A document by Jérémie Vasseur. Click on the document to view its contents.

A document by Pierre St-Laurent. Click on the document to view its contents.

The generation and propagation of Ultra Low Frequency (ULF) waves are intrinsically coupled to the cold plasma population in the terrestrial magnetosphere. During geomagnetic storms, extreme reconfigurations of the cold plasma creates a complex and dynamic system that drastically modifies this coupling. The extent and manner in which this coupling is affected remains an open question. In this report, we assess the coupling between ULF waves and cold plasmaspheric plumes during geomagnetic storms, and investigate the implications for ULF wave-driven radial transport of the outer radiation belt population. We present a series of event studies of Van Allen Probes observations. For each event, we use inferred measurements of the cold plasma density during plume crossings, in combination with magnetic and electric field observations of ULF waves. The event studies show very different, and at times contrasting, wave behaviour. This includes events where ULF waves appear to be spatially confined within plume structures. Initial estimates show that the localised patches of ULF wave power have significant implications for radial diffusion processes, and highlights the need for caution in estimating radial diffusion coefficients. We suggest that the cold plasma dynamics is an important source of uncertainty in radial diffusion models, and understanding cold plasma-ULF wave coupling is a critical area of future investigations.

The southeastern margin of the Tibetan Plateau has undergone complex deformation since the Cenozoic, resulting in a high level of seismicity and seismic hazard. Knowledge about the seismic anisotropy provides important insight about the deformation mechanism and the regional seismotectonics beneath this tectonically active region. In this study, we conduct fullwave multi-scale tomography to investigate the seismic anisotropy in the southeastern margin of the Tibetan Plateau. Broadband records at 111 permanent stations in the region from 470 teleseismic events are used to obtain 5,216 high-quality SKS splitting intensity measurements, which are then inverted in conjunction with 3D sensitivity kernels to obtain an anisotropic model with multi-scale resolution. Resolution tests show that our dataset recovers anisotropy anomalies reasonably well on the scale of 1º x 1º horizontally and ~100 km vertically. Our result suggests that in the southeastern margin of the Tibetan Plateau the deformation in the lithosphere and asthenosphere are decoupled. The anisotropy in the lithosphere varies both laterally and vertically as a result of dynamic interactions of neighboring blocks as well as lithospheric reactivation. The anisotropy in the asthenosphere largely follows the direction of regional absolute plate motion. The SKS splittings observed at the surface are shown to be consistent with the vertical integral of our depth-dependent anisotropy model over lithospheric and asthenospheric depths.

The 2022 eruption at Mauna Loa, Hawai‘i, marked the first extrusive activity from the volcano after 38 years of quiescence. The eruption was preceded by several years of seismic unrest in the vicinity of the volcano’s summit. Characterizing the structure and dynamics of seismogenic features within Mauna Loa during this pre-eruptive interval may provide insights into how pre- and co-eruptive processes manifest seismically at the volcano. In particular, the extent to which seismicity may be used to forecast the location and timing of future eruptions is unclear. To address these questions, we construct a catalog of relocated seismicity on Mauna Loa spanning 2011-2023. Our earthquake locations image complex, sub-kilometer-scale seismogenic structures in the caldera and southwest rift zone. We additionally identify a set of streaks of seismicity in the volcano’s northwest flank that are radially oriented about the summit. Using a rate-and-state friction model for earthquake occurrences, we demonstrate that the seismicity rate in this region can be modeled as a function of the stressing history caused by magma accumulation beneath the summit. Finally, we observe a mid-2019 step change in the seismicity rate in the Ka‘oiki region that may have altered the stress state of the northeast rift zone in the three years before the eruption. Our observations provide a framework for interpreting future seismic unrest at Mauna Loa.

Seismic waves contain information about the earthquake source, the geologic structure they traverse, and many forms of noise. Separating the noise from the earthquake signal is a critical first step in seismic waveform analysis. This is, however, a difficult task because optimal parameters for filtering noise typically vary with time and, if chosen inappropriately, they may strongly alter the original seismic waveform. Diffusion models based on Deep Learning (DL) have demonstrated remarkable capabilities in the restoration of images and audio signals. However, those models assume a Gaussian distribution of noise, which is not the case for typical seismic noise. Diffusion models trained on Gaussian noise do not perform well in seismic applications; therefore, we introduce a "cold" variant of diffusion models in which both clean and noisy seismic traces are restored. Here, we describe the first Cold Diffusion Model for Seismic Denoising (CDiffSD), including key design aspects, model architecture, and noise handling. We demonstrate that CDiffSD provides a new standard in performance, outperforming existing methods. Our model provides a significant advance for seismic data denoising and establishes a new state-of-the-art in the field.

Magnetotail earthward-propagating fast plasma flows provide important pathways for magnetosphere-ionosphere coupling. This study reexamines a flow-related red-line diffuse-like aurora event previously reported by Liang et al., (2011), utilizing THEMIS and ground-based auroral observations from Poker Flat. We find that time domain structures (TDSs) within the flow bursts efficiently drive electron precipitation below a few keV, aligning with predominantly red-line auroral intensifications in this non-substorm event. The diffuse-like auroras sometimes coexisted with or potentially evolved from discrete forms. We forward model red-line diffuse-like auroras due to TDS-driven precipitation, employing the time-dependent TREx-ATM auroral transport code. The good correlation (0.77) between our modeled and observed red line emissions underscores that TDSs are a primary driver of the red-line diffuse-like auroras, though whistler-mode wave contributions are needed to fully explain the most intense red-line emissions.

We revisit the nature of the ocean bottom pressure (OBP) seasonal cycle by leveraging the mounting GRACE-based OBP record and its assimilation in the ocean state estimates produced by the project for Estimating the Circulation and Climate of the Ocean (ECCO). We focus on the mean seasonal cycle from both data and ECCO estimates, examining their similarities and differences and exploring the underlying causes. Despite substantial year-to-year variability, the 21-year period studied (2002–2022) provides a relatively robust estimate of the mean seasonal cycle. Results indicate that the OBP annual harmonic tends to dominate but the semi-annual harmonic can also be important (e.g., subpolar North Pacific, Bellingshausen Basin). Amplitudes and short-scale phase variability are enhanced near coasts and continental shelves, emphasizing the importance of bottom topography in shaping the seasonal cycle in OBP. Comparisons of GRACE and ECCO estimates indicate good qualitative agreement, but considerable quantitative differences remain in many areas. The GRACE amplitudes tend to be higher than those of ECCO typically by 10%–50%, and by more than 50% in extensive regions, particularly around continental boundaries. Phase differences of more than 1 (0.5) months for the annual (semiannual) harmonics are also apparent. Larger differences near coastal regions can be related to enhanced GRACE data uncertainties and also to the absence of gravitational attraction and loading effects in ECCO. Improvements in both data and model-based estimates are still needed to narrow present uncertainties in OBP estimates.

Using 2D numerical subduction models, we compare deep slab behaviour with oceanic and continental overriding plates and a mantle viscosity structure where the lower mantle viscosity jump occurs either at 660 km or at 1000 km depth as suggested by the latest geoid inversions. We demonstrate that a strong, thick, and buoyant continental plate, combined with a 1000 km depth viscosity increase, promotes slab penetration into the lower mantle. Conversely, the same slab will deflect at 660 km depth if this subducts under an oceanic plate into a mantle where the viscosity increases at the canonical 660 km depth. To quantify these dynamics, we introduce a slab bending ratio, by dividing the deep slab tip angle by the shallow slab angle, reflecting the steepness, and sinking history of the slab. Ocean-ocean convergence models with a viscosity increase coincident with the phase transition at 660 km depth have low ratios and flattened slabs comparable to ocean-ocean cases in nature (e.g., Izu-Bonin). Coupling a continental overriding plate with a 1000 km depth viscosity increase separate from the endothermic phase change results in slabs with high ratio values, and stepped morphologies similar to that observed for the Nazca plate beneath the Southern Peruvian arc. Our results highlight that slab morphologies ultimately express the interaction between the type of overriding plate, slab-induced flow, and phase transitions, modulated by the viscosity structure of the top of the lower mantle and transition zone.

In active volcanic systems, the elevated pressurization of fluids and the movement of melt materials have an enormous influence on the stress-state of rocks and their mechanical behavior. We use seismic ambient noise to evaluate the static seismic velocity variations related to long-term volcanic deformation, and the dynamic changes associated with the 2017 Casamicciola earthquake (Mw 3.9), in the active volcanic complex of the Ischia Island (Italy). Our study reveals a significant dynamic velocity reduction mostly related to the near-surface damage, with a permanent drop near the red zone, that we posit to be related to the documented landslides and the subsidence observed immediately after the earthquake. We also report a positive long-term linear trend of velocity variations, sensitive to a generalized contraction of the Ischia Caldera that we revealed with geodetic modeling. Our results suggest a depressurization of the shallow hydrothermal system, through degassing along faults or sills.

The effect of mantle plumes is secondary to that of subducting slabs for modern plate tectonics, e.g. when considering plate driving forces. However, the impact of plumes on tectonics and planetary surface evolution may nonetheless have been significant. We use numerical mantle convection models in a 3-D spherical chunk geometry with damage rheology to study some of the potential dynamics of plume-slab interactions. Substantiating our earlier work which was restricted to 2-D geometries, we observe a range of interesting plume dynamics, including plume-driven subduction terminations, even though the new models allow for more realistic flow. We explore such plume-slab interactions, including in terms of their geometry, frequency, and the overall effect of plumes on surface dynamics as a function of the fraction of internal to bottom heating. Some versions of such plume-slab interplay may be relevant for geologic events, e.g. for the inferred ~183 Ma Karoo large igneous province formation and associated slab disruption. More recent examples may include the impingement of the Afar plume underneath Africa leading to disruption of the Hellenic slab, and the current complex structure imaged for the subduction of the Nazca plate under South America. Our results imply that plumes may play a significant role not just in kick-starting plate tectonics, but also in major modifications of slab-driven plate motions, including for the present-day mantle.

A document by Alamgir Hossan. Click on the document to view its contents.

The development of the field of geophysics is shown through the most frequent words in AGU journal article titles. Early on, the most frequent scientific words in titles were on Earth's magnetism and the effect that the sun, Earth's core, and Earth's atmosphere had on Earth's magnetism. After the 1940s, the most frequent words in titles expanded to include hydrology, water resources, oceans, and climate. The field of geophysics, as seen through the most frequent words, was also influenced by technological advances and sociopolitical events.

1. We estimate Walker Lane fault slip rates using a dense filtered and gridded geodetic velocity field and a robust multi-block model approach 2. The geodetic slip rates are independent of geologic slip rates, but 80% agree with them to within uncertainties 3. The method images off-fault deformation and vertical axis rotations providing more insight into how crustal motion drives earthquakes Hammond et al., 2023

Using Unmanned Aerial Systems (UAS) equipped with optical RGB cameras and Doppler radar, surface velocity can be efficiently measured at high spatial resolution. UAS-borne Doppler radar is particularly attractive because it is suitable for real-time velocity determination, because the measurement is contactless, and because it has fewer limitations than image velocimetry techniques. In this paper, five cross-sections (XSs) were surveyed within a 10 km stretch of Rönne Å in Sweden. Ground-truth surface velocity observations were retrieved with an electromagnetic velocity sensor (OTT MF Pro) along the XS at 1 m spacing. Videos from a UAS RGB camera were analyzed using both Particle Image Velocimetry (PIV) and Space-Time Image Velocimetry (STIV) techniques. Furthermore, we recorded full waveform signal data using a Doppler radar at multiple waypoints across the river. An algorithm fits two alternative models to the average amplitude curve to derive the correct river surface velocity: a Gaussian one peak model, or a Gaussian two peak model. Results indicate that river flow velocity and propwash velocity caused by the drone can be found in XS where the flow velocity is low, while the drone-induced propwash velocity can be neglected in fast and highly turbulent flows. To verify the river flow velocity derived from Doppler radar, a mean PIV value within the footprint of the Doppler radar at each waypoint was calculated. Finally, quantitative comparisons of OTT MF Pro data with STIV, mean PIV and Doppler radar revealed that UAS-borne Doppler radar could reliably measure the river surface velocity.

Deep learning (DL) phase picking models have proven effective in processing large volumes of seismic data, including successfully detecting earthquakes missed by other standard detection methods. Despite their success, the applicability of existing extensively-trained DL models to high-frequency borehole datasets is currently unclear. In this study, we compare four established models (GPD, U-GPD, PhaseNet and EQTransformer) trained on regional earthquakes recorded at surface stations (100 Hz) in terms of their picking performance on high-frequency borehole data (2000 Hz) from the Preston New Road (PNR) unconventional shale gas site, in the United Kingdom (UK). The PNR-1z dataset, which we use as a benchmark, consists of continuously recorded waveforms containing over 38,000 seismic events previously catalogued, ranging in magnitudes from -2.8 to 1.1. Remarkably, three of the four DL models recall a good fraction of the events and two might satisfy the monitoring requirements of some users without any modifications. In particular, PhaseNet and U-GPD demonstrate exceptional recall rates of 95% and 76.6%, respectively, and detect a substantial number of new events (over 15,800 and 8,300 events, respectively). PhaseNet’s success might be attributed to its exposure to more extensive and diverse instrument dataset during training, as well as its relatively small model size, which might mitigate overfitting to its training set. U-GPD outperforms PhaseNet during periods of high seismic rates due to its smaller window size (400-samples compared to PhaseNet’s 3000-sample window). All models start missing events below Mw -0.5, suggesting that the models could benefit from additional training with microseismic datasets. Nonetheless, PhaseNet may satisfy some users' monitoring requirements without further modification, detecting over 52,000 events at PNR. This suggests that DL models can provide efficient solutions to the big data challenge of downhole monitoring of hydraulic-fracturing induced seismicity as well as improved risk mitigation strategies at unconventional exploration sites.

Peatlands are fundamental deposits of organic carbon. Thus, their protection is of crucial importance to avoid emissions from their degradation. Peat is a mixture of organic soil that originates from the accumulation of wetland plants under continuous or cyclical anaerobic conditions for long periods. Hence, a precise quantification of peat deposits is extremely important; for that, remote- and proximal-sensing techniques are excellent candidates. Unfortunately, remote-sensing can provide information only on the few shallowest centimeters, whereas peatlands often extend to several meters in depth. In addition, peatlands are usually characterized by difficult (flooded) terrains. So, frequency-domain electromagnetic instruments, as they are compact and contactless, seem to be the ideal solution for the quantitative assessment of the extension and geometry of peatlands. Generally, electromagnetic methods are used to infer the electrical resistivity of the subsurface. In turn, the resistivity distribution can, in principle, be interpreted to infer the morphology of the peatland. Here, to some extent, we show how to shortcut the process and include the expectation and uncertainty regarding the peat resistivity directly into a probabilistic inversion workflow. The present approach allows for retrieving what really matters: the spatial distribution of the probability of peat occurrence, rather than the mere electrical resistivity. To evaluate the efficiency and effectiveness of the proposed probabilistic approach, we compare the outcomes against the more traditional deterministic fully nonlinear (Occam's) inversion and against some boreholes available in the investigated area.  

It has been said, arguably, that causality analysis should pave a promising way to interpretable deep learning and generalization. Incorporation of causality into artificial intelligence (AI) algorithms, however, is challenged with its vagueness, non-quantitiveness, computational inefficiency, etc. During the past 18 years, these challenges have been essentially resolved, with the establishment of a rigorous formalism of causality analysis initially motivated from atmospheric predictability. This not only opens a new field in the atmosphere-ocean science, namely, information flow, but also has led to scientific discoveries in other disciplines, such as quantum mechanics, neuroscience, financial economics, etc., through various applications. This note provides a brief review of the decade-long effort, including a list of major theoretical results, a sketch of the causal deep learning framework, and some representative real-world applications in geoscience pertaining to this journal, such as those on the anthropogenic cause of global warming, the decadal prediction of El Niño Modoki, the forecasting of an extreme drought in China, among others.

Southern Alaska is a collage of fault-bounded accreted terranes. The deformation history of these crustal blocks and geometric history of the bounding faults reflect both inherited features and subsequent convergent margin events. Multiple dense (<20-km spacing) arrays of broadband seismometers across southern Alaska has previously allowed for imaging of crustal structure across the region using various seismic imaging methods. Here, we employ S-toP receiver functions to investigate the crustal structure of southern Alaska for signals of dynamic tectonic activity. The subduction zone plate interface and subducting slab Moho are imaged dipping at shallow (<60-km) depths across the southernmost part of the subduction zone. Along two different transects, an inboard-dipping (~15°) boundary is imaged intersecting the trace of the Border Ranges Fault at the surface that we infer represents an unrotated inboard-dipping paleosubduction (Mesozoic) interface. This observation is combined with previous seismic imaging along both the Border Ranges Fault and the next seaward terrane-bounding fault-the Contact Fault-to buttress a known history of convergent tectonics that varies along the margin. Along with large (>10-km) crustal thickness offsets imaged across both the Denali Fault system and the Manuscript in press at upcoming AGU Monograph Tectonics and Seismicity of Alaska and Western Canada: Earthscope and Beyond 2 Eureka Creek Fault, this feature supports a Mesozoic-to-Present inboard-dipping (east and northward) subduction polarity in the region. Additionally, the Sp CCP volume reveals a 100-km x 50-km sized positive velocity gradient with depth (PVG) at ~25-km depth beneath the Copper River Basin, which we interpret as the top of a region of active underplating and/or intrusion of basaltic magmatism into the lower crust. This feature may be related to the generation of a new Wrangell Volcanic Field volcano, resulting from the underlying tear in the subducting slab.

Here we analyze the rupture process of the December 29 th , 2020 MW6.4 Petrinja earthquake (Croatia), the largest event recorded in this area characterized by a moderate strain-rate intraplate setting. We use foreshocks and aftershocks, recorded at more than 80 broadband stations located 70km to 420km from the earthquake, as empirical Green's functions (EGFs) to separate source effects from propagation and local site effects. First, we deconvolve the mainshock P-wave time windows from the EGFs in the frequency domain to obtain the corner frequency (fc). Spectral analysis based on a Brune's source model reveals a large stress drop of 24 MPa. Next, by deconvolving the Love waves in the time domain, we calculate the Apparent Source Time Functions (ASTFs). We find that the average duration of the source is ~5 s, with no significant directivity effects, indicating a bilateral rupture. To extract physical rupture parameters such as rupture velocity, slip distribution and rise time, we deploy two techniques: (1) Bayesian inversion and (2) backprojection onto isochrones of ASTFs. Both techniques show a low rupture velocity (40-50% of the shear wave velocity) and a rupture length of less than 10 km, i.e. much less than would typically be expected for a magnitude 6.4 earthquake. This apparent anticorrelation between stress drop and rupture velocity may be attributed to the complex and segmented fault system characteristic of immature intraplate settings.

In anticipation to substitute the existing manual/semi-automated methods for classifying quarry blasts, earthquakes, and noise, we developed three convolutional neural network (CNN) models. The three CNN models extract relevant features from seismograms (waveform), spectrograms (spectrum), and a combination of the two respectively. A total of 3414 samples were extracted from the three categories, 15% of the data from each category were split for testing, and the remaining data were augmented and used for training. The waveform model, spectrogram model, and combined model achieved accuracies of 95.32%, 93.13%, and 93.96%, respectively. The reliability of these models was ascertained by promising accuracies of >90% and 100% obtained for large and small datasets from testing with SCEDC data and records from the Palitana region (Gujarat) respectively. The results of this study demonstrate the potential of deep learning-based approaches for the effective classification of seismic events.

Cryoseismic studies are increasingly being used to measure intraglacial deformation, reliant on lab observations that sheared ice crystals create seismic ‘fast’ directions at predictable orientations to the flow direction. However, icy materials are often in contact with liquid phases that modify seismic properties of the aggregate. Previous studies describing seismic anisotropy in temperate ice considered how melt affects the orientation of solid ice, but not the orientation of the melt itself. We simulated microstructural shear deformation of partially molten ice with variable melt orientations, and calculated resultant seismic properties. Our results demonstrate that ≤ 3.5% 3D melt concentration changes the fast direction of a deforming icy aggregate, and higher degrees can completely overprint the solid-induced fast direction. Melt orientation is thus a key control on the seismic anisotropy of deforming partially molten ice, and solid-based methods may incorrectly estimate the magnitude and direction of subsurface flow in temperate icy bodies.

We hindcasted the seismicity rates and the next largest earthquake magnitude using seismic and hydraulic data from two hydraulic stimulation campaigns carried out in adjacent (500 m apart) ultra-deep wells in Finland. The two campaigns performed in 2018 and 2020 took place in the frame of St1 Helsinki project produced stable, pressure-controlled induced seismic activity with maximum magnitudes of MW 1.3 and 1.7, respectively. The seismicity rates were modeled using simplified physics-based approaches tailored to varying injection rates. This is the first time that this framework was applied to a cyclical injection protocol. The next largest earthquake magnitude was estimated using several existing models from the literature. Despite the close proximity of the two hydraulic stimulations and associated seismicity, we obtained strongly different parameterization of the critical model components, questioning the use of a-priori seismic hazard analysis tools in the planning of a neighboring stimulation. The differences in parameterization were attributed to the contrasting hydraulic energy rates observed in each stimulation, small differences in the structural inventory of the reservoir and resulting seismic injection efficiency, and potentially to variations in the injection protocol itself. As far as the seismicity rate model is concerned, despite a good performance during the 2018 campaign, the fit during the 2020 stimulation was suboptimal. Forecasting the next largest magnitude using different models led to a very wide range of outcomes. Moreover, their relative ranking across stimulations was inconsistent, including the situation whether the best performing model in 2018 stimulation was the worst performing one in the 2020 stimulation.

The shallow portion of a megathrust represents the zone of first contact between two colliding plates, and its rheological properties control the seismic and tsunami hazards generated by the fault. Unfortunately, underwater geodetic observations are sparse due to the high cost of obtaining geodetic data, meaning limited information is available on the interseismic behavior of this part of most megathrusts. The Rakhine-Bangladesh megathrust offers a unique opportunity to probe the behavior of the shallow megathrust as it is the only ocean-continent subduction zone where the near-trench region is fully accessible on land. Here, we use observations from ALOS-2 wide-swath imagery spanning 2015 to 2022 to conduct an InSAR timeseries analysis of the overriding plate within Bangladesh and the Indo-Myanmar Ranges. We identify a narrow pattern of alternating uplift and subsidence associated with mapped anticlines but show that it cannot be explained by plausible rates of slip on the megathrust or other fault structures. Instead, we argue that the deformation is likely caused by active aseismic folding within the wedge above a shallow decollement. We show that estimates of the decollement depth derived from a viscous folding model and the observed anticline spacing are in agreement with previous seismic observations of the decollement depth across the fold belt. We suggest that the role of ductile deformation in the overriding plate in subduction zones may be more important than previously recognized.