Search

Summary The effects of reaction kinetics of bridgmanite and ferropericlase transforming to ringwoodite on elastic properties in upwelling mantle are investigated using data of kinetic experiments and internally self-consistent thermodynamic modelling of density and seismic velocities. The kinetic experiments show inhibited grain growth of ringwoodite. At the initiation of ringwoodite growth, bridgmanite completely transforms to a metastable pyrope-bearing garnet. Ringwoodite then gradually grows from the metastable assemblage of ferropericlase and garnet. The changes in mineralogy result in a low-velocity zone directly above the 660 km seismic discontinuity due to the lower seismic velocities and densities of ferropericlase and garnet compared to ringwoodite and bridgmanite. The modelling of the effects of reaction kinetics and its effect on seismic structure at ∼660 km depth shows more sensitivity to grain size than to temperature and upwelling rate. Modelling 1-D synthetic seismograms of PP (SS) underside reflections off the kinetically inhibited backward reaction to ringwoodite shows advanced traveltimes of underside reflections off ∼660 km depth of 0.2–0.8 s (1.2–1.6 s) for upwelling rates of 50 cm yr–1 and initial grain sizes between 5 and 20 cm due to the low-velocity zone above the 660 km discontinuity. The finite width of the low-velocity layer results in frequency-dependent behaviour of PP and SS underside reflection amplitudes, with higher amplitudes towards shorter periods. The effect on the traveltimes of P-to-s conversions used for receiver function is small (

Seismic damping of near-surface deposits is an important input to site-response analysis for seismic hazard assessment. In Groningen, the Netherlands, gas production from a reservoir at 3 km depth causes seismicity. Above the gas field, an 800 m thick layer of unconsolidated sediments exist, which consists of a mixture of sand, gravel, clay and peat strata. Shear waves induced at 3 km depth experience most of their anelastic attenuation in these loose sediments. A good estimate of damping is therefore crucial for modelling realistic ground-motion levels. In Groningen, we take advantage of a large network of 200 m deep vertical arrays to estimate damping from recordings of the induced events. As a first step, we apply seismic interferometry by deconvolution to estimate local transfer functions over these vertical arrays. Subsequently, two different methods are employed. The first is the 'upgoing' method, where the amplitude decay of the retrieved upgoing wave is used. The second is the 'up-down' method, where the amplitude difference between retrieved up- and downgoing waves is utilized. For the upgoing method, the amplitude of the upgoing direct wave is affected by both elastic and anelastic effects. In order to estimate the anelastic attenuation, it is necessary to remove the elastic amplification first. Despite the fact that elastic compensation could be determined quite accurately, non-physical damping values were estimated for a number of boreholes. Likely, the underlying cause was small differences in effective response functions of geophones at different depths. It was found that the up-down method is more robust. With this method, elastic propagation corrections are not needed. In addition, small differences in in situ geophone response are irrelevant because the up- and downgoing waves retrieved at the same geophone are used. For the 1-D case, we showed that for estimating the local transfer function, the complex reverberations need to be included in the interferometric process. Only when this is done, the transfer function does not contain elastic transmission loss and Q estimation can be made without knowing the soil profile in detail. Uncertainty in the estimated damping was found from the signal-to-noise ratio of the estimated transfer function. The Q profiles estimated with the up-down method were used to derive a damping model for the top 200 m of the entire Groningen field. A scaling relation was derived by comparing estimated Q profiles with low-strain damping profiles that were constructed using published models for low-strain damping linked to soil properties. This scaling relation, together with the soil-property-based damping model, allowed up-scaling of the model to each grid-cell in the Groningen field. For depths below 200 m, damping was derived from the attenuation of the microseism over Groningen. The mean damping model, over a frequency band between 2 and 20 Hz, was estimated to be 2.0 per cent (0-50 m depth), 1.3 per cent (50-100 m), 0.66 per cent (100-150 m), 0.57 per cent (150-200 m) and 0.5 per cent (200-580 m). TURNkey (Towards more Earthquake-resilient Urban Societies through a Multi-sensor-based Information System enabling Earthquake Forecasting, Early Warning and Rapid Response actions); EU [H2020 - SC5 - 2018 - 2] Published version The work of ER has partly been funded by TURNkey (Towards more Earthquake-resilient Urban Societies through a Multi-sensor-based Information System enabling Earthquake Forecasting, Early Warning and Rapid Response actions), which is a project funded by the EU within the H2020 - SC5 - 2018 - 2 call. We would like to thank Jun Matsushima and an anonymous reviewer for a detailed and critical assessment.

Earthquake site response is an essential part of seismic hazard assessment, especially in densely populated areas. The shallow geology of the Netherlands consists of a very heterogeneous soft sediment cover, which has a strong effect on the amplitude of ground shaking. Even though the Netherlands is a low- to moderate-seismicity area, the seismic risk cannot be neglected, in particular, because shallow induced earthquakes occur. The aim of this study is to establish a nationwide site-response zonation by combining 3D lithostratigraphic models and earthquake and ambient vibration recordings. As a first step, we constrain the parameters (velocity contrast and shear-wave velocity) that are indicative of ground motion amplification in the Groningen area. For this, we compare ambient vibration and earthquake recordings using the horizontal-to-vertical spectral ratio (HVSR) method, borehole empirical transfer functions (ETFs), and amplification factors (AFs). This enables us to define an empirical relationship between the amplification measured from earthquakes by using the ETF and AF and the amplification estimated from ambient vibrations by using the HVSR. With this, we show that the HVSR can be used as a first proxy for site response. Subsequently, HVSR curves throughout the Netherlands are estimated. The HVSR amplitude characteristics largely coincide with the in situ lithostratigraphic sequences and the presence of a strong velocity contrast in the near surface. Next, sediment profiles representing the Dutch shallow subsurface are categorised into five classes, where each class represents a level of expected amplification. The mean amplification for each class, and its variability, is quantified using 66 sites with measured earthquake amplification (ETF and AF) and 115 sites with HVSR curves. The site-response (amplification) zonation map for the Netherlands is designed by transforming geological 3D grid cell models into the five classes, and an AF is assigned to most of the classes. This site-response assessment, presented on a nationwide scale, is important for a first identification of regions with increased seismic hazard potential, for example at locations with mining or geothermal energy activities.

The physical and chemical properties of the Earth’s mantle govern the cause of natural disasters, such as earthquakes and volcanoes. Since we do not have direct access to mantle materials, their properties are often inferred from laboratory measurements and surface observations (e.g. seismic data from earthquake recordings). This thesis addresses some key problems we face while utilising these data to constrain the thermal and chemical properties of the mantle. Firstly, we propose a data-driven approach based on machine learning to explain the laboratory measurements and quantify their uncertainties in the absence of an adequate physical model. Our results show that although conventional approaches based on fitting the measurements to an assumed model may appear better constrained, they could potentially provide biased results. Secondly, we use the data-driven approach to explore which thermochemical parameters can be constrained (and to what extent) with limited seismic observables- wave speeds and density. Our results show that these observables constrain temperature and major chemical parameters (silicon, magnesium, and iron), and they indicate the presence of thermochemical heterogeneities at the lowermost mantle. The dense and slow piles at the bottom of the lower mantle seen in seismic data can be explained by an enrichment in silica and iron content- characteristic feature of enstatite chondrites. The inferred heterogeneities have profound implications for the dynamics of the mantle and outer core. The methodology developed in this thesis is extremely efficient. It can easily incorporate additional observables and thus, has wide applications in the seismology and mineral physics community.

Shear properties of mantle minerals are vital for interpreting seismic shear wave speeds and therefore inferring the composition and dynamics of a planetary interior. Shear wave speed and elastic tensor components, from which the shear modulus can be computed, are usually measured in the laboratory mimicking the Earth's (or a planet's) internal pressure and temperature conditions. A functional form that relates the shear modulus to pressure (and temperature) is fitted to the measurements and used to interpolate within and extrapolate beyond the range covered by the data. Assuming a functional form provides prior information, and the constraints on the predicted shear modulus and its uncertainties might depend largely on the assumed prior rather than the data. In the present study, we propose a data-driven approach in which we train a neural network to learn the relationship between the pressure, temperature and shear modulus from the experimental data without prescribing a functional form a priori. We present an application to MgO, but the same approach works for any other mineral if there are sufficient data to train a neural network. At low pressures, the shear modulus of MgO is well-constrained by the data. However, our results show that different experimental results are inconsistent even at room temperature, seen as multiple peaks and diverging trends in probability density functions predicted by the network. Furthermore, although an explicit finite-strain equation mostly agrees with the likelihood predicted by the neural network, there are regions where it diverges from the range given by the networks. In those regions, it is the prior assumption of the form of the equation that provides constraints on the shear modulus regardless of how the Earth behaves (or data behave). In situations where realistic uncertainties are not reported, one can become overconfident when interpreting seismic models based on those defined equations of state. In contrast, the trained neural network provides a reasonable approximation to experimental data and quantifies the uncertainty from experimental errors, interpolation uncertainty, data sparsity and inconsistencies from different experiments.

With the advent of plate tectonics in the last century, our understanding of the geological evolution of the Earth system improved essentially. The internal deformation and evolution of tectonic plates remain however poorly understood. This holds in particular for the Central Mediterranean: The formerly much larger Adriatic plate is recently consumed in tectonically active belts spanning at its western margin from Sicily, over the Apennines to the Alps and at its eastern margin from the Hellenides, Dinarides towards the Alps. High seismicity along these belts indicates ongoing lithospheric deformation. It has been shown that data acquired by dense, regional networks like AlpArray provide crucial information on seismically active faults as well as on the structure and deformation of the lithosphere. The Adriatic Plate and in particular its eastern margin have however not been covered by a homogeneous seismic network yet.Here we report on the status of AdriaArray – a seismic experiment to cover the Adriatic Plate and its actively deforming margins by a dense broad-band seismic network. Within the AdriaArray region, currently about 990 permanent broad-band stations are operated by more than 40 institutions. Data of 97% of these stations are currently available via EIDA. In addition to the existing stations, 414 temporary stations from 24 mobile pools are deployed in the region achieving a coverage with an average station distance of 50 – 55 km. The experiment is based on intense cooperation between local network operators, mobile pool operators, field teams, ORFEUS, and interested research groups. Altogether, more than 50 institutions are participating in the AdriaArray experiment. We will report on the time schedule, participating institutions, mobile station pools, maps of temporary station distribution with station coverage and main points of the agreed Memorandum of Collaboration. The AdriaArray experiment will lead to a significant improvement of our understanding of the geodynamic causes of plate deformation and associated geohazards.

The first study in this thesis is designed to cover both approaches of Marchenko redatuming and Marchenko-based primary estimation in a range of subsampling scenarios, providing insights that can inform acquisition sampling choices as well as processing parameterization and quality control, e.g., to set up appropriate data filters and scaling to accommodate the effects of dipole fields, or to help ensuring that the data interpolation achieves the desired levels of reconstruction quality that minimize subsampling artifacts in Marchenko-derived fields and images. With the aim of handling short-period multiples in laterally-varying media with complex, realistic layering, a updated augmented Marchenko approach is proposed in this thesis providing reliable approximations to the problem. This is achieved through combining an adapted 1.5D approach and additional post-processing with the recently proposed Marchenko-type demultiple formula. The new approach and its limitations are demonstrated with a range of numerical models that capture varying degrees of lateral heterogeneity in the overburden. The benchmark of migrated images after removing short-period multiples suggests promising applications to field data. As a next step in the reflection-data processing chain, the migrated images are nonetheless blurred with uneven amplitudes and low resolution, despite a successful multiples removal procedure. As an important post-processing step, seismic image deblurring aims to enhance both the resolution and amplitude balance. Conventional methods developed for such an inverse problem are usually cumbersome and slowly converging, and require case-dependent user interference, e.g. in the form preconditioning, and the fine-tuning of free parameters for regularization. This thesis proposes instead to address the problem with a physics-based Machine Learning technique --- the (invertible) Recurrent Inference Machine, by explicitly employing the point spread function as the forward operator and thus prior information. With a simple almost-flat synthetic training model and very cheap computational costs, this new approach outperforms the benchmark of the invertible UNet - a commonly-used convolutional neural network architecture - in a series of tests, designed with both complex synthetic models and the field data application. Different from the other methods, the neural networks in this study are trained to map the migrated image into an impedance perturbation model, rather than the reflectivity model, making it closer to the next-level application of the impedance inversion for reservoir characterisation and time-lapse monitoring. In the last part of the thesis, a general and unified framework is proposed, which flexibly incorporates all kinds of interface conditions for any type of wave equations. With the new approach, the implementation of a wave equation solver can be benchmarked by comparing different angles of plane wave simulations to transmission/reflection coefficients from plane-wave analysis with exact boundary conditions. This is illustrated with the case of poroelastic wave equation, which has a wide application in rock physics and reservoir monitoring.

Measurements of pore pressure variations contain important information for the production of hydrocarbon reservoirs. Here, we couple surface-wave velocity changes directly to pore pressure change using sensitivity kernels. Using these kernels for realistic pore pressure changes in the Harlingen chalk gas reservoir, we modelled yearly Rayleigh-wave velocity changes. The velocity variations appear sufficiently large to be detectable on a yearly basis using passive image interferometry. This implies that, for specific formations, field-wide pore pressure monitoring is feasible with continuous seismic noise recordings at the surface.

In this study, we show that passively recorded data of nearby passing trains by a deep borehole geophone array could be linked to fluctuations of the gas-water contact in the Groningen reservoir in The Netherlands. During a period of 1.5 months, changes of inter-geophone P-wave travel times were detected by deconvolution interferometry of the recorded train signals. P-to-S converted waves, obtained by deconvolution of the horizontal component by the vertical component at individual geophones, showed simultaneous variations. The observed travel-time changes could be related to fluctuations of the gas-water contact in the observation well caused by pressure variations at a well drilling 4.5 km away. The $$\sim$$ ∼ 3.5 day delay between drilling in the reservoir and the seismic response yields a hydraulic diffusivity of approximately 5 m$$^2$$ 2 /s and suggests that the pressure front is effectively propagated over such a long distance. Our observations illustrate that downhole geophone arrays can be used to monitor changes in the subsurface if repeating noise sources are available, and that unexpected effects may occur due to drilling.

SUMMARY Receiver functions have been used for decades to study the Earth’s major discontinuities by focusing on converted waves. Deconvolution, which is the mathematical backbone of the method, is assumed to remove the source time function and the far-field dependence on structure, making it a useful method to map the nearby Earth structure and its discontinuities. Ray theory, a plane incoming wavefield, and a sufficiently well-known near-receiver background velocity model are conventionally assumed to map the observations to locations in the subsurface. Many researchers are aware of the shortcoming of these assumptions and several remedies have been proposed for mitigating their consequences. Adjoint tomography with a quasi-exact forward operator is now within reach for most researchers, and we believe is the way forward in receiver function studies. A first step is to calculate adjoint sensitivity kernels for a given misfit function. Here, we derive the adjoint source for a receiver function waveform misfit. Using a spectral element forward code, we have calculated sensitivity kernels for P-to-S converted waves using several 2-D models representing an average crust with an underlying mantle. The kernels show profound differences between P- and S-wave speed sensitivity. The sensitivity to P-wave speed is wide-ranging and related to the scattered P-wavefield which interferes with that of the P-to-S converted wave. The S-wave speed sensitivity is more local and mostly associated to potential locations of P-to-S conversion, although more distant sensitivity is also observed. Notably, there is virtually no sensitivity to impedance. We further observe the well-known trade-off between depth of the discontinuity and wave speed, but find that considering a longer waveform that includes more surface reverberations reduces this trade-off significantly.

Since the discovery of the inner core almost 100 years ago, the seismological community has found that the inner core contains significant heterogeneity in its elastic structure. This observation is significant and in many ways unexpected; we believe the inner core to be (relatively) chemically homogeneous consisting primarily of iron and nickel. Yet we observe that seismic waves which pass through the inner core travel faster in a north-south direction than an east-west direction and that the spectra of whole Earth oscillations are anomalously split in a way which is consistent with the same velocity difference. This difference in velocity between two directions through the inner core is called anisotropy, and from mineral physics we have reason to believe that this anisotropy is caused by the alignment of iron crystals which are themselves anisotropic at inner core temperatures and pressures. The primary goal of this thesis is to constrain, as well as possible, the elastic structure of the inner core. We expand upon the body wave dataset by adding new observations of paths which travel almost parallel to Earth's axis of rotation, giving us improved sensitivity to velocity in the north-south direction in the inner core. We combine our new data with other body wave datasets to produce a 3D seismic tomographic model of the inner core. This model utilised a transdimensional Markov chain Monte Carlo methodology which not only determines the best fitting anisotropy structure in the inner core, but also the uncertainties in our model and it does not require any prior assumptions on the parameterization of the inner core. The advantage of this method is significant, especially because the relatively poor sampling of the inner core means that prior assumptions on the parameterization may significantly affect the final model. In the transdimensional approach the parameterization is a part of the inversion. In our new transdimensional model we confirmed many previous observations, including an isotropic layer of 100 km thickness at the top of the inner core and that the inner core is split broadly into a western region and an eastern region. We are now able to make new robust observations, seeing for the first time that the western anisotropic zone is isolated to the northern hemisphere and that the inner most inner core exists but primarily in the eastern region. These observations are significant as it provides new insight into the mechanisms of inner core formation and dynamics, and we discuss the potential implications for inner core geodynamics. It is important in deep Earth research to bring together as many sources of information as possible. We have also measured 18 normal modes sensitive to the inner core. We used a splitting function approximation and a grid search methodology to constrain the uncertainties in the measurement. The data were then used to produce a preliminary 1D transdimensional model of inner core anisotropy using polynomial basis functions and find a model which agrees reasonably well with the spherical average of compressional anisotropy from the body wave model.

We reanalysed the induced seismicity data from the Groningen gas reservoir. We used the well-maintained induced event catalogue of the KNMI. The distributions of seismic moments and interevent times show a power law behaviour over several decades, and we find that upon increasing the magnitude threshold, these distributions remained scale-invariant. Because of this scale-invariance, we can put a constraint on the average loading of the elastic energy within the reservoir, which upon reaching a critical value gives rise to the seismic events. We find that the elastic energy roughly increases proportional to time. We also propose a new machine learning approach for declustering the seismic events, separating correlated from independent events. We find that only few events are truly independent, i.e. exponentially distributed over time. There are also few aftershocks following an Omori-type power law. The bulk of events presents a Gamma distribution for interevent times. This gives us an indication that the rigidity in the reservoir is high, but whether this results in overall correlated events should be settled with physics-based arguments rather than statistical ones.

Botswana is a country with relatively low seismic activity that experienced an unexpected Mw 6.5 earthquake on 3 April 2017. Using data from the first countrywide seismic network, we established a Botswana earthquake catalog for the period January 2014 to February 2018. Two areas of elevated seismic activity were detected. The first one is the Okavango Rift Zone in northern Botswana, an area that is known to be active. The other one is associated with the 2017 mainshock and its aftershocks in central Botswana; it follows the Paleoproterozoic suture between the Limpopo Belt and the Kaapvaal Craton. Double-difference relocation of these events revealed a reactivated fault system with a northwesterly strike with the aftershocks occurring at shallower depths than the mainshock at 29 km. The focal mechanisms of the mainshock and selected aftershocks are of normal faulting type with strikes similar to the orientation of the fault system. The unidirectional rupture of the mainshock in the lower crust combined with the upward migration of the aftershocks along the Moiyabana Fault and a thin low velocity anomaly the uppermost mantle are consistent with the events being produced by eclogitization of a dry metastable granulite facies rock by fluid intrusion from the mantle in an extensional stress regime. The Okavango Rift Zone is generally interpreted as the terminus of the southwestern branch of the East African Rift System. We suggest that the recent earthquakes in central Botswana may be considered as the southern continuation of this branch.

Similar to horizontal earthquake motions, vertical motions are amplified depedent on the local site conditions which can be critical for the safety of certain structures. Production of natural gas in Groningen, the Netherlands, results in reservoir compaction causing low magnitude, shallow earthquakes which are recorded with a borehole seismic network. These recordings form an excellent data set to understand how shallow unconsolidated subsurface geology influences the amplification behaviour of compressional waves (P-waves). First, we present borehole and single-station techniques (amplification factors, empirical transfer functions (ETF) and V/H spectral ratio implementations) to quantify vertical amplification. We show that vertical-wave incidence is a reasonable assumption. All techniques are capable of emphasising the sites with strong amplification of vertical ground motion during an earthquake. Subsequently, we compare ETF with single-station methods with the aim to develop proxies for vertical site-response using spectral ratios. In a second step, we link vertical site-response with shallow subsurface conditions, like the P-wave velocity and peat content. To better understand the amplification mechanisms, we analytically simulate P-wave propagation. In the simulations, we compute synthetic transfer functions using realistic subsurface conditions and make a comparison with the ETF. The simulations support the hypothesis that thin layers of shallow gas, originating from the Holocene peat, result in wave amplification. We observe strong vertical site-response in particular in the eastern part of Groningen, with industrial facilities and pipeline infrastructure in the region. Here, if high vertical amplifications are persistent at large earthquake magnitudes, appreciable levels of vertical loading may be expected. This study demonstrates that vertical motions should be assessed separately from horizontal motions, given that the amplification behaviour of P-waves is affected by distinctive mechanisms.

In this paper, we consider a simplified model of turbulence for large Reynolds numbers driven by a constant power energy input on large scales. In the statistical stationary regime, the behaviour of the kinetic energy is characterised by two well defined phases: a laminar phase where the kinetic energy grows linearly for a (random) time $t_w$ followed by abrupt avalanche-like energy drops of sizes $S$ due to strong intermittent fluctuations of energy dissipation. We study the probability distribution $P[t_w]$ and $P[S]$ which both exhibit a quite well defined scaling behaviour. Although $t_w$ and $S$ are not statistically correlated, we suggest and numerically checked that their scaling properties are related based on a simple, but non trivial, scaling argument. We propose that the same approach can be used for other systems showing avalanche-like behaviour such as amorphous solids and seismic events., Comment: 12 pages, 10 figures

We study the inverse scattering problem for a Schrödinger operator related to a static wave operator with variable velocity, using the GLM (Gelfand–Levitan–Marchenko) integral equation. We assume to have noisy scattering data, and we derive a stability estimate for the error of the solution of the GLM integral equation by showing the invertibility of the GLM operator between suitable function spaces. To regularise the problem, we formulate a variational total least squares problem, and we show that, under certain regularity assumptions, the optimisation problem admits minimisers. Finally, we compute numerically the regularised solution of the GLM equation using the total least squares method in a discrete sense.

Detailed imaging of accretionary wedges reveals splay fault networks that could pose a significant tsunami hazard. However, the dynamics of multiple splay fault activation during megathrust earthquakes and the consequent effects on tsunami generation are not well understood. We use a 2‐D dynamic rupture model with complex topo‐bathymetry and six curved splay fault geometries constrained from realistic tectonic loading modeled by a geodynamic seismic cycle model with consistent initial stress and strength conditions. We find that all splay faults rupture coseismically. While the largest splay fault slips due to a complex rupture branching process from the megathrust, all other splay faults are activated either top down or bottom up by dynamic stress transfer induced by trapped seismic waves. We ascribe these differences to local non‐optimal fault orientations and variable along‐dip strength excess. Generally, rupture on splay faults is facilitated by their favorable stress orientations and low strength excess as a result of high pore‐fluid pressures. The ensuing tsunami modeled with non‐linear 1‐D shallow water equations consists of one high‐amplitude crest related to rupture on the longest splay fault and a second broader wave packet resulting from slip on the other faults. This results in two episodes of flooding and a larger run‐up distance than the single long‐wavelength (300 km) tsunami sourced by the megathrust‐only rupture. Since splay fault activation is determined by both variable stress and strength conditions and dynamic activation, considering both tectonic and earthquake processes is relevant for understanding tsunamigenesis., Plain Language Summary: In subduction zones, where one tectonic plate moves beneath another, earthquakes can occur on many different faults. Splay faults are relatively steep faults that branch off the largest fault (the megathrust) in a subduction zone. As they are steeper than the megathrust, the same amount of movement on them could result in more vertical displacement of the seafloor. Therefore, splay faults are thought to play an important role in the generation of tsunamis. Here, we use computer simulations to study if an earthquake can break multiple splay faults at once and how this affects the resulting tsunami. We find that multiple splay faults can indeed fail during a single earthquake due to the stress changes from trapped seismic waves, which promote rupture on splay faults. Rupture on splay faults results in larger seafloor displacements with smaller wavelengths, so the ensuing tsunami is bigger and results in two main flooding episodes at the coast. Our results show that it is important to consider rupture on splay faults when assessing tsunami hazard., Key Points: Multiple splay faults can be activated during a single earthquake by megathrust slip and dynamic stress transfer due to trapped waves. Splay fault activation is facilitated by their favorable orientation with respect to the local stress field and their closeness to failure. Long‐term geodynamic stresses and fault geometries affect dynamic splay fault rupture and the subsequent tsunami., Volkswagen Foundation (VolkswagenStiftung) http://dx.doi.org/10.13039/501100001663, Royal Society (The Royal Society) http://dx.doi.org/10.13039/501100000288, EC | H2020 | H2020 Priority Excellent Science | H2020 European Research Council (ERC) http://dx.doi.org/10.13039/100010663, Deutsche Forschungsgemeinschaft (DFG) http://dx.doi.org/10.13039/501100001659, National Science Foundation (NSF) http://dx.doi.org/10.13039/100000001, https://github.com/TUM-I5/SWE, https://doi.org/10.5281/zenodo.6969455

The thermochemical nature of the lower mantle, and the two Large Low Shear-wave Velocity Provinces (LLSVPs) in particular, remains a topic of active debate. Exclusively imaging seismic velocity anomalies only provides limited ability to distinguish between thermal and compositional origins. Here we use whole Earth oscillations, or normal modes, to study 3D variations in mantle shear wave- (vs), compressional wave- (vp) and bulk sound velocity (vc), density (rho) and shear attenuation (qmu). These observations provide new constraints on the presence of lower mantle chemical heterogeneity. Here, we compare the two-step splitting function inversion method to the less frequently used, computationally more expensive one-step direct spectrum inversion. In theory, the one-step inversion suffers less from non-uniqueness and only requires regularization once. In practice, we find that the average spectral misfits for the one-step inversion are always lower. The ratio between vs and vp anomalies obtained from their joint inversion, proposed to be an indicator of chemical heterogeneity when exceeding a threshold predicted by mineral physics, varies significantly between the two inversion methods. The method of computing the ratio is just as important. We obtain ratios exceeding the threshold in the lower mantle only when dividing the root mean square amplitudes of our vs and vp models, although lower than some previous studies suggest. However, by taking the median ratio from a grid at each depth, we barely exceed the threshold. Instead of relying on these 1D representatives of the ratio, we infer chemical heterogeneity in certain depth ranges based on a wide spread in distributions of vs, vp anomalies and their ratio. Another constraint on the presence of chemical heterogeneity comes from the anti-correlation of vs and vc structure, found in many previous studies. We find (de-)correlation in the lower mantle in joint vs and vc inversions, with only slightly negative correlation for the two-step inversion. We shift towards more negative correlation values when extracting vc from our previously obtained vs and vp models. Lower mantle density structure has remained elusive and controversial in recent decades. In joint inversions for vs, vp, rho and discontinuity topography, we show that a basal layer of excess density is located underneath part of the otherwise lighter-than-average LLSVPs, reconciling previous observations. This dense layer is more robust in the one-step inversion and does not result from high-amplitude ghost patterns that may have plagued earlier normal mode studies. Compositional variations such as iron enrichment may explain this dense but seismically slow layer. Finally, we obtain preliminary results for imaging lower mantle shear attenuation in joint vs and qmu one-step inversions. We first select the best way of inverting for 3D qmu in synthetic tests, before applying this method to real data. Mode selection appears to be crucial, and we also need to properly account for detailed elastic structure to minimize leakage into anelastic structure. The preliminary 3D qmu model shows low attenuation in LLSVPs and high attenuation in the surrounding lower mantle, possibly indicating a dominant role for grain size.

Body wave observations of the Earth's inner core show that it contains strong seismic heterogeneity, both laterally and radially. Models of inner core structure generated using body wave data are often limited by their parameterisation. Thus, it is difficult to determine whether features such as anisotropic hemispheres or an innermost inner core truly exist with their simple shapes, or result only from the chosen parameterisation and are in fact more complex features. To overcome this limitation, we conduct seismic tomography using transdimensional Markov Chain Monte Carlo on a high quality dataset of 5296 differential and 2344 absolute P-wave travel times. In a transdimensional approach, the data defines the model space parameterisation, providing us with both the mean value of each model parameter and its probability distribution, allowing us to identify well versus poorly constrained regions. We robustly recover many first order observations found in previous studies without the imposition of a priori fixed geometry including an isotropic top layer (with anisotropy less than 1%) which is between 60 and 170 km thick, and separated into hemispheres with a slow west and a faster east. Strong anisotropy (with a maximum of 7.2%) is found mainly in the west, with much weaker anisotropy in the east. We observe for the first time that the western anisotropic zone is largely confined to the northern hemisphere, a property which would not be recognised in models assuming a simple hemispherical parameterisation. We further find that the inner most inner core, in which the slowest anisotropic velocity direction is tilted relative to Earth's axis of rotation (ζ=55∘±16∘), is offset by 400 km from the centre of the inner core and is restricted to the eastern hemisphere. We propose that this anomalous anisotropy might indicate the presence of a different phase of iron (either bcc or fcc) compared to the rest of the inner core (hcp). Published version This research was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 681535 - ATUNE) and a Vici award number 016.160.310/526 from the Netherlands Organisation for Scientific Research (NWO). LW acknowledges support from a Discovery Early Career Research Award (project number DE170100329), funded by the Australian Government.