Conference Agenda

The Southern Peninsula of Haiti, in the Caribbean region, has been the locus of two ~Mw 7 earthquakes in the last decade. This peninsula is sliced by the Enriquillo Plantain Garden Fault, a plate boundary structure commonly considered as a purely strike-slip fault system separating the North American plate from the Caribbean plate. Paleoseismological studies and geodynamic reconstructions suggest this plate boundary accommodates 7±2 mm/yr of left-lateral motion. However, the moment released during both the Léogane (Mw 7.1, 2010) and Nippes (Mw 7.2, 2021) earthquakes suggest the EPGF actually accommodates a significant amount of North-South shortening which is not yet accounted for in seismic hazard assessment. Both earthquakes show a dip over strike slip ratio of about one third, consistent with the latest GNSS velocity field and resulting block model. It therefore appears now quite clearly that the EPGF is in fact a transpressive system over which slip partitions between strike and dip slip motion. The question of the amount of partitioning over the different phases of the earthquake cycle remains unknown.

We draw an overview of deformation across this transpressional fault zone throughout the earthquake cycle. We use geodetic methods to infer surface displacement over the inter-, co- and post-seismic period that follows the Nippes 2021 Mw 7.2 earthquake. We use GNSS and Sentinel-1 A/B Synthetic Aperture Radar images from 2014 to 2021 to derive combined time series and a velocity field characterizing the long-term deformation pattern over the interseismic period. We then compare how this period of active strain accumulation relates to the surface deformation resulting from large earthquakes and moment released along the plate boundary using coseismic geodetic observations from the ALOS-2 and Sentinel-1 SAR satellites and GNSS data. We confirm partitioning is consistent between the interseismic and coseismic periods for the last two major seismic events.

In both the 2010 and 2021 earthquakes, we do not observe rupture of the main Enriquillo Plantain Garden Fault, but rather the rupture of secondary fault structures parallel to the main plate boundary. Such ruptures on secondary structures may not preclude any larger event on the main Enriquillo Plantain Garden Fault. From our analysis, we observe that the Enriquillo Plantain Garden Fault only slipped aseismically during the postseismic phase of both earthquakes. Using blind source separation method (Independent Component Analysis) on InSAR time series we show that the post-seismic deformation following both earthquakes occurred on the Miragoâne segment, a seismic gap between the 2010 and 2021 earthquakes. Besides, we also measure surface slip along an handful of other secondary mapped faults following the 2021 earthquake. We observe the time dependent behavior of post-seismic slip along these structures, consistent with the frictional behavior of rate-strengthening materials diffusing the coseismic stress perturbation. We quantify the rheological constitutive properties of the materials of each of these faults and discuss these results in the light of available geologic records.

Our results illuminate the complexity of the Enriquillo Plantain Garden Fault and associated active faults in Haiti over a fraction of the earthquake cycle, deriving key elements for a better understanding of long-term deformation in the area and for better evaluation of seismic hazard.

Velocity mapping is important for understanding how mountain belts develop. The focus of this study is Tianshan, a 2500 km-long 300 km-wide orogenic belt that was reactivated in the late Cenozoic as a far-field response to the Indo-Eurasia collision. Bounded between the rigid Tarim Basin in the south and Kazakh Platform and Junggar Basin in the north, the Tianshan mountain range absorbs a significant portion of the Indo-Eurasia oblique convergence through faulting and folding in the foreland thrust systems, intermontane basin bounding faults and conjugate strike-slip faulting that facilitates block rotation. However, how the compressional and shearing strain is partitioned between these different tectonic structures and accommodated at the different stages of a seismic cycle remains unclear, making it difficult to discriminate between end-member dynamic models governing the regional tectonics and understand the seismic hazard local populations are exposed to.

In this study, we produce the first large-scale high-resolution InSAR velocity field over the entire Tianshan mountain range to illuminate fault kinematics and improve our understanding the regional tectonics and seismic hazard. We process 7 years of Sentinel-1 data acquired between 2014 and 2022 using the automatic LiCSAR system to generate over 90,000 interferograms at 500 m resolution for 90 LiCS frames (38 ascending and 51 descending frames) that cover a total area of 2,280,000 km². We perform time series analysis on dense networks of interferograms with temporal baselines between 6 days to 1 year using the LiCSBAS software. We develop within LiCSBAS a method to detect coregistration errors and use both triplet loop errors and time series residuals to detect unwrapping errors. We further develop strategies to automatically correct, mask and discard interferograms, improving data quality across low-coherence areas. Our final networks contain only strongly connected interferograms with over 60 percent pixel coverage and at least 24 days of temporal baselines that are visually checked to be correctly unwrapped. The final networks have on average >600 interferograms per frame and have root-mean-square time series residuals under 1 radian.

We combine the InSAR velocities and compiled 2D and 3D GNSS velocities from a range of literature to generate 3D velocity fields across the entire Tianshan mountain belt. This high-resolution velocity field allows us to map active faults, compare InSAR, GNSS and geologically derived slip rates on faults, and analyse strain partitioning between frontal boundary fold-and-thrust belts, and the intermountain ranges. We also use this data set to probe the limit of the geodetic detection of surface creep by modelling velocity profiles across speculated creeping structures, hence discussing the contribution of seismic and aseismic strain accumulation during the interseismic period. By decomposing the time series into linear and seasonal variations and by correlation with land cover classification, we interpret the tectonic, hydrological, climatic and anthropogenic contributions to the vertical velocity field. These results help us better understand the seismic hazard over Tianshan and how this young plateau grows and expands under tectonic stresses.

On 21 November 2022, a MW 5.6. earthquake hit Cianjur area in West Java, Indonesia. Surprisingly, the moderate magnitude earthquake claimed 602 people’s lives, and about 167,000 people were taking shelter in refugee camps as of 30 December 2022 according to the BPBD (Regional Disaster Management Agency) of Cianjur District, Indonesia. The fatality of this event, once normalized by the amount of energy released, is 22 times larger than the Mw7.8 February 2023 Türkiye-Syria earthquake, 98 times larger than the Mw7.5 September 2018 Sulawesi earthquake and tsunami in Indonesia, and 134 times larger than the Mw7.8 April 2015 Gorkha earthquake in Nepal.

The Earth Observatory of Singapore – Remote Sensing Lab (EOS-RS) rapidly produced maps of coseismic deformation and surface disturbance (a.k.a. Damage Proxy Map, DPM) using Synthetic Aperture Radar (SAR) data. The first SAR image was acquired by ALOS-2 satellite operated by the Japan Aerospace Exploration Agency (JAXA), 11 hours after the earthquake. We generated the first DPM and disseminated it (21 hours from the data acquisition) with response agencies through Sentinel Asia network. We also engaged multi-temporal interferometric coherence analysis and produced a DPM using SAR data acquired by the Copernicus Sentinel-1 satellite operated by the European Space Agency (ESA).

The earthquake occurred on a previously unknown/unmapped fault. We combine and analyze ground observations, coseismic deformation and surface disturbance maps derived from ALOS-2 and Sentinel-1 SAR data, seismic waveforms of the mainshock and aftershocks, high-rate GNSS data, and tiltmeter data to characterize the source parameters of the earthquake and damage caused by the strong ground motion, landslides, and liquefaction, and study the potential impact of the event on the geohazards of the area.

One of the challenges of the impact and hazard analysis is that the coseismic displacements appeared as a double-couple point source even in Interferometric SAR (InSAR) observations, so it is difficult to identify the orientation of the fault plane as opposed to the auxiliary plane based on the spatial pattern of the ground displacements. Thus, we designed an inverse problem where we optimally balance the contribution of the interferograms (one from ascending stripmap mode and two from descending ScanSAR mode), so we can produce the optimal best-fit model for both cases.

Earthquakes occur when stresses on faults overcome friction resistance. Although we cannot map stress state on faults directly, satellite geodesy now gives us powerful tools to measure the slow accumulation of tectonic strain in deforming zones. Accurate maps of this strain accumulation can help constrain the spatial distribution and rates of recurrence of future earthquakes. We combine data from Sentinel-1 InSAR with sparse GNSS velocities to create the first high-resolution velocity and strain rate models for the Alpine-Himalayan Belt, stretching from Türkiye to China.

The Alpine-Himalayan Belt is a broad deforming zone accommodating distributed deformation caused by the collision of Africa, Arabia and India with Eurasia. Three quarters of the earthquakes since 1901 that killed more than 10,000 people occurred in the Alpine-Himalayan belt. Across this zone, faults are often poorly mapped and ground-based geodetic measurements of deformation from GNSS systems (such as the Global Positioning System) can be too sparse to associate strain with individual fault structures.

We process Sentinel-1 InSAR data acquired in 651 ascending and descending frames, each covering an area typically ~250x250 km, using the automatic COMET-LICSAR system (Lazecký et al. 2020). We produce small-baseline networks that typically contain the shortest 4 connections to every epoch but are augmented where appropriate by longer time-span pairs. We invert for average line-of-sight (LOS) velocities and time series at ~1 km resolution using LiCSBAS (Morishita et al. 2020), correcting for coseismic displacements for earthquakes larger than M5.5. In total, we have processed (at time of writing) data from over 131,000 acquisitions in the Alpine-Himalayan Belt, creating over 500,000 interferograms.

The LOS velocities are initially in a local reference frame that is unique for each frame. We convert these into a unified Eurasian reference frame using a compilation of GNSS data over the region. To achieve this, we use the Velmap code (Wang and Wright 2012) to carry out a joint inversion in each of several regions (Turkey/Caucasus, Iran, Afghanistan, Tien Shan, Tibet) for (i) 3D surface velocities on a triangular mesh and (ii) reference frame adjustment parameters for each InSAR frame. We calculate strain rates directly from the inverted velocity field model. In addition, we use the LOS data converted into a Eurasian reference frame to directly invert for East-West and vertical velocities at the 1 km resolution of our LOS velocities, for pixels where we have ascending and descending data available. The east-west velocities for the Tibetan region are shown in the attached figure.

Our velocity and strain rate fields reveal a variety of behaviours across the region. Vertical velocities at shorter wavelength are dominated by non-tectonic processes such as water extraction. Horizontal (east-west) velocities show concentrations of strain around major faults in regions like Anatolia and Tibet, with particularly high strain rates in locations that have experienced major earthquakes in the past 30 years, such as the location of the 2001 M7.9 Kokoxili earthquake in Tibet. In other regions, notably in Iran, strain appears more diffuse. We will end the presentation by discussing the implications of the results for our understanding of how the continents deform and for seismic hazard assessment.

References

Lazecký, Milan, Karsten Spaans, Pablo J. González, Yasser Maghsoudi, Yu Morishita, Fabien Albino, John Elliott, Nicholas Greenall, Emma Hatton, Andrew Hooper, Daniel Juncu, Alistair McDougall, Richard J. Walters, C. Scott Watson, Jonathan R. Weiss, and Tim J. Wright. 2020. 'LiCSAR: An Automatic InSAR Tool for Measuring and Monitoring Tectonic and Volcanic Activity', 12: 2430.

Morishita, Yu, Milan Lazecky, Tim J Wright, Jonathan R Weiss, John R Elliott, and Andy Hooper. 2020. 'LiCSBAS: An Open-Source InSAR Time Series Analysis Package Integrated with the LiCSAR Automated Sentinel-1 InSAR Processor', Remote Sensing, 12: 424.

Wang, H, and TJ Wright. 2012. 'Satellite geodetic imaging reveals high strain away from major faults of Western Tibet', Geophys. Res. Lett., 39: L07303.

Geodetic measurements of surface deformation can provide crucial constraints on a region’s tectonics and seismic hazard. To do so effectively, they need to be spatially dense (enough to highlight individual faults), spatially extensive (enough to capture the entirety of strain signals), temporally dense (enough that noise and nuisances can be understood), temporally extensive (enough to bring out gradual interseismic deformation), and accurate. A combination of InSAR and GNSS data is arguably the first data form that can be all five of these. In the Anatolia-Caucasus region, we are using Sentinel-IA InSAR frame velocities from the COMET LiCS system and pairing them with a high-quality GNSS velocity field to generate high-resolution maps of crustal deformation and strain rate. We find that the North Anatolian Fault is the dominant feature in the strain accumulation field, but also resolve deformation coinciding with other tectonic structures in Anatolia and throughout the Caucasus. To compare these strain rates with earthquake occurrence rates, we assemble an integrated earthquake catalogue for the region that covers many hundred years, and assess whether the moment release in earthquakes has kept up with the moment accumulation rates implied by our strain maps, focusing in particular on the North Anatolian and East Anatolian faults.