Spectacular natural hazards such as large earthquakes or volcanic eruptions are accompanied by smaller-amplitude processes that produce millimeter-scale ground deformation. Although subtle, these signals provide critical insight into the physical state of the system and their associated hazards. Interferometric Synthetic Aperture Radar (InSAR) offers the spatial resolution required to observe such deformation, but its exploitation over long time spans remains challenging due to centimeter-scale noise and systematic biases. Here, we demonstrate how state-of-the-art interferometric processing combined with a Kalman Filter–based Time Series analysis (KFTS) enables the extraction of millimeter-scale deformation from 10 years of Sentinel-1A/B data. We present two case studies: the Chaman fault system (Pakistan–Afghanistan) and the Natron rift (northern Tanzania) in the East African Rift. Careful step-by-step corrections of the interferograms include tropospheric and ionospheric corrections, azimuth shift compensation, and rigorous assessment of closure phase biases. Measurement uncertainties derived from coherence are propagated within the KFTS time-series inversion, allowing iterative estimation of phase evolution with associated uncertainties.In the Chaman fault zone, we detect aseismic deformation characterized by fault creep rates of about 5 mm/yr, as well as a slow-slip event with ~1 cm of cumulative displacement resolved using combined ascending and descending geometries. In northern Tanzania, we resolve long-term rift opening of only a few millimeters per year between 2015 and 2025, consistent with GNSS campaign measurements. We further assess the potential of Independent Component Analysis (ICA) for InSAR signal separation and discuss current limitations imposed by residual noise and vegetation-related biases.

Groundwater extraction decreases water pressure in aquifer systems, causing reversible or irreversible deformation of the water-bearing layers that manifests as recoverable or permanent displacements of the land surface, respectively. Detecting and forecasting when and where an aquifer system transitions from a reversible, poroelastic regime, to an irreversible, inelastic regime remains a crucial challenge given the complex, heterogeneous nature of aquifer systems. Here we leverage high-resolution measurements of ground deformation and groundwater levels from 2016 to 2022 to characterize both regimes at the regional scale and show that a critical transition occurred in large areas of the Sacramento Valley during California's 2020-2022 extreme drought. Our analysis reveals that, while deformation remained primarily poroelastic during the 2016-2020 interdrought period, land subsidence in areas of intense groundwater extraction accelerated abruptly in 2021, with subsidence rates exceeding the inferred poroelastic rates by several decimeters per year. Such rapid and extensive land subsidence indicates severe inelastic compaction and loss of storage capacity of the underlying aquifer system, which pose a serious threat to California's water resources and infrastructure. A comparison of present-day deformation with historical groundwater levels reveals that this abrupt transition was not predictable based on the available groundwater records alone.

While it is well documented that continental extension involves discrete tectonic or magmatic rifting events, little is known about how deformation accumulates between these events. Here we focus on strain localization across the Natron Basin, which is part of the eastern branch of the East African Rift, that experienced a major tectono-magmatic event in 2007.A cross-rift profile of horizontal GNSS velocities (2013–2017) reveals a gradual transition between the rigid Tanzanian Craton and the Somalian Plate, with ~2 mm/yr of extension distributed across ~100 km (stretching zone). Such a pattern is commonly interpreted through the lens of dislocations in an elastic half-space. Here, an east-dipping border fault locked down to ~10 km may explain the observed width of the stretching zone, provided it extends to great depths, and creeps at a constant rate of ~3 mm/yr. The extent to which this is compatible with a hot lower crust riddled with magmatic intrusions is still debatable.We thus explore an alternative model where the width of the stretching zone is entirely determined by the history of past, finite deformation, and the corresponding ambient stress state. We use a 2-D thermo-mechanical model to stretch a visco-elasto-viscoplastic brittle layer, first creating a major border fault that slips continuously, flexing its footwall and hanging wall. We then artificially “lock” this fault by instantaneously strengthening it, drastically reduce our computational time steps, and continue stretching the layer. While the system should behave as an homogeneous, elastic layer under far-field extension, i.e., produce a linear displacement profile, we obtain an arctangent-shaped profile with a characteristic stretching zone width. This suggests that strain localization is controlled by the heterogeneous distribution of pre-existing stresses. Specifically, regions of high stresses that accrued during flexure of the fault blocks are brought to failure first during inter-event stretching, prompting the localization of elasto-plastic strain in a wide zone centered on the border fault. This process explains the width of velocity gradients in rift zones without invoking a deep, continuously creeping fault. We therefore suggest that long-term stress buildup plays a key role in short-term strain localization, and discuss its implications for active deformation in magma-rich continental rift settings like the Natron Basin.

Over the past two decades, InSAR evolved from the occasional processing of single interferograms over arid terrains to monitoring continuous time series of SAR acquisitions at the continental scale. Challenges, including atmospheric phase screen mitigation, automatic careful SAR image co-registration, ionospheric phase screen corrections, or discontinuous acquisition planning, were met through various technical and methodological advances by many research groups globally. The resulting methodologies now allow us to image a vast range of processes, from sudden large earthquakes to continuous subsidence involving metric to millimetric displacements. In addition to the ability to process datasets over continental scales, we can now measure natural signals of a few millimeters over distances lower than a kilometer. In recent years, we proposed technical solutions to issues that were seriously impeding our ability to measure small, millimeter-scale displacements over natural terrains. First, I will discuss the early development of tropospheric corrections using numerical weather models and highlight some of the most recent tools and methods stemming from there. Second, I will illustrate our approach to tackle the issue of continuously incoming SAR acquisitions, which we addressed by developing a data assimilation-based method involving a Kalman filter. This tool allows the rapid update of pre-existing time series of deformation as new SAR images are available while carefully propagating forward some of the uncertainties associated with time series analysis. Third, I will show how we handle the automatic denoising of InSAR time series using a fully convolutional neural network, allowing us to detect sub-millimeter tectonic fault slip with no prior knowledge of the faults. Fourth, I will present some recent developments about the effect of fading signals and time-dependent coherence evolution over temperate regions, depending on land covers. All these developments allowed us to image surface deformation processes, including several continuously creeping faults globally, transient tectonic slow slip events, intriguing post-seismic deformation signals, and strong subsidence patterns.

Along a plate boundary, why deformation and seismic hazard distributes across multiple active faults or along a single major structure remains unknown. The transpressive Chaman plate boundary (CPB) is a wide faulted region of distributed deformation and seismicity that accommodates the differential motion between India and Eurasia at 30 mm/year. However, main identified faults, including the Chaman fault, only accommodate 12 to 18 mm/year of relative motion and large earthquakes (Mw > 7) occurred east of them. We use Interfero-metric Synthetic Aperture Radar to locate the missing strain and identify active structures. The current displace-ment is partitioned between the Chaman fault, Ghazaband fault and a recent, immature but fast fault zone to the east. Such partitioning matches known seismic ruptures and results in the ongoing widening of the plate boundary, potentially controlled by the depth of the brittle-ductile transition. The CPB illustrates the impact of geological time scale deformation on today's seismic activity.

Slow, aseismic slip plays a crucial role in the initiation, propagation, and arrest of large earthquakes along active faults. In addition, aseismic slip controls the budget of elastic strain in the crust, hence the amount of energy available for upcoming earthquakes. The conditions for slow slip include specific material properties of the fault zone, pore fluid pressure, and geometrical complexities of the fault plane. Fine scale descriptions of aseismic slip at the surface and at depth are key to determine the factors controlling the occurrence of slow, aseismic versus rapid, seismic fault slip. We focus on the spatial and temporal distribution of aseismic slip along the North Anatolian Fault, the plate boundary accommodating the 2 cm/yr of relative motion between Anatolia and Eurasia. Along the eastern termination of the rupture trace of the 1944 M7.3 Bolu-Gerede earthquake lies a segment that slips aseismically since at least the 1950s. We use Sentinel 1 time series of displacement and GNSS data to provide a spatio-temporal description of the kinematics of fault slip. We show that aseismic slip observed at the surface is coincident with a shallow locking depth and that slow slip events with a return period of 2.5 years are restricted to a specific section of the fault. In the light of historical measurements, we discuss potential rheological implications of our results and propose a simple alternative model to explain the local occurrence of shallow aseismic slip at this location.

The Chaman plate boundary between India and Eurasia is a wide faulted region in Pakistan and Afghanistan, hosting distributed seismicity. Along the western edge of the deforming region, the Chaman fault currently accommodates less than 15 mm/yr of slip, while the differential left-lateral motion between both tectonic plates is close to 30 mm/yr. In the past century, significant earthquakes have ruptured structures east of the Chaman fault, including the 1931 Mach earthquake and 1935 Quetta earthquake with magnitudes (Mw) greater than seven. We aim to identify where strain focuses so that active structures likely to rupture in large earthquakes are outlined. We use ground velocities computed from 6 years-long InSAR time series in ascending and descending line of sights to map gradients of deformation in the Kirthar ranges. InSAR data reveals that most of the current plate boundary strain focuses east of the Chaman and Ghazaband fault in the central axis of the ranges. We model velocities along profiles across the plate boundary as the surface expression of left-lateral slip on several vertical faults: the Chaman fault, the subparallel Ghazaband fault, the Hoshab fault and one to three unknown faults to the east. We localise strain in the continuation of the Ornach Nal in the south and along the Quetta-Kalat fault which is thought to have hosted the 1935 Quetta earthquake (Mw 7.7). Three discrete portions of the Ghazaband fault slip with rates close to 10 mm/yr. Our description of partitioning matches known seismic ruptures, and makes sense in a geodynamical and geological perspective. We propose a tectonic model of the plate boundary evolution with an eastward migration of strain.

While some faults remain locked for tens to hundreds of years, some active faults slip slowly, either continuously or episodically. The discovery of slow, generally silent, slip at the turn of the century led to a profound modification of our understanding of the mechanics of faulting, shedding light on the dynamics of fault slip. Such dynamics areis controlled by the past history of stress along the fault plane (i.e. historical ruptures), fluids circulating in the crust and the rheology of the crust and fault plane. Understanding the influence of these different factors requires dense observations, as suggested by the large range of spatial and temporal scales involved in the control of the slip velocity along a fault. Specifically, the smallest scales of slow slip have beenwere inferred by the observation of tremors or low frequency events, interpreted as the chatter of a fault plane while it slips slowly. We are missing direct observations of such kilometer-scale slow slip events and continental creeping faults are an obvious target for such observationsfor such observations. Aseismic slip along the North Anatolian Fault was recognized in the 1960’s by the observation of offset man-made features without earthquakes recorded. Following these early observations, multiple geodetic studies focused on recording aseismic slip and analyzed the average rate of shallow slow slip in the vicinity of the town of Ismetpasa. GPS, InSAR and creepmeter data all converge toward an aseismic slip rate reaching 1 cm/yr in places, with significant along- strike variations. Furthermore, earlyHowever, creepmeter measurements in the 80’s, confirmed by records from a more recent instrument, suggest aseismic slip is currently episodic, occurring in bursts of slip. Recent InSAR data from the Cosmo-SkyMed constellation captured a month-long slow slip event with a maximum of 2 cm/yr of slip. We propose to analyze the geodetic record to search for slow slip events over the 2015-2020 period. We take advantage of a dense network of continuous GNSS stations installed in 2017 and of time series of Sentinel 1 SAR data to identify at least 3 slow slip events along the North Anatolian Fault. Thanks to the dense temporal sampling of the GNSS records, we describe faithfullyobserve the onset of slow slip. We use a deep learning algorithm to extract the surface signature of the slow slip events from the InSAR time series, highlighting a slow rupture front propagating along strike. We compare the occurrences of slow slip events with the local fault geometry, the average distribution of kinematic coupling and the historical seismicity. We discuss the mechanical implications of such detailed description of slow slip along an active fault. In conclusion, while slow slip rate averaged over periods longer than 2-3 years seems constant over the last 40 years, identification of slow slip events suggests this apparently constant rate results from slow slip events over multiple spatial and temporal scales.

Surface fault slip can be continuously monitored at fine spatial resolution from space using InSAR. Based on 5 years of observations (2014-2019), we describe and interpret the InSAR time series of deformation around the Chaman fault, a major strike-slip fault along the boundary between the Indian and Eurasian plates. Aseismic slip was observed on two >100 km long segments, reaching a maximum of 1 cm/yr. In between, a fault segment delimited by a restraining and releasing bend in the fault trace hosted three Mb 4.2, Mw 5.1 and Mw 5.6 earthquakes in our observation period. These earthquakes were followed by significant postseismic slip with characteristic duration between 1.5 to 3 years. Postseismic to coseismic surface slip ratios reach at least 0.6-1.2. In addition, aseismic slip was observed in close spatio-temporal relationship with those earthquakes. Finally, we argue that we detect numerous micro-slip events of Mw<3, although with large uncertainty. We provide an extensive description of the various modes of slip along this plate boundary fault and discuss the mechanical implications of such entangled behavior.

The 700‐km‐long Chaman fault (CF) marks the western edge of the plate boundary between India and Eurasia. Although global plate models predict 2.3–3.6 cm/yr left‐lateral motion between both plates, the fault is known to have hosted few earthquakes in historical times. Recent geodetic measurements attested the presence of aseismic slip locally. To detail the interplay between fast and slow slip along the CF, we build three Interferometric Synthetic‐Aperture Radar time series of ground deformation covering the whole fault length over 5 years (2014–2019). We find that most of the active fault trace slips aseismically and continuously. From south to north, we identify three creeping fault portions: the Nushki, Central, and Qalat segments of lengths between 80 and 130 km. The loading rate is 1.2 ± 0.3 cm/yr for the two southernmost portions, while it is about 0.7 ± 0.2 cm/yr for the Qalat segment. The Central segment and the nearby locked segments have hosted the largest known historical earthquakes on the CF, and three moderate magnitude earthquakes in our observation period. We image these earthquakes for which modeled slip at depth ( M w 5–5.6), time series of surface slip and deformation patterns argue toward large triggered aseismic slip. The June 2018 event displays postseismic moment 3–15 times greater than coseismic moment. Over the two decades covered by geodetic observations, continuous or triggered aseismic slip dominates along most of the fault and co‐locates with earthquakes. We observe that fault geometrical complexities delimit active segments and may be responsible for the kilometer‐scale intertwining between seismic and aseismic events.

Systematically characterizing slip behaviours on active faults is key to unraveling the physics of tectonic faulting and the interplay between slow and fast earthquakes. Interferometric Synthetic Aperture Radar (InSAR), by enabling measurement of ground deformation at a global scale every few days, may hold the key to those interactions. However, atmospheric propagation delays often exceed ground deformation of interest despite state-of-the art processing, and thus InSAR analysis requires expert interpretation and a priori knowledge of fault systems, precluding global investigations of deformation dynamics. Here, we show that a deep auto-encoder architecture tailored to untangle ground deformation from noise in InSAR time series autonomously extracts deformation signals, without prior knowledge of a fault’s location or slip behaviour. Applied to InSAR data over the North Anatolian Fault, our method reaches 2 mm detection, revealing a slow earthquake twice as extensive as previously recognized. We further explore the generalization of our approach to inflation/deflation-induced deformation, applying the same methodology to the geothermal field of Coso, California.

Earth orbiting satellites, such as Sentinel 1A-B, build up an ever-growing set of synthetic aperture radar images of the ground. This conceptually allows for real-time monitoring of ground displacements using Interferometric Synthetic Aperture Radar (InSAR), notably in tectonically active regions such as fault zones or over volcanoes. We propose a Kalman filter for InSAR time series analysis (KFTS), an efficient method to rapidly update preexisting time series of displacement with data as they are made available, with limited computational cost. KFTS solves together for the evolution of phase change with time and for a parametrized model of ground deformation. Synthetic tests of the KFTS reveal exact agreement with the equivalent weighted least squares solution and a convergence of descriptive model parameter after the assimilation of about 1 year of data. We include the impact of sudden deformation events such as earthquakes or slow slip events on the time series of displacement. First tests of the KFTS on ENVISAT data over Mt. Etna (Sicily) and Sentinel 1 data around the Chaman fault (Afghanistan, Pakistan) show precise (0.05 mm) retrieval of phase change when data are sufficient. Otherwise, the optimized parametrized model is used to forecast phase change. Good agreement is found with classic time series analysis solution and GPS-derived time series. Accurate estimates are conditioned to the proper parametrization of errors so that models and observations can be combined with their respective uncertainties. This new tool is freely available to process ongoing InSAR time series.

The San Andreas Fault creeping section is generally considered as slipping continuously and aseismically, at a rate of about 35 mm/yr. However, recent studies, using either Global Positioning System (GPS) network or Interferometric Synthetic Aperture Radar (InSAR) data, have highlighted spatial and temporal variations of slip rate. Here, we combine GPS, InSAR, creepmeter and seismicity data over the 2008-2018 period, taking advantage of their complementary spatial and temporal resolutions, to detail a comprehensive picture of episodic acceleration and deceleration slip patterns. For this purpose, we use a variational Bayesian Independent Component Analysis (vbICA) decomposition to separate geodetic deformation due to non-tectonic sources from signals of tectonic origin. The fault slip kinematics is reconstructed by linear inversion of each Independent Component related to transient tectonic activity. We document aseismic slip acceleration transients and discuss their origin.