towards WSW, with a rupture velocity of ~ 2100 ± 130 m s−1 and ~ 27 s rupture duration. The main rupture area, with a length of ~ 26 ± 5 km, only covered 70 % of the former seismic gap, leaving a smaller, unbroken segment of ~ 30 km length to the SE with positive stress change. The subsequent aftershock sequence extended over a broader region of ~ 70 km in length, spreading to both sides of the mainshock rupture patch into the regions experiencing a stress increase according to our Coulomb stress modeling. Our results support the hypothesis of a shallow locking depth of the Anatolian micro-plate, which has a possible implication to the seismic bursts along the EAF and alternating seismic activity on the North Anatolian and the East Anatolian faults.

The Pamir orogen, Central Asia, is the result of the ongoing northward advance of the Indian continent causing shortening inside Asia. Geodetic and seismic data place the most intense deformation along the northern rim of the Pamir, but the recent 7 December 2015, Mw7.2 Sarez earthquake occurred in the Pamir's interior. We present a distributed slip model of this earthquake using coseismic geodetic data and postseismic field observations. The earthquake ruptured an ∼80 km long, subvertical, sinistral fault consisting of three right-stepping segments from the surface to ∼30 km depth with a maximum slip of three meters in the upper 10 km of the crust. The coseismic slip model agrees well with en echelon secondary surface breaks that are partly influenced by liquefaction-induced mass movements. These structures reveal up to 2 m of sinistral offset along the northern, low-offset segment of modeled rupture. The 2015 event initiated close to the presumed epicenter of the 1911 Mw∼7.3 Lake Sarez earthquake, which had a similar strike-slip mechanism. These earthquakes highlight the importance of NE trending sinistral faults in the active tectonics of the Pamir. Strike-slip deformation accommodates shear between the rapidly northward moving eastern Pamir and the Tajik basin in the west and is part of the westward (lateral) extrusion of thickened Pamir plateau crust into the Tajik basin. The Sarez-Karakul fault system and the two large Sarez earthquakes likely are crustal expressions of the underthrusting of the northwestern leading edge of the Indian mantle lithosphere beneath the Pamir.

The northern Tien Shan of Kyrgyzstan and Kazakhstan has been affected by a series of major earthquakes in the late 19th and early 20th centuries. To assess the significance of such a pulse of strain release in a continental interior, it is important to analyze and quantify strain release over multiple time scales. We have undertaken paleoseismological investigations at two geomorphically distinct sites (Panfilovkoe and Rot Front) near the Kyrgyz capital Bishkek. Although located near the historic epicenters, both sites were not affected by these earthquakes. Trenching was accompanied by dating stratigraphy and offset surfaces using luminescence, radiocarbon, and 10Be terrestrial cosmogenic nuclide methods. At Rot Front, trenching of a small scarp did not reveal evidence for surface rupture during the last 5000 years. The scarp rather resembles an extensive debris-flow lobe. At Panfilovkoe, we estimate a Late Pleistocene minimum slip rate of 0.2 ± 0.1 mm/a, averaged over at least two, probably three earthquake cycles. Dip-slip reverse motion along segmented, moderately steep faults resulted in hanging wall collapse scarps during different events. The most recent earthquake occurred around 3.6 ± 1.3 kyr ago (1σ), with dip-slip offsets between 1.2 and 1.4 m. We calculate a probabilistic paleomagnitude to be between 6.7 and 7.2, which is in agreement with regional data from the Kyrgyz range. The morphotectonic signals in the northern Tien Shan are a prime example of deformation in a tectonically active intracontinental mountain belt and as such can help understand the longer-term coevolution of topography and seismogenic processes in similar structural settings worldwide.

Understanding the mechanisms by which earthquake cycles produce folding and accommodate shortening is essential to quantify the seismic potential of active faults and integrate aseismic slip within our understanding of the physical mechanisms of the long‐term deformation. However, measuring such small deformation signals in mountainous areas is challenging with current space‐geodesy techniques, due to the low rates of motion relative to the amplitude of the noise. Here we successfully carry out a multitemporal Interferometric Synthetic Aperture Radar analysis over the North Qaidam fold‐thrust system in NE Tibet, where eight Mw> 5.2 earthquakes occurred between 2003 and 2009. We report various cases of aseismic slip uplifting the thickened crust at short wavelengths. We provide a rare example of a steep, shallow, 13‐km‐long and 6‐km‐wide afterslip signal that coincides spatially with an anticline and that continues into 2011 in response to a Mw 6.3 event in 2003. We suggest that a buried seismic slip during the 2003 earthquake has triggered both plastic an‐elastic folding and aseismic slip on the shallow thrusts. We produce a first‐order two‐dimensional model of the postseismic surface displacements due to the 2003 earthquake and highlight a segmented slip on three fault patches that steepen approaching the surface. This study emphasizes the fundamental role of shallow aseismic slip in the long‐term and permanent deformation of thrusts and folds and the potential of Interferometric Synthetic Aperture Radar for detecting and characterizing the spatiotemporal behavior of aseismic slip over large mountainous regions., Journal of Geophysical Research: Solid Earth, 124 (7), ISSN:2169-9313, ISSN:0148-0227, ISSN:2169-9356

International audience; An increasing number of observations supports temporal clustering behaviour of earthquakes within fault systems. As earthquake occurrence is mainly controlled by the crustal stresses, it is crucial to determine their spatio-temporal evolution to understand the generation of catastrophic seismic sequences. A possible way to constrain these variations is to measure the surface displacement field induced by seismic sequences. However, the observation time of modern satellite geodesy (InSAR/GPS) is short compared to the duration of an earthquake sequence. Thus, the goal of this paper is to extend the temporal range of observations of a seismic sequence. We focus on the largest earthquakes of the 1936–1997, Northeast Lut, Iran, sequence that is composed of 11 Mw ≧ ∼ 6 earthquakes. Using subpixel correlation of historic (KH9) and recent (Sentinel-2) optical satellite images, we measure for the first time the surface displacement field of the 1979 Mw 7.1 Khuli-Boniabad earthquake, which broke the eastern part of the Dasht-e-Bayaz fault. Using subpixel correlation of optical (SPOT2-4) and SAR (JERS-2) images, we also measure the surface displacement field of the Mw 7.2 Zirkuh earthquake, which ruptured the Abiz fault. We found that both earthquakes have a mean slip of 2.5 m but the Khuli-Boniabad earthquake broke two main segments (total rupture ∼ 60 km), whereas the Zirkuh earthquake broke three main segments (total rupture ∼ 125 km). We suggest that the differences are controlled by the maturity of the faults, the Dasht-e-Bayaz fault being less mature than the Abiz fault. Furthermore, we succeed to measure offsets up to 2.60 m for the 1979 Mw 6.6 Korizan earthquake that broke the northern part of the Abiz fault. It is the first time that the surface displacement field for such a small historic earthquake has been measured using optical correlation. Finally, our study confirms the potential of historical optical imagery for retrieving surface displacements for past earthquakes (pre-modern geodesy era).

In permafrost areas, seasonal freeze-thaw cycles result in upward and downward movements of the ground. For some permafrost areas, long-term downward movements were reported during the last decade. We measured seasonal and multi-year ground movements in a yedoma region of the Lena River Delta, Siberia, in 2013–2017, using reference rods installed deep in the permafrost. The seasonal subsidence was 1.7 ± 1.5 cm in the cold summer of 2013 and 4.8 ± 2 cm in the warm summer of 2014. Furthermore, we measured a pronounced multi-year net subsidence of 9.3 ± 5.7 cm from spring 2013 to the end of summer 2017. Importantly, we observed a high spatial variability of subsidence of up to 6 cm across a sub-meter horizontal scale. In summer 2013, we accompanied our field measurements with Differential Synthetic Aperture Radar Interferometry (DInSAR) on repeat-pass TerraSAR-X (TSX) data from the summer of 2013 to detect summer thaw subsidence over the same study area. Interferometry was strongly affected by a fast phase coherence loss, atmospheric artifacts, and possibly the choice of reference point. A cumulative ground movement map, built from a continuous interferogram stack, did not reveal a subsidence on the upland but showed a distinct subsidence of up to 2 cm in most of the thermokarst basins. There, the spatial pattern of DInSAR-measured subsidence corresponded well with relative surface wetness identified with the near infra-red band of a high-resolution optical image. Our study suggests that (i) although X-band SAR has serious limitations for ground movement monitoring in permafrost landscapes, it can provide valuable information for specific environments like thermokarst basins, and (ii) due to the high sub-pixel spatial variability of ground movements, a validation scheme needs to be developed and implemented for future DInSAR studies in permafrost environments.

International audience; In this study, we reestimate the source model of the 1997 Mw 7.2 Zirkuh earthquake (northeastern Iran) by jointly optimizing intermediate‐field Interferometry Synthetic Aperture Radar data and near‐field optical correlation data using a two‐step fault modeling procedure. First, we estimate the geometry of the multisegmented Abiz fault using a genetic algorithm. Then, we discretize the fault segments into subfaults and invert the data to image the slip distribution on the fault. Our joint‐data model, although similar to the Interferometry Synthetic Aperture Radar‐based model to the first order, highlights differences in the fault dip and slip distribution. Our preferred model is ∼80° west dipping in the northern part of the fault, ∼75° east dipping in the southern part and shows three disconnected high slip zones separated by low slip zones. The low slip zones are located where the Abiz fault shows geometric complexities and where the aftershocks are located. We interpret this rough slip distribution as three asperities separated by geometrical barriers that impede the rupture propagation. Finally, no shallow slip deficit is found for the overall rupture except on the central segment where it could be due to off‐fault deformation in quaternary deposits.

Volcanoes commonly inflate or deflate during episodes of unrest or eruption. Continuum mechanics models that assume linear elastic deformation of the Earth’s crust are routinely used to invert the observed ground motions. The source(s) of deformation in such models are generally interpreted in terms of magma bodies or pathways, and thus form a basis for hazard assessment and mitigation. Using discontinuum mechanics models, we show how host-rock fracturing (i.e. non-elastic deformation) during drainage of a magma body can progressively change the shape and depth of an elastic-deformation source. We argue that this effect explains the marked spatio-temporal changes in source model attributes inferred for the March-April 2007 eruption of Piton de la Fournaise volcano, La Reunion. We find that pronounced deflation-related host-rock fracturing can: (1) yield inclined source model geometries for a horizontal magma body; (2) cause significant upward migration of an elastic-deformation source, leading to underestimation of the true magma body depth and potentially to a misinterpretation of ascending magma; and (3) at least partly explain underestimation by elastic–deformation sources of changes in sub-surface magma volume.