November 2023 – Events:

We would like to bring to your attention the following upcoming events within the community that may be of interest to you and your colleagues. Additional details are below.

  • UBC will be hosting the 7th International Symposium on Three-Dimensional Electromagnetics (3DEM-7) from Nov 13 to 17th. DETAILS BELOW
  • MDRU & DMEC are organizing a one day short course titled “EM for Geologists” to be held on Nov 17. DETAILS BELOW
  • BCGS Monthly talk – next item.

November 23, 2023 – Monthly Talk

Speaker: Jorge Lopez-Alvis, University of British Columbia

Title: Using convolutional neural networks to classify UXO with multi-component electromagnetic induction data

Date: Thurs, Nov 23, 2023

Time: 4:30pm to 5:30pm PDT

Location: In-person: Room 111 – 409 Granville Street, Vancouver, BC, V6C 1T2

Virtual: via Zoom, to be announced.

Note: This will be an in-person meeting

Abstract

Electromagnetic induction (EMI) methods are commonly used to classify unexploded ordnance (UXO) in both terrestrial and marine settings. Modern time-domain systems used for classification are multicomponent which means they acquire many transmitter-receiver pairs at multiple time-channels. Traditionally, classification is done using a physics-based inversion approach where polarizability curves are estimated from the EMI data. We developed a convolutional neural network (CNN) that classifies UXO directly from EMI data. The architecture of our CNN produces high-resolution results and can handle the multiple transmitter-receiver pairs and the per-line acquisition of multicomponent systems. The CNN is trained using synthetic data generated with a dipole forward model considering all possible UXO and clutter objects. In this presentation, we will discuss how we structure the data to be input into a neural network, how we design the network, and how training is performed. Our approach was tested on data acquired with the UltraTEMA system in the Sequim Bay marine test site. For this test dataset, our CNN-based approach detects all UXOs and classifies more than 90% as the correct type while also discriminating ~70% of the clutter. An overview of our workflow applied to this dataset and some discussion of the classification results will be presented.

Bio:

Jorge is a Postdoctoral Fellow from the Geophysical Inversion Facility (GIF) group at UBC. He has a background in near-surface geophysics and using machine learning in geophysics. His current research involves developing an approach for using neural networks to classify UXO directly from electromagnetic data. He completed a joint PhD program at Ghent University and University of Liège where he explored the use of deep generative models (or generative A.I.) to produce geologically realistic images from geophysical data.


3DEM-7 Symposium: Nov 13-17, 2023,   https://3dem-7.geosci.xyz/

The 7th International Symposium on Three-Dimensional Electromagnetic (3DEM-7) will be held on Nov 13-15, 2023 and we are looking forward to welcoming the EM community to Vancouver!

Registration is open at https://3dem-7.geosci.xyz/#registration and the abstract submission deadline has been extended to Oct 20: https://3dem-7.geosci.xyz/#abstracts-call. If you plan to attend, we appreciate early registration so that we can orchestrate the catering.

We are still seeking sponsors, so if your organization is interested in sponsoring 3DEM-7, please reach out to Lindsey Heagy (lheagy@eoas.ubc.ca) and Eldad Haber (ehaber@eoas.ubc.ca).

If you are in Vancouver and are willing to billet students / postdocs for the event, please contact Lindsey (lheagy@eoas.ubc.ca). There are some folks with limited travel funds who would like to attend, but require additional support.

Workshop: EM for Geologists (MDRU/DMEC): Nov 17, 2023

A one day short course covering electromagnetic methods, case studies, etc. See below link for the schedule and to register.

https://www.mdru.ubc.ca/training/sc125-em-for-geologists-workshop/

STUDENTS NOTE: Condor Consulting will cover the registration fee ($60) for any student who wants to attend. DMEC will cover $200 for any out-of-town student who attends to defray expenses (receipts required).

Guidelines for drones in geophysics

BCGS is pleased to support an initiative that provides guidelines for geophysical uses of drone technology. Drone Geoscience LLC has generously prepared a website with guidelines for using drones for geophysics.

See also our “Drones” page, under the “Resources” menu, or review materials from the BCGS drone workshop from May 2022.

Thanks are extended to Ronald S. Bell, Senior geoDRONEologist & Geophysicist with Drone Geoscience LLC.

February 2023 – Monthly Talk

Speaker: Dr. Mengli Zhang, Research Associate,  Center for Gravity, Electrical, and Magnetic Studies (CGEM), Colorado School of Mines

Title: Efficient geophysical data acquisition using ergodic sampling: Non-linear relationship between information sampling ability (ISA) and number of samples

Date: Thursday February 23, 2022

Time: 4:30pm to 5:30pm PST

Location: Virtual. Zoom link will be distributed via our newsletter in advance of the talk. Contact info@bcgsonline.org if you would like to attend but did not receive the newsletter with link (sent February 21, 2023).

Abstract:

Geophysicists use difference tools such as data display, modeling, and inversion to image subsurface of the earth. The denser the data are, the more details of earth model we can obtain. The price we pay for denser data is of course the higher cost for acquisition, especially for 3D data. We may default to an implicit assumption that the resolution of our earth model is linearly dependent upon the number of samples we can collect for geophysical data. This assumption may be rooted in Nyquist sampling theory. However, Nyquist sampling theory is a sufficient but not necessary condition. We have re-examined the necessity of such dense sampling in geophysical data acquisition and developed an ergodic sampling method and shows that the number of samples has a non-linear relationship with the information sampling ability (ISA). In contrast to Nyquist sampling, which requires a sufficient but larger than necessary sample set, ergodic sampling only acquires the core subset of samples that is both necessary and sufficient to gather the same information. Therefore, ergodic sampling can significantly decrease the number of samples compared with Nyquist sampling. We present our new sampling theory and demonstrate its application in the geophysical data acquisition. Our simulation and field data example show that the cost can be reduced by a factor up to 10. Equivalently, this result also means that it is possible to acquire 10 times more information when the same number of samples used in the traditional equi-spaced sampling is deployed using the ergodic sampling strategy.

Bio:
Dr. Mengli Zhang is a Research Associate in the Department of Geophysics at Colorado School of Mines. She is a geophysicist specialized in optimization of the exploration cycle from acquisition, interpretation, to discovery by incorporating economic factors. She is also an expert on efficient and economical multi-geophysical data acquisition using ergodic sampling theory. She obtained her BS in Information Engineering and MS degree in Information and Communication Systems from Xi’an Jiaotong University in China. She earned an MS degree in Geoscience from the University of Texas at Dallas, USA, and PhD degree in Geophysics from Colorado School of Mines. She has 10 years of industry experiences, first as a research geophysicist and then as a project manager and as Chief Geophysicist in the eastern Ordos Basin for China National Petroleum Corporation, where she applied information analyses to increase gas reservoir discoveries. She worked closely with geologists to select locations of more than 500 drilled wells, perform post-drilling analyses throughout the life cycle of wells including the production stage, and to improve interpretation and targeting methodology based on drilling and production results. Her current research has applications to the information-based economic geophysical data acquisition, which has the potential to significantly reduce the cost of exploration for energy and metals and to accelerate discoveries.