Location: ES 2093 (22 Russel Street) |
Time: Tuesday afternoons (4-5 pm during school year, 4:30-5:30 for KEGS meeting, 3-4 pm during summer, unless otherwise noted)
Schedule (email liuqy@physics if you have any visitor who would like to give a talk):
In spring: 1/7 (Yuan Xu), 1/21; 2/4, 2/25 (Johnson Ha); 3/5 (inverse short course by Michael Zhdanov), 3/6 (special Lecture by Michael Zhdanov), 3/17 (Yajing Liu), 3/24, 3/31
Dates reserved for KEGS talks:
Physics Colloquium (Zoback): 01/30
Other important Dates: 09/05: fall classes start; 11/4-8: fall reading week; 12/04: last day of fall classes; 01/06: spring classes start; 02/17-21: spring reading week; 04/03: last day of classes.
Other relevant seminar series at UofT: ES seminar(Thu noon), ES Rockfest (Fri 3pm, run by grads), Physics Colloquium (Thu 4pm), Brewer-Wilson Seminar (Fri noon, run by grads), Center for Global Change Science (CGCS) Seminars (Tue 4 pm)
For a list of Geoscience events across GTA, check out the GTA Geoscience Events Calendar. Also check out the websites for Toronto Geological Discussion Group (TGDG), Canadian Exploration Geophysical Society (KEGS) and KEGS Foundation.
ES seminar: Agricultural Geophysics: New Opportunities in applications of Geophysics
Date/Time: Jan 23rd 2020, NOON
Speaker: Prof. Erasmus Oware
Affiliations: University of Baffalo
Abstract: See ES seminar website
Physics Colloquium: Utilizing Physics-Based Models to Manage the Risk of Injection-Induced Seismicity Associated with Unconventional Oil and Gas Production
Date/Time: Jan 30th 2020, 4-5 pm
Speaker: Prof. Mark Zoback
Affiliations: Stanford University
Abstract: The occurrence of earthquakes induced by wastewater injection and hydraulic fracturing is well documented in areas of unconventional oil and gas development around the world. In this talk I will describe how and why these earthquakes occur and outline the steps that can be taken to minimize the occurrence of such earthquakes. One key step is identification of potentially problematic faults prior to fluid injection (or hydraulic fracturing) another is related to understanding how pressure diffusion and/or poroelastic stress transfer affects the likelihood of triggering fault slip.
KEGS Workshop on "Principles & Pitfalls of Geophysical Inversion"
Date/Time: March 5, 2020, 8:30-12
Speaker: Prof. Michael Zhdanov
Affiliations: University of Utah and TechnoImaging (2020 Collett Distinguished Lecturer)
Abstract: See KEGS website for details on how to sign up.
Special Geophysics Seminar: Effective-medium inversion of induced polarization data for mineral exploration and mineral discrimination: principles and case histories
Date/Time: March 6, 2020, 12-1 pm
Speaker: Prof. Michael Zhdanov
Affiliations: University of Utah and TechnoImaging
Date/Time: 2020, 4-5 pm
Title: 2019-20 CSEG Distinguished Lecture: Crustal Fluids, Friction and Faults: What can we learn from injection-induced earthquakes?
Date/Time: January 14, 2020, 4:30-5:30 pm
Speaker: Prof. David Eaton
Affiliations: University of Calgary
Induced earthquakes - seismic events that are triggered by human activities - have been linked to various anthropogenic processes including deep underground mining, impoundment of a large surface water reservoir behind a dam, and subsurface injection or withdrawal of fluids. Several energy technologies, such as shale-gas development and enhanced geothermal systems, rely on subsurface fluid-injection processes that mimic certain naturally occurring phenomena. The deployment of these energy technologies has led to felt seismicity in some areas where certain necessary conditions are met, notably the presence of a pre-existing fault network and a hydraulic pathway connecting it to the injection source. Passive-seismic monitoring is a rapidly developing geophysical technique used to characterize fracture growth, fluid diffusion and fault activation across a range of temporal and spatial scales. Recent investigations of induced seismicity are yielding surprising new insights about fluid transport, ground motion, and the frictional behaviour of faults. Examination of induced events could therefore aid in understanding natural earthquakes in intraplate regions and, more generally, fluid-driven processes in the Earth's crust.
Title: Synthetic transmit aperture ultrasound imaging (STA) in biomedical applications
Date/Time: January 7th 2020, 4-5 pm
Speaker: Prof. Yuan Xu
Affiliations: Department of Physics, Ryerson University
Ultrasound imaging has become one of the most popular medical imaging modalities, mainly because it is non-invasive, real-time, and low-cost. Traditional ultrasound imaging is implemented by scanning the object with a focused ultrasound beam. I will discuss three challenges in medical ultrasound imaging: the high system complexity of 3D ultrasound imaging, the heterogeneity of sound speed in biological tissues, and the low signal-to-noise ratio in ultrasound images due to speckles. Synthetic transmit aperture (STA) imaging has been investigated in my lab to address these challenges. First, we used sources larger than a wavelength, rather than point elements as in the traditional STA, to reduce the system complexity significantly while maintaining a reasonable imaging quality. We used spatial response function in the image reconstruction to improve the image quality. Second, normalized cross correlation was used in STA to estimate the relative arrival delay error and the aberration profile due to acoustic speed heterogeneity. An algorithm was also proposed to maximize the brightness over a region of interest of the reconstructed image to find the absolute delay error with the data of a single image frame. The potential application of the delay estimation to mapping the speed of sound in soft tissues will also be discussed. Lastly, spatial and temporal compounding techniques have been developed to improve the image contrast-to-noise ratio by reducing the background speckle variations in STA.
Title: Computational seismology in Taiwan: recent achievements and future challenges
Date/Time: Nov 26 2019, 4-5 pm
Speaker: Dr. Shiann-Jong Lee
Affiliations: Institute of Earth Sciences (IES), Academia Sinica, Taiwan
Abstract: My recent research has two main topics: (1) Numerical inversion of precise source
rupture model; (2) Forward simulation of high resolution 3-D wave propagation. By
collecting all the numerical source models and numerical wave propagation
simulation results, a numerical earthquake database of hazardous earthquakes,
including the past, recent events and future scenario earthquakes, in Taiwan will be
established. The collection of numerical earthquake models can apply not only on the
study of earthquake source physics but also on the earthquake engineering. The
long-term goal is to realize ground motion prediction as well as to give contribution
for earthquake mitigation and seismic hazard assessment.
2019 Tuzo Wilson Lecture: Exploring the Earth's Interior by Full Seismic Waveforms
Date/Time: Nov 19th 2019, 7:30 pm at Isabel Bader Theatre
Speaker: Prof. Qinya Liu
Affiliations: Department of Physics & Department of Earth Sciences, University of Toronto
KEGS Nov meeting: Integrated geophysical imaging of crystalline crust in Canada's Superior Archean province: Progress with the Metal Earth Project.
Date/Time: Nov 12th 2019, 4:30 pm
Speaker: Prof. Mostafa Naghizadeh
Affiliations: Assistant Professor of Exploration Geophysics, Harquail School of Earth Sciences, Laurentian University
Abstract: The Metal Earth project acquired ~1000 km of deep seismic reflection profiles in Canada's Superior Archean province. These surveys cover from Rainy River near the Manitoba-Ontario border in the Wabigoon geological subprovince to Chibougamau in eastern Quebec in the Abitibi geological subprovince. Along the same transects, Gravity and Magnetotelluric (MT) data were also acquired for the purpose of integrated interpretation of geophysical data. Metal Earth regional-scale transects, covering up to 70 km offsets, target mineralizing fluid pathways throughout the crust, whereas higher spatial-resolution reflection surveys target structures at mine camp scales. Because Metal Earth was proposed to map and compare entire Archean ore and geologically similar non-ore systems, regional sections cover the entire crust to the Moho in the Abitibi and Wabigoon greenstone belts. The processing workflow of Metal Earth's crustal-scale seismic data was focused on robust static solutions, detailed velocity analysis, minimal trace smoothing, and high-resolution imaging. Where the new sections overlap with previous Lithoprobe surveys, a clear improvement in reflector detection and definition is observed. Improvements are here attributed to the increased bandwidth of the signal, better estimates of seismic wave speeds used in processing, and especially more accurate migrations of the data. The inverted Gravity, Magnetic, and MT models were integrated into the interpretation of seismic images, revealing possible mineralizing fluid pathways extending to the surface which could be considered as potential mineral prospecting targets. In this talk, I will present the latest integrated geophysical interpretation of the Superior province from the Metal Earth and Lithoprobe data, utilizing electrical resistivity, density, and seismic reflectivity models to identify the nature of anomalies.
Special Geophysics Seminar: The Forgotten Holocaust - Geophysical Investigations at World War 2 Extermination Sites in Lithuania
Date/Time: November 5th, 2019, NOON
Speaker: Dr. Alastair McClymont
Affiliations: Advisian, Calgary, Alberta
Abstract: Between 1941 and 1944, Einsatzgruppe Nazi death squads conspired with local nationalists to systematically exterminate as many as 250,000 Jews and other persecuted ethnic groups from the region that now forms Lithuania. Most of those killed were shot and buried in mass graves that are scattered across the country, many of which are only known to exist through eye witness testimony or by accidental exhumation of human remains. Because of Jewish cultural sensitivities, traditional archaeological excavations of the suspected grave sites are prohibited. Non-invasive geophysical methods, including electrical resistivity tomography (ERT), ground-penetrating radar (GPR), ground magnetics, terrain conductivity mapping, and unmanned aerial vehicle (UAV) methods have provided useful information to help document the past history of these sites in lieu of, and prior to, targeted archaeological excavations. In this talk I will present results from geophysical surveys conducted between 2016 and 2019 at sites including the Ponary Extermination Camp, the former HKP562 Work Camp in Vilnius, and the IXth Fort in Kaunas. This work has been featured in two recent documentary films and has helped Lithuanian researchers in their efforts to not only memorialize the victims of the Holocaust but tell the story of their lost Jewish culture.
2019 SEG Honorary Lecturer: Azimuthal P-P seismic measurements: Past, present, and future
Date/Time: Oct 23rd (Wednesday), 4-5 pm
Speaker: Heloise Lynn
Affiliations: Lynn Inc., La Veta, Colorado, USA
Abstract: See SEG Honorary lecture website
Special Geophysics Seminar: Exploring of Offshore Ground Water - A Case Study from Canterbury Bright, New Zealand
Date/Time: Oct 16th (Wednesday), NOON
Speaker: Dr. Marion Jegen-Kulcsar
Abstract: Vast offshore bodies of fresh and moderately brackish groundwater have been documented up to 100 km from modem shorelines. These aquifers could potentially relieve increasing stress on aquifers in heavily populated coastal areas. Yet little is known about their exact shape and dynamic, and thus whether they could be used sustainably. I will present a case study from the Canterbury Bight in New Zealand, In which we integrate controlled-source electromagnetic and seismic reflection data, borehole information and hydrological modelling to quantitatively characterise a previously unknown offshore freshened groundwater (OFG) system near Canterbury, New Zealand.
Title: Seismic velocity estimation: a deep recurrent neural-network approach
Date/Time: October 15, 2019, 4-5 pm
Speaker: Prof. Gabriel Fabien-Ouellet
Affiliations: Département des génies civil, géologique et des mines, Polytechnique Montréal
Abstract: Seismic reflection imaging is a powerful geophysical technique that uses reflected seismic waves to characterize the structure of the subsurface. It can be applied at all scales, and have been used to detect the Moho, find new mining targets and delineate near-surface variations for engineering studies. However, seismic surveys require extensive data processing that is labour intensive and costly.
In this talk, we ask the question: can seismic data processing and imaging be automated with machine learning? Due to the sheer complexity of this problem, the answer remains elusive. However, part of the solution can be gleaned from smaller, simpler sub-problems. The machine learning community has fully embraced this strategy--- one of the most successful examples being the famous MNIST problem wherein simple handwritten digits must be recognized by an algorithm.
In this talk, we argue that velocity analysis of seismic reflection data has the potential to become a similar canonical example for the seismic community. More precisely, we define the problem as the estimation of normal move-out (NMO) and interval velocity from common midpoint (CMP) seismic gathers in the idealized synthetic approximation of a flat layered Earth. This problem is small enough so that training can be performed with modest computational resources but is still directly applicable to a stacking approach for seismic imaging. We investigate estimating the NMO-derived velocity for this model with a deep, recursive neural net. The architecture of the network mimics traditional semblance analysis in that steps such as semblance computation and picking are replaced with different
`layers' of convolutional, recursive and recurrent neural nets. We show that the precision of the predicted velocity profiles using the network approach is favorable compared to traditional semblance analysis. We additionally show that the implemented network is applicable to processing real data despite being trained on synthetics. We finally argue that the NMO velocity estimation problem possesses all the appropriate attributes needed to serve as a testbed for new machine learning approaches to seismic processing.
KEGS October meeting: A rapid model for generating synthetic gamma ray detector responses
Date/Time: Oct 8th, 2019, 4:30 pm
Speaker: Dr. Nicolas Martin-Burtart
Affiliations: Radiation Solutions
Abstract: Providing correct nuclide identification is critical for first responders and two main methods can be used to achieve that goal: finding peaks and template matching. Finding peaks only uses a small portion of the available information in a spectrum and shielding can attenuate or eliminate some peaks of a nuclide, making it harder to identify.
Template fitting allows taking the whole spectral response into account, making up for different geometries, but generating appropriate spectra is challenging. The shielding around the source, the scattering happening between the source and the detector as well as the source geometry itself will result in different spectral shapes making it difficult to model. All simulations need to be redone when switching from one detector to another: the size of the detector and its nature will dramatically change the response. The spectrum recorded by the detector is the result of a process which can be described as the product of three functions: Output = Detector_Response x Shield x Source.
A new approach has been developed, where tabulated MCNP mono-energetic simulations in an empty space were done beforehand for each of those three functions, describing the transport and interaction of the photons. A set of 100 energies, spread between 30 keV and 3000 keV following a function of energy were identified. Detector responses were obtained with bare sources simulations. The fluence rate out of a mono-energetic source surrounded by 0.1 cm of different shielding materials was also computed.
To determine the detector response function of a 1024 channel detector, 1024x1024 matrices were then computed by interpolating those previously made simulations, generating detector response to a mono-energetic bare source and fluence rates for each shielding configuration. To simulate a given nuclide configuration, a 1024 vector consisting of the line intensities was generated as an input. The spectral response in void will then be obtained by multiplying the different matrices.
Analytical methods have been used in conjunction to compute the fluence rate of infinite sources in different media (air, soil, special nuclear materials) using cross sections in specific materials. This approach seemed to be efficient at modeling SNMs, medical isotopes and was used to adjust the backscattering components for sources in air. X-rays could also be adjusted using analytical models.
A Single Value Decomposition was finally applied to select the most relevant components of the different scenarios for a given nuclide in order to keep the identification engine efficient and limit the number of cases per nuclide in the library.
The presentation will go over the different parts of the tools and the comparison between actual spectra and synthetically generated ones.
JOUBIN-JAME LECTURE: New applications of microscopy techniques in geoscience studies
Date/Time: October 1st, 2019, 4-5 pm
Speaker: Prof. L. Flora Sun
Affiliations: China University of Petroleum (Beijing)
Abstract: Microscopic techniques, including X-ray CT, electron microscopes, etc., have been used in geological studies for decades. In the past ten years as unconventional reservoirs and enhanced oil recovery gaining more importance, new applications of these techniques have been developed. Combined with the latest progress in microscopic instruments and computer science, the so-called digital rock physics is helping us better understand the mechanisms in the rocks on various scales. This lecture will provide an overview on the following topics: * Microscopic techniques of observing and analyzing rocks; *Geological phenomena observed using multiple techniques (case study: a profile of a shale gas reservoir); *Fundamental issues towards digital rock physics: image processing, segmentation, and statistics; *Multi-scale, multi-parameter simulations using the rock models extracted from rock images; *Discussion: how can we use the modeling results for reservoir studies?
A short course will be offered on October 22nd (to be confirmed). We will have some hand-on practice on image processing and rock physical statistics. An educational version of computer program will be ready for downloading before the course. All are welcome!
KEGS TORONTO SEP PRESENTATION: Downhole Physical Rock Properties and Televiewer Measurements - Borehole Data with a Purpose (Pizza and light refreshments will be provided by KEGS Foundation)
Date/Time: September 24, 2019, 4:30 pm
Speaker: Pamela Patraskovic
Affiliations: Program Manager, DGI Geoscience
Abstract: Downhole physical rock properties and televiewer measurements allow for a continuous and quantitative characterization of the subsurface that is valuable throughout a project lifecycle. Physical rock properties measured in situ include electrical, magnetic, gamma, density, NMR and acoustic velocity methods. These have multiple applications, including constraining geophysical inversions, characterizing mineralization, lithology and alteration, stratigraphic correlation and determining mechanical and elastic properties of formations. Acoustic and optical televiewer data allows for the precise strike and dip, depth and aperture of geotechnical and structural features to be obtained. Televiewer data provides accurate and rapid inputs for structural models, can recover information in areas of low RQD and provides a continuous, digital record of the entire borehole. Additionally, the televiewer can be run in both diamond (core) and reverse circulation boreholes. In cored boreholes (DDHs), televiewer data provides an alternative to oriented core, as well as more complete information in zones of low recovery or RQD. Televiewer data acquired in RC holes allows for the recovery of valuable structural and geotechnical information, as well as provides the potential to replace part of a diamond drill program with less expensive RC drilled boreholes. Downhole physical rock properties and televiewer measurements can be easily integrated with other geological and geophysical data sets, adding value to projects. Recent developments using advanced statistical characterization and AI to assist in interpretation are further enhancing the utility of comprehensive borehole measurements.
Title: The Physics and Statistics of Earthquakes
Date/Time: September 17th, 2019, 4-5 pm
Speaker: Prof. Robert Shcherbakov
Affiliations: Department of Earth Sciences and Department of Physics and Astronomy, University of Western Ontario
Abstract: The majority of earthquakes occur unexpectedly and can trigger subsequent sequences of events that can culminate in more powerful earthquakes. This self-exciting nature of seismicity generates complex clustering of earthquakes in space and in time. This clustering is a result of several physical mechanisms operating in the seismogenic crust. One of them is the triggering by preceding earthquakes that can lead to a cascade of events with a complicated branching structure. This is typically exemplified by ubiquitous aftershock sequences. The occurrence of aftershocks is a signature of physical systems exhibiting relaxation phenomena. They are observed in various natural and experimental physical systems. Empirical observations reveal that earthquakes obey several well defined non-trivial empirical laws in magnitude, temporal, and spatial domains. In many cases their characteristics follow scale-invariant distributions. The occurrence of aftershocks displays a prominent temporal behavior due to time-dependent mechanisms of stress and/or energy transfer.
In this presentation, I will overview several fundamental aspects of the occurrence of earthquakes and aftershocks. I will present a mechanical model that mimics the behavior of a seismogenic fault and explains the origin of aftershocks and the corresponding empirical statistical laws. I will show that the nonlinear viscoelasticity plays a critical role in the triggering of aftershocks. The proposed model also suggests that the power-law rheology of the fault gauge and underlying lower crust and upper mantle controls the decay rate of aftershocks. I will also present recent results related to the statistics and forecasting of earthquakes. Within statistical description of earthquakes, the problem of constraining the magnitude of the largest expected earthquake during a future time interval is of critical importance in mitigating earthquake hazard. I will address this problem by developing a methodology to compute the probabilities for such extreme earthquakes to be above certain magnitudes. The suggested approach can be implemented in current or future operational earthquake forecasting schemes, where the constraints on the magnitudes of future large earthquakes are taken into account. Here is a related article on Nature Communications.
Title: Physics of injection-induced earthquakes unveiled by seismic wave analysis and numerical
Date/Time: August 27, 2019, 3-4 pm
Speaker: Prof. Yihe Huang
Affiliations: University of Michigan
It is well known that fluid injection can induce earthquakes, but how fluid induces earthquakes initially and then contributes to the sequences of earthquakes following afterwards are still puzzling. Fluid pressure is usually considered as the major driving force when injection wells are close to the fault, whereas stress loading due to poroelastic deformation can take over when injection wells are more distant. We tackle three questions related to the physical processes of iinduced earthquakes: 1) How large is the change of fluid pressure or poroelastic stress? 2) Can fluid migration leave a signature in earthquake characteristics and ground motions? 3) Are earthquakes always a direct response to fluid injection?
Using spectral ratio approaches based on empirical Green's functions, we show that stress drops of induced earthquakes are indistinguishable from those of tectonic earthquakes in the central US, indicating that most stress released by induced seismicity has been accumulated by tectonic activities and the fluid-induced fault stress change is relatively small compared to the absolute fault stress levels. We also explore the magnitude-frequency distribution (MFD) and spatial-temporal evolution of induced seismicity by improving detections of small earthquakes using single-station waveform template matching. We find significant temporal variations in the MFD with respect to Gutenberg-Richter statistics that suggest temporal changes in deformation patterns and source mechanisms. By forward-modeling the rupture directivity of major induced earthquakes, we show that the 2016 Fairview earthquake that occurred near a high-pressure injection zone ruptured toward injection wells, whereas the 2011 Prague and 2016 Cushing earthquakes ruptured away. Such observations are consistent with numerical simulations with energy-based crack propagation and fluid flow, which suggest that high-pressure injection and low-stress faults can together favor ruptures propagating back to injection wells. We also find through simulations of induced earthquakes that aseismic slip plays an important role in inducing earthquakes besides fluid pressure and poroelastic stress. Small fluid-induced stress perturbations can trigger substantial aseismic slip that may either advance or delay subsequent earthquakes. Finally, our analysis of very small induced earthquakes in an injection experiment in France revealed their low stress drops, supporting the notion that there was tremendous amount of aseismic deformation during the experiment.
Title: Full intensity waveform inversion
Date/Time: August 6, 2019, 3-4 pm
Speaker: Prof. Yike Liu
Affiliations: Institute of Geology and Geophysics, Chinese Academy of Sciences
Abstract: Many full waveform inversion schemes are based on the iterative perturbation theory to fit the observed waveforms. When the
observed waveforms lack of low frequencies, those schemes may encounter
convergence problems due to cycle skipping if the initial velocity model
is far from the true model. To mitigate this difficulty, we propose a
new objective function that fits the seismic waveform intensity so the
dependence of the starting model can be reduced. The waveform intensity
is proportional to the square of its amplitude. Forming the intensity is
a nonlinear operation, which separates the original waveform spectrum
into an ultra-low frequency part and a higher frequency part, even for
data originally do not have low-frequency contents. Therefore,
conducting multi-scale inversions starting from ultra-low-frequency
intensity data can largely avoid the cycle-skipping problem. We
formulate the intensity objective function, the minimization process and
the velocity updating gradient. Numerical examples demonstrate that the
proposed method is very promising and can invert for the model using
data lacking of low frequency information. Here is the related article
on Geophysics in 2018.
Title: Linking kinematic source process with megathrust strength and dynamic ground motion prediction along the Nicoya Peninsula subduction megathrust
Date/Time: July 30th, 2019, 3-4 pm
Speaker: Prof. Hongfeng Yang
Affiliations: Earth System Science Programme, Chinese University of Hong Kong
Abstract: Frictional strength and slip weakening distance are crucial for earthquake generation and rupture propagation. However, in-situ frictional properties remain elusive on natural faults. Here we first constrain the frictional parameters along the megathrust that was ruptured during the 2012 Mw 7.6 Nicoya earthquake by spontaneous rupture simulations with constraint from kinematic source models and near-field GPS records. Our preferred dynamic model indicates a relative low fracture energy: 0.4X10^6 J/m2, with the D0~0.2 m and the strength drop ~4.0 MPa during the 2012 Nicoya earthquake. We then test the determinism of interseismic locking models in term of future rupture scenario and ground motion prediction along the Nicoya Peninsula megathrust. To do so, we estimate initial megathrust stress from locking, then initiate spontaneous ruptures, testing different nucleation points. For ruptures initiating at the hypocentre of the 2012 Nicoya earthquake, we find scenarios that approximate the coseismic slip distribution and final earthquake moment magnitude derived from seismic and geodetic observations. However, we find that only ~40% of nucleations develop into large earthquakes of Mw>7.2 based on present interseismic locking models. Of these events, those nucleated from deeper depths have a tendency for larger-amplitude shallow slip, suggesting increased tsunami potential. Furthermore, irrespective of the input locking models we do not observe rupture scenarios of earthquakes with intermediate magnitudes between 6 and 7, a result consistent with observations outside of the aftershock period in Nicoya. The results of hypocentre-dependent earthquake magnitudes and tsunamigenic potential not only pose challenges in estimating rupture extents from locking models, but also underscore the significance of quantitatively evaluating seismic and tsunami hazard in subduction zones. Here is the related article
on EPSL in 2019.