Conference Agenda

Coupling between tectonic plates in subduction zones can be evaluated using various geodetic data, but GNSS data are commonly used worldwide. While GNSS shines with its high temporal resolution, it only captures motion at a single point. In contrast, Sentinel-1 SAR data provides time series with higher spatial resolution and temporal resolution of up to 6 days, making it a valuable yet underutilized tool in determining surface displacement caused by seismic cycles in subduction zones worldwide.
In this study, we examine the Hikurangi subduction zone located in the North Island of New Zealand from November 2014 to December 2022. We processed three ascending track and one descending track to cover the whole North Island, from Auckland to Wellington. During this period, we observed both aseismic and seismic deformation, including the Kaikoura earthquake (Mw 7.8, November 2016), and the postseismic slip associated, and Manawatu’s slow slip events (Mw ~ 7 to 7.5, 2014-2015). We first show the utility of using InSAR data in geodetic inversions to resolve the interseismic coupling in as much detail as possible. We show using resolution tests that in a region where the GNSS network has a good spatial coverage, it is very difficult to capture small-scale locked asperities without InSAR.

We then focus on the inversion of InSAR and GNSS data to understand the impact of the Kaikoura crustal earthquake on the coupling. We investigate the inter-slow slip period between 2018 to 2022. By integrating InSAR velocity maps, our recovered coupling map for this time period highlights the significance of incorporating InSAR data to map out where the plate interface is coupled. Our results show a plate coupling consistent with the inter-slow slip model computed using only GNSS data (Wallace, 2020). We observe a strongly coupled area in the region of large slow slip events, but also near the trench.
Finally, we aim to explore how a coupling map using heterogeneous green’s function can differ from the standard approach using InSAR and GNSS data worldwide. We present one of the first model using InSAR and GNSS with heterogeneous green’s function to assess the coupling along a subduction zone.

The north-western Tibetan Plateau is characterized by a complex interaction between major fault systems, such as the left-lateral Altyn Tagh (ATF) and Karakax faults, the Longmu Gozha Co fault system (LGCF), which implies left-stepping en échelon fault strands connected by normal relay zones, as well as thrust faults at the northern front of the Western Kunlun (WK) range. The seismicity of the region is unevenly distributed and rather rare, with however the occurrence of several large earthquakes (> Mw 6) both on strike-slip and normal faults, and on blind thrusts sheets. GNSS-derived velocities are currently too sparse, with relatively high uncertainties, to constrain the regional kinematics and slip behavior of all individual faults. The present work is based on the Sentinel-1 radar data archive, and InSAR, and aims to better quantify the present-day fault kinematics and slip partitioning in this area.

We present an InSAR time-series analysis with a high spatial and temporal resolution to monitor the main active structures of this region over six years. This analysis is however challenging due to glaciers, which alter the coherence of the signal, and due to strong topographic gradients inducing atmospheric phase delays where tectonic deformation is expected. We rely on the ForM@Ter LArge-scale multi-Temporal Sentinel-1 InterferoMetry (FLATSIM, doi:10.24400/253171/FLATSIM2020) service (Thollard et al., 2021) to process the 5 ascending and 5 descending tracks covering our area and provide corresponding displacement time series in the line of sights. We use parametric signal decompositions in order to separate tectonic from non-tectonic signals. We then use spatial redundancy and complementary geometries to identify the horizontal and vertical components of each signal. We finally derive a regional linear velocity map representative of tectonic motions, masking the remaining non-tectonic signals.

These first InSAR-based velocities allow to discuss the partitioning of deformation and slip between the various fault systems, as well as the degree of locking of some of these active structures. While no tectonic signal could be unambiguously detected across the Karakax fault, several strike-slip strands within the ATF and LGCF systems could be identified, for the first time, as currently accumulating strain. Our InSAR-derived slip rates appear consistent with those derived from the few morphotectonic studies within their uncertainties (~ 5 mm/yr). Our results reveal lateral variations of tectonic loading and surface creep rates, as well as of locking depths along the ATF-LGCF system, highlighting a likely interplay between localized deformation near the ATF and LGCF junction, and more diffuse deformation near that of the LGCF and Karakorum fault. In the light of several recent morphotectonic studies, our results suggest that the LGCF may be the most recent and active western branch of the ATF.

The earthquake loading cycle is the repeated process of stress accumulation on a fault, and its subsequent release during earthquakes. Improved understanding of how strain evolves throughout the cycle in continental fault zones will allow us to have a better idea of their seismic hazard. These changes in strain at the surface of the earth can be measured using geodetic techniques, such as interferometric synthetic aperture radar (InSAR). Then, we can estimate the stress changes on the fault at depth through comparing these observations with dynamic forward models.

Previous studies have focussed on using geodesy, primarily GNSS & InSAR, to measure small fractions of the cycle of strain accumulation in major strike-slip fault zones, such as the North Anatolian and San Andreas Faults (Hussain et al., 2018), or on megathrusts (Ingleby et al., 2020). This observational bias towards long, large-offset, structurally mature faults is likely due to the large scale of their associated deformation, both in terms of spatial extent, and ease of observing associated surface deformation.

However, there have been fewer studies on smaller earthquakes. Therefore, current understanding of mechanisms occurring during the seismic cycle, and fault structure at depth, is based on observations of major continental-scale faults. These concepts may not necessarily be applicable to smaller-offset faults, particularly those with dip-slip geometry (Ingleby and Wright, 2017).

Here, we leverage a combination of legacy and contemporary SAR data to study the seismic cycle on previously-overlooked continental dip-slip faults. With data from three generations of SAR satellites, ERS, Envisat and Sentinel-1, we are currently constructing a 30-year time-series of coseismic and postseismic deformation following the 1995 Mw 6.5 normal-faulting Grevena earthquake, Greece. Early results suggest there has been postseismic afterslip, and we are searching for evidence of postseismic viscoelastic relaxation. I will build a dynamic model of the Grevena fault using the PyLith software (Aagard et al, 2013), and vary geometric and viscoelastic parameters to estimate the range of potential models which may explain the geodetic observations.

In future, we will combine geodetic observations and forward modelling for dip-slip faults in a range of global settings. This will allow us to constrain the structure of the fault, and assess the seismic hazard of dip-slip faults. I have compiled a list of different regions which could be interesting to compare, including normal faulting earthquakes in Greece, Turkey, Basin & Range (USA), the East African Rift, and Tibet.

In Northern Andes, oblique subduction of the Nazca plate below the South America Plate induces a northward motion of the North Andean Sliver, at a rate of ~10 mm/yr with respect to Stable South America. In Ecuador in particular, the associated strain is mainly accommodated along the large Chingual-Cosanga-Puna-Pallatanga (CCPP) fault system, which hosted several magnitude 7+ earthquakes in the historical period. A recent study using block-modeling of GNSS data (P. Jarrin, PhD) raises important questions about the partitioning and the localization of the deformation both inside and at the limits of the North-Andean sliver. Therefore, time-series analysis of Sentinel-1 InSAR data, at a 120m resolution, would complement the existing geodetic dataset of observation of low-rate crustal motions in this region.

Taking advantage of 7 to 8 years of Sentinel-1 archive using both ascending and descending tracks, we compute time-series of InSAR data for the whole Interandean region of Ecuador (~100 by 400 km), using the NSBAS processing chain. Because processing of InSAR data in this Equatorial region raises several challenges, such as low-coherence due to vegetation, ionospheric and tropospheric noise, and fading signals, we develop strategies to mitigate the noise terms. By using an optimized interferogram network, ECMWF-ERA5 weather model for tropospheric correction, improved weighting during multilooking using colinearity, and a temporal decomposition of the time-series, we produce the first InSAR velocity maps of the Ecuadorian Cordilleras.

Parallel analysis of GNSS time-series showed that this area is undergoing a strong post-seismic deformation due to the Pedernales megathrust earthquake (April 2016, Mw7.8). This post-seismic effect creates up to 5 mm/yr of extension across the cordilleras and hides the long-term interseismic strain we are interested in. To deal with this issue, we first extract InSAR velocity starting from mid-2017, when the non-linear post-seismic phase is over. We then reference the Sentinel-1 velocity maps to a 3-dimensional GNSS velocity field extracted on the same time-span. The two datasets show a consistency of about 1.4-1.8 mm/yr. Finally, we remove the interpolated GNSS post-seismic velocity field to obtain equivalent inter-seismic velocity fields for the post-2017 period. We then compare them to a block model derived from pre-2016 GNSS horizontal data, in order to characterize velocity gradients across active faults accommodating the motion of North Andean Sliver.

As earthquakes represent the releasing of strain, knowing how strain accumulated along active faults is essential for geodynamic and earthquake studies. In this research, we focus on mapping and modeling the strain field of the central-eastern segment of the Altyn Tagh Fault (ATF), which is one of the longest active strike-slip faults in the world and mainly accommodates the tectonic deformation between the Tibetan Plateau and the Tarim Basin, and the strain rate is calculated over a total area of ~ 600,000 km2 around the fault using both Sentinel-1 InSAR and GNSS data. We use the LiCSAR processing system to produce interferograms on 7 ascending tracks and 6 descending tracks, with nearly 180 epochs between October 2014 and July 2022 are used in each track. To reduce the impact of phase biases and nontectonic seasonal signals, we combine both short temporal (< 4 months) and 1-year to 7-year long summer-to-summer baseline interferograms in the network, which generates an average of nearly 2000 interferograms in each LiCSAR frame (a track includes 1 or 2 frames). We use the Generic Atmospheric Correction Online Service (GACOS) to reduce the tropospheric delay in the unwrapped phase. Time-series analysis is applied using LiCSBAS. We estimate 78 3D GPS velocities using the data measured during 1998-2021 from the Crustal Movement Observation Network of China-I/II and then solve for the best-fit model of strain rates for the central-eastern Altyn Tagh fault zone based on both InSAR and GNSS velocities. To understand how the strain is generated, we also model the strain field using both Bayesian inversion method and finite element method. Our results reveal significant variations in strain accumulation along the central-eastern ATF, which we think may have a close relationship with the active bends or stepovers along the fault. By comparing the strain rate distribution with the historical earthquake data, our result could provide important reference for future seismic risk assessment and earthquake prediction in this region. Additionally, by comparing with previous geodetic and geological investigation results, our study could bring some new thoughts and directions for future research about the ATF and other active faults.