DLP Webinar: Airborne Micro-Tem and Deep Learning Inversion for Base of Sand Ultra-Resolution Mapping
| Instructor: | Dr Daniele Colombo |
|---|---|
| Duration: | 30 min + Q&A |
| Discipline: | Near Surface |
| Main topics: | Near surface, airborne transient electromagnetics, seismic corrections, deep learning inversion |
| Language: | English |
Next Delivery: 24 October 2023, 13:00 CEST
Request this Webinar Overview of all scheduled webinars Read Paper in EarthDoc Register Here
Attending webinars and access to recent EarthDoc material is free of charge for EAGE members, join here.Description
A key component in the seismic exploration of sand-covered areas is the characterization of sand dune velocities and the corresponding kinematic and dynamic corrections. Standard practices involve the mapping of the sand depositional basal surface (base of sand) followed by the application of empirically derived sand velocity functions for correcting the kinematic distortions. We developed a novel ultra-resolution imaging approach for the base of sand that utilizes helicopter-borne micro-transient EM (microTEM) to achieve a depth resolution of a few meters with continuous spatial coverage. A self-supervised ML inversion scheme is developed for general geophysical data and it is applied to microTEM data. Statistical sampling techniques under the name of Active Learning (AL) are applied to heavily reduce the samples needed for training. The dynamic learning model is supported by an iterative feedback loop between the inversion process and a ML system progressively adapting to the characteristics of the field data. The coupled physics-ML inversion scheme through AL sampling provides optimal generalization for field data applications. Results indicate an ultra-resolution mapping for the base of sand that reveals a complex paleo topography for the sand depositional surface. Seismic processing is enhanced to de-risk the exploration of deep low-relief structures.
Participants' Profile
Explorers, geologists, geophysicists, seismic processors, machine learning practitioners, inversion experts, EM geophysicists, airborne EM service providers.
About the Lecturer
Daniele Colombo works as Principal Scientist in EXPEC Advanced Research Center, Saudi Aramco. He holds a Ph.D. from University of Milan where he studied crustal and global seismology for the seismo-tectonic analysis of the Costa Rica’s subduction zone
using teleseisms and local earthquakes. Daniele had a successful career in the geophysical service industry such as WesternGeco/Schlumberger where he occupied technical and managerial roles. He then moved to Aramco upstream research in 2009 in the
Geophysics Technology Division of EXPEC ARC. Daniele has a track record of inventions and novel developments in geophysics related to joint inversion velocity model building, depth imaging in complex geology, surface-consistent FWI, surface-to-borehole
electromagnetics for waterflood monitoring, microseismic analysis for fracture characterization, physics-driven deep learning inversion. He authored 46 patents, more than 100 peer-reviewed publications and is the recipient of several awards, among
which the most recent include the 2020 AAPG Robert R. Berg Outstanding Research Award, 2022 Honorable Mention for Best Paper in Geophysics Journal, the 2020 and 2022 World Oil Award for Best Exploration Technology. Daniele’s current interests are
related to seismic and electromagnetic inversion, physics-driven machine learning, microseismic fracture characterization, seismic full waveform inversion and Multiphysics reservoir monitoring.