Yuqing Xie:Modelling for possible mechanisms of the compressional branching in the 2012 Off-Sumatra earthquake The 2012 Off-Sumatra earthquake is the largest recorded intra-plate strike-slip earthquake. It occurred in the diffuse deformation zone in the Indo-Australia Plate, rupturing multiple oblique fault segments. Back-projection studies of the 2012 Mw 8.6 Off-Sumatra earthquake reveal a counter-intuitive rupture pattern that two episodes of rupture branching into compressional fault segments with elevated dynamic normal stress, leaving the dilatational segments unbroken or broken with a delay. To explain the compressional branching, we propose several hypotheses: (1) the stress is lowered on the dilatational branch by a recent event, (2) a low frictional coefficient, and (3) the influence of fault geometry. We check these hypotheses by conducting quasi-dynamic simulations of earthquake cycles and a one-way coupled dynamic simulation using a rate-and-state friction law. We reproduce the compressional branching when the angle between two faults is 75 degrees and the static frictional coefficient is 0.2, due to a combination of the three mechanisms. /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Tian Feng: Combining CNN and RNN in Seismic Phase Picking: The recent expansion of seismic data and computing resources enables flourishing applications of deep learning in seismology. Many studies aim at automatically picking P and S arrivals, especially those buried in noises. Dozens of deep-learning-based models prove to be efficient in detecting phases of local events (<300km). Most of them take seismograms/spectrograms as input data and output the probability of P/S/background phases. Previous studies treat inputs as images and focus on the Convolutional Neural Network (CNN). CNNs employ filters within convolutional layers to extract features from inputs, and are often used in computer vision to recognize objects and patterns in images. Recent studies notice that input seismograms/spectrograms are sequential data, more like audio. Recurrent Neural Networks (RNN) are designed to interpret temporal or sequential information, which is widely used in speech recognition. However, RNNs cannot be stacked into very deep models, and training an RNN is a very difficult task. We solve this problem by extracting features from a pre-trained CNN model and then training an RNN on top of it. Compared with models trained directly from CNNs (PhaseNet and GPD), our hybrid model has higher accuracy and precision in the testing set. Besides, the hybrid model is more robust against low SNR of waveforms and far hypocentral distances, which indicates better generalization ability.
Abijah Simon: Cenozoic wedge tectonics as a crustal thickening mechanism in eastern Tibet A fundamental issue in studies of the eastern Tibet plateau involves clarifying the mechanism of crustal deformation that has resulted in its present-day crustal thickness of over 50 km (Wang et al., 2010; Xu et al., 2016). The two main endmember models debated for the Cenozoic construction of eastern Tibet are (1) the thrust belt model, which attributes crustal thickening to brittle shortening of the upper crust, and (2) the channel flow model, which attributes crustal thickening to vertical inflation of the mid-lower crust with only minor upper-crustal shortening. In this study, we combine detailed geologic field mapping and seismic reflection profile interpretation to create a new pure-shear wedge tectonic model that can explain the structural evolution of this region. In our model, an westward-tapering thrust wedge with Moho-offsetting discrete shear zones is responsible for the crustal thickening and uplift in the northern part of eastern Tibet. This model may also explain along-strike variations in geology and geomorphology along the eastern margin. /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Ashley Schoenfeld: Stressing Out: A look at the stress regime of Enceladus’ south pole Enceladus is a small moon of Saturn (radius ~250 km), distinguished by its uniquely active south pole. The moon’s geological activity is expressed as cyclic eruption of plumes sourced from a series of parallel ‘‘tiger-stripe” fractures (TSF) (Porco et al., 2006). The cyclic nature of the plume’s eruption has been attributed to diurnal variations of tidal stresses acting on the moon. There is, however, a mismatch in timing between Cassini’s observations of peak eruption and what is predicted by the theory of tidally modulated cracks (Nimmo et al., 2014); as well, the tiger stripes erupt throughout the entirety of its orbit around Saturn, even when the cracks are predicted to be under compression (Nimmo et al., 2014). To explore these discrepancies, we assume that the peak eruption time occurs under the condition of transtension and investigate the total stress as a result of both tidal and tectonic stresses. We find tectonically derived stresses to be non-trivial when compared to the tidal-stresses, dominating the stress field at the south pole. Similarly, we find the sum of these stresses to be transtensional in nature, suggesting that tectonic stresses may explain why the tiger stripe fractures deviate from the behavior predicted by a purely tidal model.
Justin Higa: Incipient breakup of the Isla Ángel de la Guarda microcontinent, Gulf of California, México Faults on microcontinents record the dynamic evolution of plate boundaries. However, most microcontinents are submarine and difficult to study. Here, we show that the southern part of the subaerial Isla Ángel de la Guarda (IAG) microcontinent is the likely site of an incipient plate boundary reorganization in the Gulf of California that may lead to future microcontinent breakup. To characterize the kinematics of this reorganization, we integrated remote fault mapping using high-resolution satellite- and drone-based topography with neotectonic field-mapping and 13 luminescence ages from sediment deposits offset or trapped by faults. Onshore, N-S striking normal faults occur along strike of a nascent offshore spreading center in the North Salsipuedes Basin, west of IAG. Late Pleistocene and Holocene luminescence ages indicate onshore fault activity in the last ~50 ka. These observations imply that the North Salsipuedes Basin is kinematically linked with active faults onshore IAG. Thus, crustal extension across southern IAG may connect to and reactivate extinct plate boundary structures east of IAG in the Upper Tiburón basin and play an important role in reorganizing the Pacific-North America plate boundary. ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
Tong Zhou: Array-based seismic waveform coherency measurement, simulation, and application in evaluating back-projections Array-based seismic source imaging methods, e.g., back-projections, are routinely applied to the large earthquake investigation and earthquake hazard assessment. The back projection (BP) method locates the high frequency energy radiators in the fault region with the alignment and stacking of the coherent seismic phases, which represents the fault rupturing process. However, what the BP radiators actually images is still under debate, along with the accuracy and resolution of the BPs, especially when waveform complexity presents. Two candidates that are sensitive to BPs include the slip acceleration and the rupture speed change. To test the BP methods and distinguish what fault properties the BP most sensitive to, we need to simulate the real incoherency waveforms. In this work, we first measure the coherence fluctuation with time, frequency and interstation distance in an array with moderate size earthquakes which are supposed to be treated as point sources at teleseismic distance. Then we propose two methods: (1) multiple plane waves and (2) multiple Born scatterers to fit the fluctuation of coherence statistically. Such semi-analytical methods capture the statistical features of the incoherent coda waves, and save significant computation time compared to numerical simulations. With this method, we are able to test the resolvability of various issues including earthquake nucleation, bilateral propagation, fault jumping, supershear transition, etc. and evolve the methods for improving BP images, e.g., slowness calibration, artifacts mitigation, and even to build possible relationships between asperity sizes and coherence fluctuations. /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Erik Weidner: Radial Anisotropy in the Indian Ocean from Higher Mode Surface Waves and a Hierarchical Transdismensional Approach A fully non-linear transdimensional hierarchical Markov Chain Monte Carlo approach was developed and applied to fundamental and higher mode Love and Rayleigh wave dispersion data to constrain radial anisotropy in the Indian Ocean. We obtained three-dimensional tomographic models of shear-wave velocity and anisotropy with quantitative uncertainties down to the mantle transition zone. We compared these models to results from regularized linear inversions of the same data set obtained with and without prior constraints on 410- and 660-discontinuities topography. We found that the undulations of these discontinuities had little effect on the resulting models. We also compared results from non-linear joint and separate inversions of Love and Rayleigh waves and tested the effect of depth parametrization on the models.
The Importance of Paleobathymetry in Understanding the Long-Term Carbon Cycle through Variations in Carbonate Compensation Depths throughout the Last 100Myr Seafloor spreading and cooling of oceanic lithosphere results in a constantly evolving bathymetry. First order changes occurred throughout the Cenozoic as plates speeds slowed by a factor of 2 and major plates reorganized. We evaluate the role of period-accurate bathymetry distributions at global and basin scales on the carbonate compensation depth (CCD). To analyze the effects of bathymetry on the carbon cycle we focus here on the Late Paleocene. We find a strong bathymetric dependence on the CCD at global and basin scale levels using the LOSCAR earth system model. Steady states snapshots at 60 Ma reveal that the Indian, Pacific, and Atlantic basin CCDs are ~1km deeper than previous estimates, while the Tethys CCD is over 2km deeper. Variations in the initial riverine flux, an uncertain climate parameter, potentially reconciles global CCD predictions with ocean core sample data. Our study demonstrates the need to reconcile the interpretative climate parameters used within climate modeling with realistic bathymetric reconstructions. The addition of evolving bathymetry proves to be necessary when studying the long-term climate and carbon cycles through models such as LOSCAR, GEOCLIM, and DCESS. Consequently, chosen bathymetry reconstructions need to be rationalized for, and amongst, climate studies to improve interpretations of these cycles.
Jewel Abbate: Models of thermal vs. compositional convection in Earth’s core Buoyancy-driven convection in Earth’s liquid outer core is sourced from both thermal and compositional variations induced by the crystallization of the inner core. The turbulent fluid motion resulting from these variations is the primary power source of Earth’s dynamo, and thus the primary focus in both numerical and laboratory studies for generating dynamo models. Most often, these models are built upon a framework accounting for only thermal buoyancy as a source of convective turbulence, but recent measurements of the thermal and electrical conductivity of high-pressure and high-temperature iron suggest compositional anomalies are the primary driver of modern core convection. It remains unexplored exactly how thermally vs. compositionally driven core flow may differ in terms of dynamo physics and therefore change currently acknowledged models. Here we compare results from numerical and laboratory experiments of rotating thermal convection in liquid gallium (a low viscosity metal with thermal properties comparable to that of Earth’s core) to experiments in high viscosity silicone oil (a fluid with thermal properties comparable to compositional properties of Earth’s core). Using silicone oil thermal convection as a proxy for core compositional convection allows for a direct comparison of models of core flow, with which we can further shed light on exactly how Earth’s dynamo is maintained. /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Boontigan Kuhasubpasin: The effect of lithospheric structure on the lithospheric stress field The state of stress in the lithosphere controls many geological processes from the local scale to plate tectonics. The sources of lithospheric stress range from mantle flow to lithospheric heterogeneity. In this study, I focus on understanding the sensitivity of the stress field to lithospheric heterogeneity by examining different models for lithospheric structure and assumptions regarding compensation and lithospheric mantle density. I use the crustal and mantle structure from Crust 1.0 and a thermodynamically determined lithosphere to calculate the gravitational potential energy (GPE) and mean outward tractions. The gradient of deviatoric stresses that balance it is solved using the finite element package ABAQUS. I compare our results for azimuth of the most compressive stress and inferred regimes to the observations from the World Stress Map 2016, and previous work using CRUST 2.0. Our study confirms how important the uncertainty in lithospheric structure weighs on our understanding of the state of stress of the lithospheric plates.
The Missing Oceanic Supershear Earthquakes Supershear earthquakes are of broad interest because they produce destructive ground shaking and their occurrence and sustained propagation give rise to important questions in rupture dynamics. Previously, only eight shallow supershear earthquakes have been reported with strong near- and/or far-field seismic evidence. Here we report seven previously undocumented oceanic supershear earthquakes based on rupture speeds determined by Slowness-Enhanced Back-Projection and far-field Rayleigh Mach wave identification. Our findings indicate that the oceanic supershear earthquakes are no less than their continental counterparts and increase the percentage of supershear earthquakes to 14.5% among all large shallow strike-slip events. We observe a wider range of stable supershear rupture speeds than theoretical predictions, possibly explained by the presence of fault damage zones and slip obliqueness. We attribute the commonness of supershear ruptures in nature to larger seismogenic width, bimaterial effect of oceanic-continental boundaries, and dynamic weakening mechanisms that promote both supershear transitions and propagations.
Xiyuan Bao: Are hotspots hotter than ridges? The temperature and composition of the mantle domains sourcing hotspots and mid-ocean ridges are fundamental to understanding their origin. Are hotspots sourced from compositionally distinct reservoirs than ridges through active, therefore hotter upwellings deep in the mantle? Geochemical signals seem to indicate so. For example, primordial 3He/4He signals as high as 50 times the present atmospheric ratio (Ra) can be found in ocean island basalts (OIBs, erupted at hotspots) , and these signals are correlated with higher buoyancy flux (plume strength) and lower seismic shear-wave velocity in both the upper mantle and lowermost mantle. In contrast, mid-ocean ridge basalts (MORBs) have lower 3He/4He values around 8 Ra, which is consistent with a different source reservoir shallower than that of OIBs. Olivine geothermometry suggests 100-300 °C of excess potential temperature at hotspots compared to ridges. Here we re-examine the temperature of oceanic hotspots and ridges from seismic tomographic models, using a self-consistent velocity to temperature conversion. Assuming the upper mantle and plume source are compositionally homogeneous and consist of depleted MORB mantle (DMM), we find that up to 60% oceanic hotspots are not resolvably hotter than ridges regardless of their proximity to each other. However, hotspots with high 3He/4He and high buoyancy flux are significantly hotter than ridges. These conclusions are robust even if there are compositional differences between plume sources and DMM. /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Leslie Insixiengmay: Earth’s Early Magnetic Field Powered by Exsolution of Silicates from Liquid Iron The Earth’s magnetic field is generated by the geodynamo: the process in which the rotation and convection of liquid iron in the outer core generates a magnetic field. Convection in the liquid outer core is driven by a combination of release of latent heat and gravitational energy from the inner core. Paleomagnetic observations show that the Earth’s magnetic field dates back at least 3.45 billion years. However, thermal evolution models suggest that the Earth’s inner core began to crystallize only one billion years ago. While we have a convincing explanation for what has powered the magnetosphere for the last billion years, it is not clear what powered it prior to the solidification of the inner core. Recent studies suggest that the reaction of mantle and core material from the Moon-forming giant impact could play a key role in understanding the formation of the Earth’s early magnetic field. Here, we focus on understanding silicate exsolution from liquid iron by running first-principles molecular dynamics (MD) simulations. The calculations implement density functional theory (DFT) through the Vienna ab initio Simulation Package (VASP), allowing us to observe how the system behaves at an atomic scale. We compute chemical compositions, reaction rates, reaction mechanisms, solubility, and the energy released by exsolution of silicate components from liquid iron.
The Earth's molten outer core and its magnetic dynamo are prevalent geophysical characteristics that exist among many planets in the solar system and other planetary systems. Turbulent flows in the Earth's molten outer core, driven by convection, generate a planetary-scale, nearly axial, and dipole-dominated magnetic field. The behaviors of strongly turbulent convection in the presence of strong Lorentz forces are mostly unknown. Thus, we present results of laboratory experiments on heat transfer of non-rotating Rayleigh-Bénard convection of liquid gallium in the presence of a vertical magnetic field. The experiment is carried out in two cylindrical containers with diameter-to-height aspect ratio Gamma = 1 and 2 for 10^ 6 <~ Ra <~ 10^8 and 0 <~ Ch <~ 10^5. Combined with the results from the previous studies, our experiment shows a more complete picture of near-onset to supercritical behaviors of heat transfer in liquid metal magnetoconvection (MC) over a large range of parameter space (10^ 3 <~ Ra <~ 10^9).