Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.

org/page/policies/terms
DOI:10.1190/leedff.2024.43.issue-1

SM
Special Section:
Microseismic monitoring

ISSN 1070-485X
January 2024 · Volume 43, No. 1
 We Give You The Best Image™
· True Amplitude consistency in all the modules.
· VTI, TTI and Orthorhombic Anisotropy.
· OVT Gathers.
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

· All imaging modules use Topography.

· Converted waves, OBC, VSP geometry.
· Gamma migration scans.
· Anti-aliasing:
bandpass.
triangles, boxcar, multiple

· Diffraction Imaging.
· Grid based tomography with horizon constraints.
· 360 degrees Wide Azimuth Tomography using
Extended Gathers.
· Updates using OVT Gathers.
· Targets, grid of targets, sparse target lines.
· Single and multiple parameter residual
velocity estimation.
· Automatic or manual residual velocity picking.
· Anisotropic parameter and velocity updates.
· Updates from topography.
· Automatically
points.
generated back-projection

· Layer freezing or global updates.

DOI:10.1190/leedff.2024.43.issue-1

· Global and local error visualization and QC.

· Automatic Vrms estimation.
· Optimization
nodes.
of data distribution on parallel

· Job monitoring. Add and subtract nodes

during runtime.

A Complete Depth Imaging System · Adaptive run-time parameter optimization.

· 5-D Beam Forming using PWD filters.
Z-Terra Software Solutions Modern 3D/4D Visualization and
Imaging Intuitive Graphical User Interfaces
ZTK Kirchhoff PSTM and PSDM. Anisotropic VTI, TTI and Orthorhombic Traveltimes. OVT Gathers. · Intuitive Graphic Interface on each module.
ZTK-DI Diffraction Imaging in Kirchhoff time and depth. Specularity Gathers.
ZRTM Two Way Reverse Time Wave Equation Migration.
· Base Map Viewer for input data and grid
definition.
ZFBM Gaussian Beam and Fast Beam Migration. · Beams and rays QC and visualization tools.
Z5D 5-D Interpolation using Azimuth Move Out (AMO).
ZSM Wave Equation Multiples Imaging Migration.
· Overlay of migration operator on input data
visualization.
ZFA Spectral Decomposition and Super Resolution Thin Layers Inversion. · 3-D/4-D synchronized visualization, input data,
gathers, stack, velocity model, travel-times,
Velocity Model Building Base Map.
ZTomo Grid based tomography constrained by horizons. Anisotropic updates.
ZTomo-360 Wide azimuth tomography using Extended Gathers.
ZBT Beam Tomography. High Resolution Tomography using Beam Migration.
ZESTANI Anisotropic Velocity Model Building starting from well logs.
ZMod Acoustic Modeling, Wave-Equation Illumination.
ZFWI Full Waveform Inversion in Data Domain. 281.945.0000
ZEOS Visualization and Velocity Model Building. Contact us today at sales@z-terra.com.

HOUSTON THE HAGUE RIO DE JANEIRO BUENOS AIRES KUALA LUMPUR

17171 Park Row, Suite 247 Stationsplein 6, 5th Floor Av. Nilo Pecanha 50 Av. Lima 575, 8th floor Dialog Tower, No. 15, Jalan PJU 7/5,
Houston, TX 77084 Voorburg 2270 JB Edificio De Paoli, Sala 1616 Buenos Aires C1073AAK Mutiara Damansara, 47810 Petaling Jaya,
USA The Netherlands Rio de Janeiro, RJ Brazil Argentina Selangor Darul Ehsan, Malaysia.
 WORK WORK

HARDER
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

SMARTER
Conventional processing
and imaging workflow DUG MP-FWI Imaging
(pre-processing, model building and (using field-data input)
Kirchhoff depth migration)

Vp/Vs Vp/Vs

Brine Sand Shale

DOI:10.1190/leedff.2024.43.issue-1

Gas Sand

P-impedance P-impedance

Less scatter = less uncertainty!

Equivalent crossplots of simultaneous AVA inversion results

BEX MC3D data courtesy of Multi-Client Resources

Quantitative interpretation requires true amplitudes and high signal-to-noise ratio.

DUG Multi-parameter FWI Imaging delivers superior pre-stack reflectivity outputs

for lithology and fluid prediction. It’s a complete replacement for the conventional
processing and imaging workflow. Ready to work? Send us a signal!

SURANCE SURANCE
AS AS
COMPASS
COMPASS

SE
SE

Let’s talk—contact info@dug.com

RVICES
RVICES

•
IS O • IS
dug.com/fwi
•

O 27001
•

9001
 The Leading Edge
Table of Contents
Special Section: Microseismic monitoring Departments
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

4..................... Editorial Calendar

7 ������������������Introduction to this special section: Microseismic monitoring
J. Le Calvez and E. Ay 5..................... President’s Page
56................... IMAGE '23 Review
8 ������������������Advances in hydraulic fracture characterization via borehole microseismic and strain monitoring
Y. Altowairqi, H. Alqatari, D. Colombo, and E. Turkoglu 58................... Erratum
59................... Announcements
16 ����������������Detection of microseismic events in continuous DAS data using convolutional neural networks
N. Boitz and S. Shapiro 59................... Board Report
60................... Reviews
24 ����������������Real-time passive seismic monitoring using DAS — Today’s solutions and remaining challenges
T. Mizuno and J. Le Calvez 61................... Membership
63................... Meetings Calendar
64................... Seismic Soundoff
30 ����������������Archaeological geophysical investigation of Uzun Rama Steppe kurgans, Goranboy
K. Bayramov, G. Alizada, S. Mammadov, V. Azimov, M. Abdullayev, C. Jodry, and M. Bano
On the cover: An artistic composition
37 ����������������Characterization of anisotropy in basin-scale subsurface using teleseismic receiver function analysis of a geophone and geophysical data
DOI:10.1190/leedff.2024.43.issue-1

Y. Li and A. Nikulin using Figure 2b from Mizuno et al. Cover

designed by Maria Gee.
46 ����������������A decade of technology advancement in seismic processing: A case study from reprocessing legacy sparse OBC data in
Kashagan Field
J. Park, C. Hyslop, A. Ouzounis, A. Wawrzynski, T. Vdovina, S. Lee, K. Hasner, W. D. Ibanez, S. Schreuder, Z. Tapalov,
A. Gabdullin, and N. Yergaliyev

Episode 208: ❘
Listen free On demand
Pioneering Seismic Imaging for
Energy and Sustainability
with Biondo Biondi

Episode 207:
Advancing Geosciences –
How SEG Foundation Makes a
Difference
with David Bartel, SEG Foundation Chair

Episode 206:
Arthur Cheng on His SEG Journey
from Student to President
with Arthur Cheng
seg.org/podcast

2 The Leading Edge January 2024

 The Leading Edge
SEG Board of Directors
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

PRESIDENT TREASURER DIRECTOR AT LARGE

Arthur C. H. Cheng Mike Mellen Olga I. Nedorub
The Chinese University of Hong Kong Houston, TX, USA ConocoPhillips
Houston, TX, USA Houston, TX, USA

PRESIDENT-ELECT PAST PRESIDENT DIRECTOR AT LARGE

John Eastwood Kenneth M. Tubman Catherine Truffert
Calgary, AB, Canada SAExploration IRIS Instruments
Houston, TX, USA Orléans, France

FIRST VICE PRESIDENT DIRECTOR AT LARGE DIRECTOR AT LARGE

Mauricio Sacchi Sergio Chávez-Pérez Constantine Tsingas
University of Alberta Mexican Petroleum Institute Saudi Aramco
Edmonton, AB, Canada Ciudad de México, Mexico Dhahran, Kingdom of Saudi Arabia
DOI:10.1190/leedff.2024.43.issue-1

SECOND VICE PRESIDENT DIRECTOR AT LARGE CHAIR OF THE COUNCIL

Marianne Rauch Ana Curcio Allen J. Bertagne
Houston, TX, USA Proingeo SA BRT Energy Advisors LLC
Buenos Aires, Argentina Houston, TX, USA

VICE PRESIDENT, PUBLICATIONS DIRECTOR AT LARGE

Kyle Spikes Lillian G. Flakes
The University of Texas at Austin GeoSoftware
Austin, TX, USA Richardson, TX, USA

The Leading Edge® (Print ISSN 1070-485X; Online ISSN 1938-3789) is published monthly by the Society of Exploration Geophysicists, 125 W. 15th St., JIM WHITE, Executive Director
Suite 100, Tulsa, Oklahoma 74119 USA; phone 1-918-497-5500. Periodicals postage paid at Tulsa, OK and at additional mailing offices. SCOTT SUTHERLAND, Managing Director, Business and
Geoscience Technology
Print subscriptions for professional members of the Society in good standing are included in membership dues paid at World Bank IV rates. JENNIFER COBB, Managing Director, Publications and Membership
Dues for Active and Associate Members for 2024 vary depending on the three-tiered dues structure based on World Bank classification of the JENO MAVZER, Director, Journals
member’s country of citizenship or primary work residence. Students may purchase print subscriptions by paying a discounted fee during dues STEVE BROWN, Managing Editor
renewal or by contacting members@seg.org. Nonmembers may obtain online subscriptions by contacting books@seg.org. Copies of single issues KELSY TAYLOR, Associate Editor
of The Leading Edge® may be purchased by inquiry to books@seg.org. MARIA GEE, Graphic Designer
Corporations, universities, and other institutions can purchase access to The Leading Edge® via standalone subscriptions or one of several SARAH WEATHERS, Publishing Platform Manager
available subscription package options. SEG’s subscription pricing features three academic and three corporate tiers, providing affordable STACY BAKER, Publishing Platform Analyst
access to applied-geophysics content for institutions of any size. SEG subscription packages offer significant savings on the price of single-
publication subscriptions. Special pricing is available for organizations with multiple sites and for institutional consortia. For more information on Editorial information: 1-918-497-5503, sbrown@seg.org
package pricing, including single-site and multisite institutional subscriptions, please contact Patrick Riley at sales@seg.org or 1-918-497-5531. Advertising information and rates:
Institutional subscriptions for print delivery are available only with the purchase of a digital subscription. Rates are subject to change without 1-918-497-5524
notice. For more information, please visit https://seg.org/subscriptions. Kathy Gamble, kgamble@seg.org
Advertising rates for The Leading Edge® will be furnished upon request. No advertisement will be accepted for products or services that cannot Subscription information:
be demonstrated to be based on accepted principles of the physical sciences. For more information, contact kgamble@seg.org. Members, members@seg.org
Statements of fact and opinion are made on the responsibility of the authors and advertisers alone and do not imply an opinion on the part of Nonmembers, books@seg.org
the officers or members of SEG. Copyright 2024 by the Society of Exploration Geophysicists. The Leading Edge®, SEG®, and the SEG logo are Institutions, Patrick Riley, 1-918-497-5531,
registered marks of the Society of Exploration Geophysicists. All rights reserved. Material may not be reproduced without written permission. priley@seg.org
Printed in the USA. POSTMASTER: Send changes of address to
The Leading Edge
125 W. 15th St., Suite 100
Tulsa, OK 74119 USA

January 2024 The Leading Edge 3

 The Leading Edge THE LEADING EDGE
Editorial Calendar EDITORIAL BOARD
CHAIR
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Issue Special section theme Due date Guest editors Chester J. Weiss
Sandia National Laboratories
February 2024 Future of applied geophysics past due Yaoguo Li Albuquerque, NM, USA
Michael Wilt1 cjweiss@sandia.gov
Chester J. Weiss1
March 2024 Imaging faults and fractures past due Lily Horne
Molly Turko Heather Bedle
Heather Bedle1 University of Oklahoma
Norman, OK, USA
April 2024 Gravity, electrical, and magnetic past due Alan Morgan hbedle@ou.edu
methods Jiajia Sun
Maurizio Fedi
Irina Filina
May 2024 Rock physics past due Gregor Baechle Niels Grobbe
University of Hawai‘i at Mānoa
Jeremie Dautriat Honolulu, HI, USA
Laurent Louis1 ngrobbe@hawaii.edu
June 2024 Subsurface uncertainty 1 Feb 2024 David Lubo-Robles
Matt Walker
Madhav Vyas1
July 2024 General submissions 15 Feb 2024 TLE Editorial Board Joël Le Calvez
SLB
Chester J. Weiss1 Sugar Land, TX, USA
DOI:10.1190/leedff.2024.43.issue-1

jcalvez2@slb.com
August 2024 Geophysics and sustainability 15 Mar 2024 Julia Correa
Aleksei Titov
Vladimir Kazei1
September 2024 Focus on the Mediterranean region 15 Apr 2024 Walter Rietveld Madhav Vyas
Ivica Mihaljevic BP America
Maha Khattab Houston, TX, USA
madhav.vyas@bp.com
Ramesh Neelamani1
October 2024 Geophysical methods 15 May 2024 Rich Krahenbuhl
in archaeology Michael Wilt1
November 2024 Optical fiber 15 Jun 2024 Erkan Ay Vladimir Kazei
Aramco Americas
Joël Le Calvez1 Houston, TX, USA
vladimir.kazei@aramcoamericas.com
December 2024 Reservoir characterization 15 Jul 2024 Satinder Chopra
Tom Davis
Heather Bedle1
1
TLE Editorial Board coordinator Laurent Louis
Aramco Americas
Houston, TX, USA
laurent.louis@aramcoamericas.com

Ramesh (Neelsh) Neelamani

ExxonMobil
Houston, TX, USA
ramesh.neelamani@exxonmobil.com

TLE publishes special sections and standalone articles covering all aspects of applied geophysics and related
disciplines. Submission of articles is open to all. Please submit articles via the online manuscript submission
system at https://mc.manuscriptcentral.com/tle. Submission instructions are available at https://library.seg.org/
Michael Wilt
TLE-authors. For full descriptions of special section themes, see https://library.seg.org/TLE-sections. TLE Editorial Lawrence Berkeley National Laboratory
Board coordinators work with guest editors to coordinate and support the review process and also may serve as Berkeley, CA, USA
mwilt@lbl.gov
guest editors. For additional assistance, contact tle@seg.org.

4 The Leading Edge January 2024

 President’s Page
Maintaining the pipeline
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

of geophysics talent
This month’s author:
Joseph H. Y. Ma, SEG Community Manager

S tudents and emerging professionals

often commit to career choices that
align with their expectations of incentives,
published by the Society of Petroleum
Engineers in 2021 (SPE, 2021) suggested
a decreasing trend in compensation levels
suitable candidates for future opportunities.
In larger organizations, the requisition
procedures involve many departments and
their values, and their passions. Curiosity for submanagerial-level professionals since approvals and can take up to 2–6 months
about earth tectonics, an ambition to 2018. Recent substantial retrenchments to complete. Administrative screening,
advance outer space planetary exploration, and hiring freezes (in 2015, 2018, and alongside technical and behavioral evalu-
or a desire to work in the energy and min- 2020) have caused our community to lose ations during role fitting, adds another layer
eral industries that support the world’s two to three cycles of intakes and graduates. of challenge. It is not uncommon to hear
economic growth are but a few of the factors A mere decade is more than enough time of suitable profiles being overlooked or
that have motivated young talents to pursue to create an impression among students screened out and never getting a chance
the study of geophysics and ultimately majoring in the geosciences that their career for consideration.
DOI:10.1190/leedff.2024.43.issue-1

dedicate their careers to it. However, recent prospects are suboptimal. This concern is These job-search difficulties can be
years have seen a noticeable decline in both only compounded by increasing higher discouraging to those seeking employment
student enrollment in university geophysics education costs globally. in the energy industry. Some who miss out
programs and young professionals pursuing on securing employment use it as an oppor-
geophysics-related professions. According Hiring practices tunity to further their studies and expand
to data in the Australian Geoscience For those who have finished their aca- their knowledge in the geosciences, but
Tertiary Education Profile, in 2020 there demic training, finding a suitable oppor- many will move on to other job markets.
were roughly 30% fewer earth science tunity and kick-starting a geophysics career As a result, the industry misses out on
undergraduates in the United States and is no easy task. In an unstable employment existing talent.
United Kingdom than there were in 1980 climate, industry and research institutes
(Cohen, 2022). are much more prudent in their hiring The retreat of globalization
Still, many brilliant young minds are practices, tending to prioritize the recruit- Some students are attracted to geo-
passionate about the geosciences, but they ment of experienced senior professionals physics because it is a field that embraces
face difficulties in choosing this profes- rather than committing the time and global opportunities, encourages global
sional path. What extra effort can we, as a resources to staffing junior positions that collaboration, and tackles global chal-
professional community, make to maintain will require more on-the-job training. lenges. The COVID-19 pandemic sty-
the geophysics talent pipeline? Let’s begin In the competitive landscape for mied some of this appeal. For instance,
by understanding the issues that next- entry-level positions, those who took the many scholarship and employment
generation geophysicists face. initiative in building their careers early opportunities are now reserved for
in their academic journey find that the domestic candidates. Secondly, obtaining
Career prospects benefits of their efforts are diminishing. a visa is more difficult. Those seeking to
As a profession, geophysics in the past Traditionally, internships and industrial pursue higher education and develop their
40 years has been driven in large part by project collaborations have been seen as career in other countries now face greater
the needs of the oil and gas industry, but valuable opportunities for students and limitations. Industry is also becoming
what once drew talent to the field has emerging researchers, allowing them to less willing to offer visa sponsorship or
become something of an inhibiting factor. establish work relationships and confront expatriation to acquire talents across the
Prone to technological advancement, eco- real-world challenges. However, investing globe. Attendance at international confer-
nomic cycles, and geopolitics, the oil and extra time and effort beyond regular aca- ences and workshops, which are impor-
gas industry is a volatile one. This implies demic commitments no longer guarantees tant platforms for professional develop-
it is a lot harder for professionals in the a return offer from the organization where ment and networking, also constitutes a
industry to receive renumeration and career a placement occurred. challenge. Take the recent IMAGE ’23
stability comparable to that enjoyed by Furthermore, not all companies have as an example. The event is as successful
previous generations. A salary survey a talent pool system in place to consider as ever, but a number of papers were

January 2024 The Leading Edge 5

 withdrawn and speakers replaced, mostly versatile training that can lead to many Additionally, a Student Chapters cam-
due to visa issues. Co-organized with the extended opportunities in science, engi- paign has been launched to encourage close
American Association of Petroleum neering, and computing. Very few disci- engagement among 200 student chapters
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Geologists, IMAGE is contracted to be plines exceed geophysics in its ability to in 49 countries. Established in 2020, the
held in Houston until at least 2025. expose students to mathematics, physics, Early-Career Subcommittee of the SEG
Perhaps consideration can be given to programming, instrumentation, commu- Research Committee is another great
rotating the event to various continents nications, project planning, and hands-on example. It provides input on the research
to better support members and geophysics fieldwork simultaneously. This training, direction of SEG while putting a spotlight
practitioners worldwide. together with a curiosity about the earth, on early-career views. The subcommittee
will make geophysics graduates competi- is also currently working with other SEG
Emergence of new opportunities tive in an era when many new challenges groups, including the Emerging
The energy industry is under consis- and opportunities emerge. For instance, Professionals International Committee, to
tent public scrutiny related to environ- some professionals who have exited the enhance technical career support and to
mental impact. With their greater aware- energy industry in the past few years have revitalize meeting and workshop technical
ness of social-environmental topics, a found their skill sets highly adaptable to programs. All of these initiatives are meant
younger generation may not be as readily alternative career pathways such as data to provide extra platforms for student and
attracted to “mainstream” geophysics as science, medical imaging, and geotechnical early-career participation.
past generations were. Nevertheless, engineering, to name but a few. Even in What else should we do? It’s up to
adversity often brings with it opportunity. the established energy and mining indus- everyone in the community to propose ideas
As a study of the earth’s properties and tries, the need for geophysics specialists is and pitch in.
processes, geophysics finds itself still not going away. What must be strength- Sustaining the talent pipeline will
relevant to the world’s energy transition ened is the support and attention we give require the combined efforts of academia,
and climate mitigation initiatives. Hot to emerging professionals. industry, and professional societies. It is
DOI:10.1190/leedff.2024.43.issue-1

topics related to the energy transition, going to take some corner turning and
such as geothermal energy production, SEG’s early-career support on the rise long-term investment, but let us recognize
hydrogen production, and carbon capture SEG is solidifying the support it the challenges and needs of the next gen-
and storage, may rekindle a new genera- provides students and early-career profes- eration and do our part to ensure those
tion’s interest in pursuing geophysics as a sionals. Existing programs such as SEG needs are met and that the talent is nur-
field of study, followed by seeking a career EVOLVE, the SEG/Chevron Student tured. There can only be a community when
in the energy industry. More attention is Leadership Symposium, Geoscientists there are dedicated members.
also being given to climate sciences, for without Borders, and the SEG Women’s
which geophysics is the backbone. Network Committee continue to offer Acknowledgments
Cryosphere geophysics, water resources exposure and support to students of vari- Discussions with Heming Wang,
sciences, and volcanology are examples. ous aspirations. The online SEG Daniel Afolabi, Silas Adeoluwa Samuel,
Geophysics graduates must be made aware Community, which launched in 2022, is Vayavur Rajesh, Ziyi Francis Yin, and a
of alternative career choices where their providing a convenient platform for few anonymous SEG early-career mem-
expertise can make a meaningful impact, knowledge and information exchange bers contributed to the writing of this
such as government laboratories, nongov- among its more than 2200 participants article. The author is grateful for their
ernmental organizations, and technology — half of whom are under the age of 30. contributions.
start-ups. The online SEG Community is one area
where experienced professionals could References
Embrace and adapt to changes help immediately by taking a more active Cohen, D. R., 2022, Australian geoscience
Transformation is required to react to role in assisting students and early-career tertiary education profile 2003–2021:
Australian Geoscience Council, https://
the changes the community is facing. Let professionals with knowledge of com-
www.agc.org.au/resources/reports/austra-
us take this as an opportunity to reposition munity dynamics and emerging opportu- lian-geoscience-council-report/, accessed
in a way that maintains the developmental nities. The next generation would value 11 December 2023.
support afforded to our next generation of this type of mentorship greatly. Anyone SPE, 2021, SPE salary survey report 2021,
geophysicists and SEG members. Rather can become a member of the online SEG https://www.spe.org/en/industry/oil-and-
than career-orientation driven, we can now Community by simply signing up at gas-salary-survey/, accessed 11 December
think of geophysics as fundamental and https://seg.org/seg-community. 2023.

6 The Leading Edge January 2024

 Introduction to this special section: Microseismic monitoring
Joël Le Calvez1 and Erkan Ay2
https://doi.org/10.1190/tle43010007.1
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

M icroseismic monitoring is an important technique for

understanding the behavior of subsurface structures and
assessing the risks associated with human activities that can induce
high-resolution images of the subsurface, while joint inversion
combines data from multiple sources, such as seismic, electro-
magnetic, and geodetic data, to create more accurate and com-
seismic events. However, it is also a challenging problem due to prehensive models of subsurface behavior.
the low-energy radiation of the microseismic events, which can To lead off this special section, Altowairqi et al. present recent
make them difficult to detect and locate accurately. In recent years, advances in hydraulic fracture characterization that leverage both
several innovations in technology and data analysis have addressed borehole-based fiber-optic-acquired microseismic data and strain
the limitations faced in the past. data, perhaps helping bridge the gap between stimulated reservoir
One major innovation in microseismic monitoring has been volume and estimated stimulated volume.
utilizing the development of distributed acoustic sensing (DAS) Another innovation in microseismic monitoring has been
technology. DAS uses fiber-optic cables to detect and measure the development of machine learning algorithms for data analysis.
tiny subsurface vibrations caused by microseismic events. DAS Machine learning can be used to analyze large volumes of data
technology has several advantages over traditional seismometers, from multiple sensors and identify patterns and correlations that
including a higher frequency spectrum, greater spatial resolution, may be difficult for humans to detect. For example, machine
and the ability to monitor larger areas with fewer sensors. learning algorithms can be used to identify microseismic events
In addition to technological innovations, there have also been that are similar in location, magnitude, and frequency, which
developments in the use of microseismic monitoring for geome- can help improve the accuracy of event detection and location.
chanical modeling and reservoir simulation. By combining micro- Machine learning can also be used to predict the likelihood of
DOI:10.1190/leedff.2024.43.issue-1

seismic monitoring data with other geophysical and geologic data, future microseismic events based on historical data and other
such as well logs and seismic surveys, engineers can create detailed factors. Such predictions can help operators optimize production
models of subsurface structures and behavior. These models can and manage risk.
be used to optimize production and reduce the risk of induced Boitz and Shapiro present how convolutional neural networks
seismicity by simulating different scenarios and evaluating their enable DAS-acquired microseismic event detection, while Mizuno
impact on the subsurface. and Le Calvez discuss a new DAS-based real-time processing
Despite these innovations, microseismic monitoring remains workflow that combines image-based machine learning and
a challenge due to the complexity of the subsurface and the traditional processing approaches updated for DAS-acquired data.
variability of microseismic events. Microseismic events can be Microseismic monitoring has wide-ranging applications in
influenced by factors such as stress changes, fluid flow, and rock industries such as oil and gas, mining, geothermal energy, and
properties, which can make them difficult to predict and under- civil engineering. Recent innovations in technology and data
stand. In addition, the small size and low energy of microseismic analysis have improved the accuracy and resolution of microseismic
events can make them difficult to detect and locate accurately, monitoring, but there is still much to learn about the behavior of
especially in noisy or heterogeneous subsurface environments. subsurface materials and the factors that influence microseismic
To address these challenges, researchers are exploring new events. Ongoing research in microseismic monitoring is focused
techniques for data acquisition and processing, such as passive on developing new techniques for data acquisition and processing,
seismic imaging and joint inversion of multiple data types. as well as improving our understanding of the subsurface through
Passive seismic imaging uses ambient seismic noise to create geomechanical modeling and reservoir simulation.

1
SLB, Sugar Land, Texas, USA. E-mail: jcalvez2@slb.com.
2
Shell, Houston, Texas, USA. E-mail: erkan.ay@shell.com.

Special Section: Microseismic monitoring January 2024 The Leading Edge 7

 Advances in hydraulic fracture characterization
via borehole microseismic and strain monitoring
Yazeed Altowairqi1, Hala Alqatari2, Daniele Colombo1, and Ersan Turkoglu1
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

https://doi.org/10.1190/tle43010008.1

Abstract related to the direction of the wave arrival, and the absence of
We analyze the impact and challenges associated with micro- polarization information. The full utilization of FO in microseismic
seismic monitoring in the development of unconventional reservoirs. monitoring remains the subject of ongoing research in contrast to
We detail advanced methods that address some of these challenges the well-established seismological analysis framework utilized for
and discuss emerging data analysis and acquisition technologies geophone data analysis. A combination of geophone arrays and
such as those provided by fiber-optic monitoring. During the FO installations reduces the uncertainties and enhances the quality
development of unconventional resources, it is important to under- of the microseismic analysis and the related fracture characteriza-
stand the stimulated rock volume obtained from the hydraulic tion. Utilization of dense arrays incorporating geophones and point
fracturing treatment. Microseismic monitoring is one of the sources receivers, or DAS, enhances the sampling and fidelity of the
of information for estimating stimulated rock volume in which recorded wavefields and generates opportunities for characterizing
accurate hypocenter calculations provide the horizontal and vertical geologic features at unprecedented scales. Densely sampled wave-
fracture geometry and extent. We developed hypocenter waveform- fields provide an avenue for analysis that could not be performed
based relocation and collapsing techniques and demonstrated these via conventional acquisitions such as the sampling of the near-field
on synthetic and field data to obtain an enhanced description of strain (Luo et al., 2021), and the development of image-based
the fracture geometry. We further provide the initial analysis of a location approaches (Wang and Alkhalifah, 2018). Machine
DOI:10.1190/leedff.2024.43.issue-1

recently acquired data set composed of multiple monitoring wells learning techniques can be also utilized to explore new aspects of
instrumented with geophones and fiber optics operating simultane- signal enhancement for dense seismic arrays records.
ously. Microseismic events recorded by the multiwell dual geophone/ In this article, we detail the development of microseismic
fiber setup show meaningful signal-to-noise ratio to enable a joint hypocenter location techniques applied to conventional micro-
interpretation. Additional strain and temperature data provided seismic data recorded with geophone arrays and introduce the
by dynamic fiber-optic measurements extend the interpretation initial analysis of a recent simultaneous acquisition of a hybrid
capabilities of the hydraulic fracturing phenomenon while providing geophone and FO monitoring setup. This contribution provides
an avenue for advanced research on fracture characterization. an overview of the challenges and opportunities related to the
utilization of microseismic data in the development of unconven-
Introduction tional reservoirs. We first analyze the opportunities presented by
During the development phase of unconventional resources, recently developed hypocenter locations methods and detail the
it is necessary to understand natural and hydraulic fracture geometry field results of a recent hybrid acquisition utilizing geophones and
to estimate the stimulated rock volume and to evaluate the well FO in separate horizontal wells. The study involves five parallel
performance. Microseismic monitoring using surface and/or wells. Among those wells, two lateral wells were simultaneously
downhole geophone arrays has become more commonly utilized stimulated and monitored using a dense downhole geophone array
in hydraulic fracturing operations due to its ability to determine deployed in a horizontal well drilled in the middle. The micro-
the fracture network and map the fracture geometry. Recent seismic activity was also captured by FO cables deployed in two
developments in downhole acquisition involve fiber-optic (FO) additional horizonal wells drilled side by side next to the stimulated
sensing, including low-frequency distributed acoustic sensing wells. The ongoing study reviews the current practices and intro-
(DAS), distributed strain sensing (DSS), and distributed tempera- duces recent advances and future opportunities in acquisition and
ture sensing (DTS). The utilization of FO can improve microseismic data analysis technologies for enhancing hydraulic fracture descrip-
monitoring acquisition by providing wide recording apertures, tion and characterization.
high spatial resolution, and low operational costs. Such measure-
ments can also be used to record strain rate or strain for hydraulic Method
fracturing monitoring in an offset well to provide additional Waveform-based relative relocation technique. The developed
parameters for fracture characterization. Among the shortcomings methodology represents a postprocessing detection and relocation
of FO recordings, however, are generally lower sensitivity compared technique that capitalizes on the concept of earthquake doublets,
to three-component (3C) geophone arrays, variable sensitivity i.e., a pair of seismic event waveforms exhibiting similar

Manuscript received 31 August 2023; accepted 2 November 2023.

1
Saudi Aramco, EXPEC Advanced Research Center, Dhahran, Saudi Arabia. E-mail: yazeed.altowairqi@aramco.com; daniele.colombo@aramco.com;
ersan.turkoglu@aramco.com.
2
Saudi Aramco, Unconventional Resources Exploration and Characterization, Dhahran, Saudi Arabia. E-mail: hala.alqatari@aramco.com.

8 The Leading Edge January 2024 Special Section: Microseismic monitoring

 characteristics in terms of P-and S-wave polarization as well as
focal mechanism at a given station (Arrowsmith and Eisner, ​FD​  −1​  ​​(​T ij​  * ​​  ​( f )​​Sij​  * ​​  ​( f )​)
2006). These doublets are often found near one another, forming ​Cij​ ​(τ)​ = ​  ____________
  
  
_____________  ​​, (1)
​√  
​∑t​​  ​Tij​  2​  (​ t )​​​∑ t​​  ​Sij​  2​  (​ t )​​ ​
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

part of the same sequence of seismic activity, such as fracture

systems (Slunga et al., 1995). By analyzing waveform similarities
between earthquake doublets, it becomes possible to accurately where ​​FD​  −1​  ​​ denotes the inverse Fourier transform, and ​​Tij​  * ​​ ​(f )​​ is the
determine the relative spatial locations of the microseismic events. complex conjugate of the Fourier transform of T(t) for a receiver
An effective approach to conduct this analysis entails computing station i and receiver component j. The subsequent step involves
the cross-correlation function between a set of well-located events, averaging the normalized cross correlograms for receiver compo-
denoted as template events, and a group of target events (Gibbons nents. For a P-wave and an S-wave template waveform, this can
and Ringdal, 2006; Reyes-Montes et al., 2009). Template events, be respectively formulated as
in contrast to target events, exhibit high signal-to-noise ratio
(S/N) and high magnitudes, with their spatial location and origin ​Ci​P​​ = ​1_3​​(​Ci​P,N
​ ​+ ​Ci​P,E ​ ​)​
​ ​+ ​Ci​P,Z (2)
time determined using a conventional approach, e.g., the absolute-
based location technique. In addition to having high-quality ​​Ci​  S​  ​ = 21_​ ​(​Ci​  S,N ​  ​)​​,
​  ​+ ​Ci​  S,E (3)
waveform signals, the selection of template events should be made
based on their distinctive waveform characteristics (Caffagni where ​Ci​  P,N
​  ​​, C
​​ i​  P,E
​  ​​, and C
​​ i​  P,Z
​  ​represent the normalized P-wave cross
et al., 2016). Consequently, template events should be spatially correlogram of the north, east, and vertical components for receiver
distant from one another to maximize the detectability of doublet station i, respectively, and ​​Ci​  S,N ​  a​​ nd ​​Ci​  S,E
​  ​​ denote the S-wave cross
events originating from distinct seismic sources and to reduce the correlogram of the north and east components, respectively. The
occurrence of duplicated detections, namely target events that are next step is to create a 3D mesh around each template event with
detected by two or more template events (Meng et al., 2018). a predefined grid size related to velocity and frequency. Each grid
One of the advantages offered by the developed master relative node represents a potential source location where a target event
DOI:10.1190/leedff.2024.43.issue-1

location technique, when contrasted with the absolute location can be relocated. The theoretical traveltime differences between
techniques, resides in its capability to detect and accurately locate the template event location and each potential source location are
weak microseismic events. Moreover, this method exhibits robust- calculated for each receiver station. The following step involves
ness in the presence of noisy waveform signals and inaccuracies aligning and stacking the averaged cross-correlation sequences
in the velocity model under the assumption that similar events, based on the corrected traveltimes associated with each potential
occurring in the same location or with distances significantly source location. This corrected traveltime is determined by con-
smaller than the source-receiver distance, share the same errors sidering the picked arrival time in the template waveform and the
in the velocity model (Geller and Mueller, 1980). These attributes theoretical traveltime difference between template and potential
make this approach particularly suitable for downhole acquisition source locations. It is expressed as follows:
geometry, where events with low S/N and local heterogeneities
in the velocity model are frequently encountered. ​τik​ ​ = τ​ i​  o​  ​(​rm​ ​)​+ ​τi​  c​  ​(​rk​​)​− ​τi​  c​  ​(​rm​ ​) ​, (4)
As the presented technique is implemented to detect additional
weak microseismic events and to improve the location precision where ​​τi​  o​  ​(​rm​ ​)​​ represents the picked arrival time of the template
of the already-located event hypocenters, we will focus in this event and ​​τi​  c​  ​(​r​k​)​​ and ​​τi​  c​  ​(​rm​ ​)​​ denote the theoretical arrival time
article on the relative location aspects. The workflow follows the calculated between receiver i and the potential source location rk
approach of Meng et al. (2018), which starts with microseismic and theoretical arrival time between receiver i and the template
events detected and located by conventional processing approaches, location rm. Equation 4 is then used to calculate the stacked cross
namely the short-term average to long-term average method and correlogram for each potential source location as
the absolute-based location technique, respectively. Subsequently, _ _
a set of reference events (template events) is deployed based on C​(​rk​​, t)​= _
​ ​1 ​​∑ i=1
2N
N
​ ​Si​​(t + ​τik​S ​)​,
​C ​Pi​​(t + ​τik​P​)​+ C (5)
the selection criteria discussed earlier. These template waveforms _ _
are then utilized to compute the cross-correlation coefficient where N represents the number of receiver stations, and C ​
​​​  ​  Pi​  ​​ and C ​
​​​  ​  Si​  ​​
sequence between each template event and the continuous time indicate the P-wave and S-wave cross correlograms, respectively.
signal of the rest of the events (target events). To achieve this, The coherence function in equation 5 can be constrained by
P- and S-wave templates are created by extracting waveform back-azimuth or incidence angle as follows:
segments based on the respective P- and S-wave arrival times for
​​{ω ​Φik​ ​​ t + ​τ ik​P ​​ ​+ ​(1 − ω)​____________​}​​, (6)
_ _
​C ​​Pi​​ t + ​τik​P​​ ​+ ​C ​​Si​​ t + ​τik​S​​ ​
each event. The sliding-window cross-correlation function is then ​C​(​rk​​​, t)​= _
​1 ​∑ i=1
2N
N
( ( ) ( )
2 )
calculated between the P-wave/S-wave template waveforms and
the time signal of the target events across all receiver stations and where
receiver components. If T (t) and S(t) represent the template
waveform signal and target waveform signal, respectively, the ​Φik​ (​ t)​= exp​{− 50 ​​(​(​Φik​  C​  ​− ​Φim ​  o ​)​  ​)​​​  2​}​​,
​  C ​)​  ​− ​(​Φik​  o ​​  − ​Φim
normalized cross-correlation formula can be expressed as
(Arrowsmith and Eisner, 2006)

Special Section: Microseismic monitoring January 2024 The Leading Edge 9

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 1. Schematic diagram of the waveform-based relative relocation method. The generated 3D grid is centered at the selected template event.

where ​​Φik​  C​​ and ​​Φ​  imC​   ​​ represent the theo-
retical back-azimuth or incidence angle
calculated for the potential source
location and template event with
respect to receiver i, respectively, and​​
Φ​  iko​  ​​ and ​​Φim
​  o ​​ denote the polarization
angle of template and target waveform,
respectively. An event is detected if the
cross-correlation coefficient exceeds a Figure 2. Schematic diagram illustrating the iterative approach of the collapsing and remapping technique.
predefined threshold and is subse-
DOI:10.1190/leedff.2024.43.issue-1

quently located to the grid node with the maximum cross-

correlation value (Figure 1).
Collapsing and remapping technique. The collapsing and remap-
ping approach (Jones and Stewart, 1997) is proposed to be employed
following the detection and location phase of microseismic data
processing, or it can be applied in cascade after the waveform-based
relative relocation technique. The primary objective is to leverage
the inherent location uncertainties associated with each micro-
seismic event, with the aim of generating simplified and interpre-
table microseismic cloud structures. Location uncertainty can be
expressed in the form of confidence ellipsoids, which are three-
dimensional geometric representations of data uncertainty. In the Figure 3. Schematic diagram of the synthetic survey configuration: (a) map view and cross-
context of seismic event locations, these ellipsoids are used to section views from the (b) y and (c) x axes of the acquisition. The red circles represent
quantify spatial uncertainty associated with microseismic events. the microseismic events locations in the form of two vertical fractures. The blue triangles
indicate the location of the geophone array.
The uncertainty in microseismic event locations can arise from
multiple factors, including operational errors, inaccuracies in the microseismic event locations. The relocation iterations are repeated
velocity model, presence of ambient noise in the data, suboptimal until the normalized movement of each microseismic event con-
acquisition configurations, and inconsistencies in the arrival-time forms to a chi-square distribution with three degrees of freedom,
picks of P- and S-wave signals (Warpinski, 2009). indicating an optimal fit (Jones and Stewart, 1997).
We present the collapsing method as an intermediate analysis
step preceding the interpretation of the microseismic cloud shape, Synthetic application
aiming to mitigate the potential of misinterpretation of the caus- We compare the performance of the waveform-based relative
ative structures. This is particularly valuable when dealing with location and the Geiger techniques when applied to synthetic
microseismic clouds that exhibit diffused or scattered behavior, microseismic data. The Geiger method employs an iterative
which can be incorrectly attributed to a complex source system approach to determine the absolute source location of an earth-
(Cipolla et al., 2011). quake (Geiger, 1912). Full elastic synthetic seismograms are
The process of the collapsing method follows an iterative generated based on a downhole acquisition geometry. A
workflow (Figure 2), which starts with the selection of a target 3C 12-geophone array is deployed along a lateral monitoring well
event for relocation. Then, the target event is moved toward the to capture the seismic response from a lateral treatment well
center of gravity of a group of events that fall within the boundary drilled 396 m away. A total of 461 evenly distributed microseismic
of its uncertainty ellipsoid. This group includes the target event events are modeled as a pair of parallel vertical fractures, positioned
itself. The relocation process is applied iteratively to the remaining with a separation distance of 18.3 m. Figure 3 illustrates the
microseismic events, forming a single iteration of the remapped source-receiver geometry.

10 The Leading Edge January 2024 Special Section: Microseismic monitoring

 For each microseismic event, a shear-tensile dislocation relative relocation technique. Both methods can generate event
moment tensor is used in conjunction with a triangular source-time locations in close proximity to the actual locations. Events
function. Each synthetic microseismic recording is modeled as a located by the relative approach exhibit a better lateral contain-
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

16 s trace and sampled at 1000 Hz. A controlled level of noise is ment compared to those located using the Geiger method. Both
added to the waveforms, where the magnitude of the noise is a location results exhibit a degree of sparsity in the vertical
function of the largest S-wave radiation at the source sphere, so direction. Figure 7 represents a histogram depicting the distribu-
different components at different locations are expected to have tion of location errors obtained from both methods in the lateral
different S/N due to radiation pattern, geometrical spreading, and vertical directions. While comparable errors are observed
and wavefield particle motion. (Figure 4). in the vertical direction, the relative relocation method dem-
We first ran the Geiger location algorithm on all the micro- onstrates a smaller error, approximately three times smaller
seismic events to obtain initial locations. Figure 5 illustrates than that of the Geiger method.
the results of these locations denoted by blue crosses. Figure 6 The collapsing and remapping approach is also applied to the
represents the microseismic hypocenters located using the microseismic location events obtained with the Geiger method.
The results of the relocation process are shown in Figure 8. After
10 iterations, it can be appreciated how the collapsed microseismic
events provide a more accurate description of the fracture geometry
relative to the original hypocenter locations.

Field data applications

Case 1. The first case is related to employing the collapsing
and remapping technique to event hypocenters determined
through an absolute location scheme. Figure 9 illustrates the
collapsing results obtained from two different microseismic trials
DOI:10.1190/leedff.2024.43.issue-1

conducted to evaluate hydraulic fracture propagation and geom-

Figure 4. Microseismic event waveforms modeled at 12 3C receivers.
etry in an unconventional reservoir. The results show the outcomes

Figure 5. Microseismic location results using the Geiger approach. (a) Map view and cross-
section views from the (b) y and (c) x axes of the acquisition. The red circles represent the true
source locations. The blue crosses indicate the locations determined by the Geiger method.

Figure 7. Histograms illustrating location error distributions produced by the Geiger method
and the relative relocation method in (a) and (b) the horizontal direction and (c) and (d) the
vertical direction.

Figure 6. Microseismic location results using the relative relocation approach. (a) Map Figure 8. Comparison between the synthetic test hypocenter locations from the Geiger
view and cross-section views from the (b) y and (c) x axes of the acquisition. The red circles approach (original) and the application of the postprocessing collapsing technique (collapsed).
represent the true source locations. The blue crosses indicate the locations determined by The green crosses indicate the true source locations. The geometrical description of the
the waveform-based relative location. fracture is enhanced by the postprocessing application of the collapsing method.

Special Section: Microseismic monitoring January 2024 The Leading Edge 11

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 10. Hydraulic fracturing monitoring program where the stimulated wells were
monitored simultaneously by a wireline array of 3C geophones in the center between the
two stimulated wells, and by two monitoring wells instrumented with FO located on the
outer flanks: (a) microseismic hypocenter locations from one stage obtained from the
processing of the 3C geophone array data alone, (b) microseismic locations obtained by
using the DAS data alone from two DAS cables.

Table 1. Field-detected microseismic events from DAS cables and geophone array.
Figure 9. Collapsing results obtained from two different microseismic trials conducted to
evaluate hydraulic fracture propagation: (a) and (c) two different areas with microseismic Fiber-optic Fiber-optic Colocated Total Geophone
DOI:10.1190/leedff.2024.43.issue-1

locations obtained with a standard method; (b) and (d) locations after the application of the Well C 452 40 40 492 3646
collapsing method.
Well A 239 290 239 529 2625
of collapsing, indicating that the microseismic stages demonstrate Total 691 330 279 1021 6271
a confined behavior to suggest possible planar zones of weaknesses,
i.e., lineation.
Case 2. The second field example illustrates a case in which a and at three times the distance (farther FO) for each case. The
hybrid acquisition of geophone arrays and FO cables was performed DAS cable, installed as wireline and coupled to the well casing
in multilateral wells. Two treatment wells were drilled into a by gravity, extends the entire length of the well including the
different stacked landing zone than the monitoring wells, with vertical and horizontal sections. Each fracking stage is simultane-
the monitoring wells located in a lower stratigraphic layer below ously monitored by the geophone array and by the two DAS cables
the target zone. The horizontal sections of the treatment and on the opposite sides of the treatment well to provide optimal
monitor wells are parallel to each other. A total of 87 stages were spatial and azimuthal coverage. DAS acquisition for microseismic
stimulated using slickwater fracturing on both wells with almost monitoring used a gauge length of 10 m where the interrogator
symmetric frac design (well spacing, number of clusters, spacing sampling rate was set at 8 kHz.
of clusters and stages, as well as the stimulated volumes), except Examples of joint microseismic recordings are shown in
for a couple of stages where the number of clusters and frac volume Figure 11. The plot shows the recording of a good quality
were increased. Key objectives for each acquisition are to investigate microseismic event (MW = –0.22) detected by both the geophone
the number, magnitude, and type of microseismic events activated array and the two DAS cables (FO) at different distances. It
during each frac stage. Some of the qualitative parameters analyzed can be observed that the geophone array provides the full
are the microseismic cloud shape or envelop, the fracture geometry, wavefield recording on the three components to enable the
the off-stage clustering, the event cloud overlaps and the event wavefield rotation and the characterization of the radiation
density. Important enhancements of the acquisition technology pattern from the source. The S/N is high, while the aperture/
are provided by the utilization of FO in the form of DAS, DSS, azimuthal coverage is small, such that event locations must
and DTS as separated deployed cables, where the acquisition utilize the azimuthal/propagation angle information as per the
geometry is displayed in Figure 10. Geiger method if no other “migration” type methods are utilized
In the recent hydraulic fracturing monitoring program, the (Gajewski et al., 2007). On the DAS side, the wavefield is
stimulated wells were monitored simultaneously by a wireline continuously sampled over kilometers including the heel section
array of 3C geophones deployed in the observation well in the of the monitoring wells, hence with enormous aperture, while
center (i.e., in between the two stimulated wells) while two no wavefield polarization information can be obtained. The
monitoring wells instrumented with FO cables were located on data sets are complementary. The joint analysis of the data is
the outer flanks. The two FO wells were active during the treat- expected to provide high-resolution joint hypocenter locations
ment of each well such that the DAS data are respectively at equal as well as better characterization of the velocity model and of
distance as the geophone array to the stimulated well (closer FO) the source characteristics.

12 The Leading Edge January 2024 Special Section: Microseismic monitoring

 Event detection performed independently on the geophone detected and subsequently located. The analysis of joint geophone-
and DAS cables has indicated a much larger sensitivity of the fiber data is ongoing, and it is expected to enhance both detectability
geophone array relative to the closer DAS, where the monitoring and location accuracy of small-magnitude events by application of
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

wells are located at the same distance to the stimulated well the described techniques. The availability of simultaneous acquisi-
(Table 1). The DAS cables have independently detected up to 16% tion of microseismic data from both 3C geophones and DAS may
of the microseismic events identified by the geophone array. While enable additional joint analysis, for example, the development of
these detections can reproduce a good number of hypocenter dedicated denoising techniques with geophone recordings used as
locations by simultaneously utilizing the data from both the DAS templates, as well as the development of velocity of displacement-
cables, several small-magnitude microseismic events cannot be strain relations for magnitude and moment tensor evaluations.
The simultaneous acquisition of
temperature (DTS) and slow strain
DAS data (DSS) along the FO wells
provides additional information relative
to the propagation of fractures
approaching and/or intersecting the
monitoring wells. Strain data are char-
acterized by different phases of positive
strain (i.e., tensile/fiber extension) as
the fracture approaches the cable and
intersects it during fracture growth
(Figure 12b), and negative strain (i.e.,
compression/fiber contraction) after
pumps shut down because of fracture
DOI:10.1190/leedff.2024.43.issue-1

closing and fluid leak off (Richter et al.,

2019). The slow strain DAS data further
illustrate an uneven distribution of
strain episodes along the horizontal
section of the well (MD) as well as
repeated reactivation episodes during
different stimulation stages. Such
Figure 11. Records of a good S/N microseismic event (MW = –0.22) from the two DAS cables and from the geophone array. behavior of an uneven distribution and

Figure 12: Compilation of results from FO monitoring: (a) hypocenter locations using the DAS data alone (two DAS recordings) with the observed cumulative strain data, (b) slow strain
DAS data (DSS) from FO-2 showing an uneven distribution of strain episodes along the horizontal section of the well (MD) as well as the repeated reactivation episodes during different
stimulation stages, (c) temperature response (DTS) from FO-2 showing a strong correlation to the strain data (arrow). Lines in (a) indicate the stimulation stage with the corresponding
strain episodes observed on the DSS.

Special Section: Microseismic monitoring January 2024 The Leading Edge 13

 reactivation is indicative of the possible presence of natural fracture 3C geophone arrays could be further utilized for calibrating
systems. A clear temperature response is also observed (Figure 12c) advanced analysis of DAS data.
showing a good correlation with the strain results (arrow). The The ongoing interpretation work is developing along the
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

compilation of the microseismic hypocenter locations (obtained discussed objectives of enhancing the microseismic hypocenter
using the DAS data alone — two cables) and of the cumulative locations by utilizing mirrored observations from geophone and
strain data is shown in Figure 12a. Some correlation is inferred DAS arrays. In doing so, we utilize as much as possible the
between the microseismic event alignment, and the cumulative advanced location approaches of pattern matching/relative loca-
strain recorded on the DAS cables. Lines (i.e., vectors) are used tions, as well as collapsing technology. The integration of the
to connect the stimulation stages with the corresponding strain seismological data with strain and temperature measurements
events observed on the DSS. The variable orientation of the vectors from FO is expected to enhance the interpretation capability
supports the hypothesis of existing fracture reactivation but does while bridging the gap between seismological studies, engineering
not necessarily provide the information of the fracture network applications, and geomechanical analysis. The acquired data sets
geometry. It should also be noted that for existing fracture networks further provide an avenue for research related to moment tensor
the extensional strain observed on the fiber may occur aseismically, analysis and possible template-based denoising applications via
i.e., without causing a shear failure marked by a microseismic deterministic or machine learning approaches. The integration of
event. High accuracy hypocenter locations and possible determina- geophone and FO data is expected to reduce the uncertainty in
tions of the source mechanism/moment tensor would enable a terms of frac geometry, well spacing, and stacked landing as well
better integration of microseismic with the dynamic strain data. as providing optimized stimulation designs.
The calculation of moment tensor from the combination of the
two DAS cables and of 3C observations from the geophone array Conclusion
is part of an ongoing study. Microseismic monitoring of hydraulic fractures is a valuable
practice that can provide important data to describe the distribution
Discussion of the stimulated rock volume for the development of unconven-
DOI:10.1190/leedff.2024.43.issue-1

The current main utilization of the microseismic tool by tional reservoirs. Current practices focus on the geometrical
reservoir engineers is related to the geometrical characterization description of the stimulated or generated fractures as inferred
of fractures that is obtained by applying some empirical rules from the microseismic hypocenter locations. Obtaining accurate
(e.g., half length, fracture height) to the distribution of located hypocenter locations is therefore of primary importance to reservoir
hypocenters. The predicted stimulated rock volume is used to engineers for improving operations. Technologies addressing
guide the development of the unconventional reservoirs in terms advanced location techniques while estimating the corresponding
of multilateral well spacing for optimal permeability and drainage uncertainties may represent one of the most impactful outcomes
enhancement. Such utilization of microseismic data is focused for the near future. Multiwell monitoring scenarios, such as the
on reliable hypocenter locations, for which we may expect that one described with horizontal well acquisition with geophones
significant errors may be introduced by inaccurate phase identifica- and FO, provide the necessary aperture and azimuthal coverage
tions as well as by inaccurate velocity estimations. Enhancing the to enable robust hypocenter locations. The acquisition setup also
reliability of microseismic hypocenter locations is, therefore, one provides an avenue for advanced studies and research, for example
of the first priorities to be addressed in the development of new for the study of the source moment tensor. FO is a rich source of
technology. In particular, the combined utilization of relative data complementing the highly sensitive geophone arrays. This
location methods based on waveform pattern match followed by may enable more accurate analyses of the microseismic phenomena
collapsing techniques can provide a high accuracy and weighted once more research is developed in this direction. Hybrid acquisi-
description of microseismic hypocenter locations for enhanced tions involving geophones and FO, like the one presented, provide
geometrical fracture characterization. The identified multiplets vast amounts of data from DAS as well as from additional proper-
of microseismic events can be further utilized to generate com- ties such as strain and temperature. The seamless integration of
posite high S/N templates through stacking. Such templates FO data with seismological data will help bridge the gap between
characterizing specific radiation patterns can be further utilized seismologists and reservoir engineers for the evaluation of uncon-
as a diagnostic tool for the source mechanism or moment tensor ventional reservoirs. Such studies provide an avenue for a better
where recent developments have shown the potential of utilizing understanding of the microseismic phenomena for a multitude of
machine learning techniques to overcome the limitations involved utilizations from seismological studies to applications for sustain-
in the deterministic inversion for the moment tensor (Mascarell ability objectives such as geothermal, carbon capture and storage,
et al., 2021). The microseismic moment tensor evaluation represents and gas storage monitoring.
a natural evolution for the fracture characterization objective, a
task that is still elusive and difficult to achieve for acquisition Data and materials availability
technology utilizing a single observation well. Multiwell monitor- Data associated with this research are confidential and cannot
ing such as the described hybrid geophone-DAS acquisition may be released.
prove to represent an ideal scenario for extending the limited
aperture of 3C geophones arrays. The full wavefield records from Corresponding author: daniele.colombo@aramco.com

14 The Leading Edge January 2024 Special Section: Microseismic monitoring

 References
Arrowsmith, S. J., and L. Eisner, 2006, A technique for identifying microseismic multiplets and
application to the Valhall field, North Sea: Geophysics, 71, no. 2, V31–V40, https://doi.
org/10.1190/1.2187804.
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Caffagni, E., D. W. Eaton, J. P. Jones, and M. van der Baan, 2016, Detection and analysis of
microseismic events using a matched filtering algorithm (MFA): Geophysical Journal
International, 206, no. 1, 644–658, https://doi.org/10.1093/gji/ggw168.
Cipolla, C., S. Maxwell, M. Mack, and R. Downie, 2011, A practical guide to interpreting
microseismic measurements: North American Unconventional Gas Conference and
Exhibition, SPE-144067, https://doi.org/10.2118/144067-MS.
Gajewski, D., D. Anikiev, B. Kashtan, E. Tessmer, and C. Vanelle, 2007, Localization of
seismic events by diffraction stacking: 77th Annual International Meeting, SEG, Expanded
Abstracts, 1287–1291, https://doi.org/10.1190/1.2792738.
Geiger, L., 1912, Probability method for the determination of earthquake epicenters from the
arrival time only: Bulletin of Saint Louis University, 8, 60–76.
Geller, R. J., and C. S. Mueller, 1980, Four similar earthquakes in central California: Geophysical
Research Letters, 7, no. 10, 821–824, https://doi.org/10.1029/GL007i010p00821.
Gibbons, S. J., and F. Ringdal, 2006, The detection of low magnitude seismic events using
array-based waveform correlation: Geophysical Journal International, 165, no. 1, 149–166, ®
https://doi.org/10.1111/j.1365-246X.2006.02865.x.
Jones, R. H., and R. C. Stewart, 1997, A method for determining significant structures in a
cloud of earthquakes: Journal of Geophysical Research. Solid Earth, 102, no. B4, 8245–8254,
https://doi.org/10.1029/96JB03739.
Luo, B., G. Jin, and F. Stanek, 2021, Near-field strain in distributed acoustic sensing-based
microseismic observations: Geophysics, 86, no. 5, P49–P60, https://doi.org/10.1190/
geo2021-0031.1.
DOI:10.1190/leedff.2024.43.issue-1

Mascarell, M. C., D. Rovetta, D. Colombo, and E. Turkoglu, 2021, Moment tensor prediction
from borehole microseismic data with machine learning: 82nd Annual Conference and
Exhibition, EAGE, Extended Abstracts, https://doi.org/10.3997/2214-4609.202112513.
Meng, X., H. Chen, F. Niu, Y. Tang, C. Yin, and F. Wu, 2018, Microseismic monitoring of
stimulating shale gas reservoir in SW China: 1. An improved matching and locating
technique for downhole monitoring: Journal of Geophysical Research. Solid Earth, 123,
no. 2, 1643–1658, https://doi.org/10.1002/2017JB014488.
Reyes-Montes, J. M., W. S. Pettitt, B. Hemmings, J. R. Haycox, J. R. Andrews, and R. P. Young,
2009, Application of relative location techniques to induced microseismicity from hydraulic
fracturing: SPE Annual Technical Conference and Exhibition, SPE-124620, https://doi.
org/10.2118/124620-MS.
Richter, P., T. Parker, C. Woerpel, Y. Wu, R. Rufino, and M. Farhadiroushan, 2019, Hydraulic
fracture monitoring and optimization in unconventional completions using a high-resolution
engineered fiber-optic distributed acoustic sensor: First Break, 37, no. 4, 63–68, https://
doi.org/10.3997/1365-2397.n0021.
Slunga, R., S. T. Rögnvaldsson, and R. Bödvarsson, 1995, Absolute and relative locations of
similar events with application to microearthquakes in southern Iceland: Geophysical
Journal International, 123, no. 2, 409–419, https://doi.org/10.1111/j.1365-246X.1995.
tb06862.x.
Wang, H., and T. Alkhalifah, 2018, Microseismic imaging using a source function independent
full waveform inversion method: Geophysical Journal International, 214, no. 1, 46–57,
https://doi.org/10.1093/gji/ggy121.
Warpinski, N., 2009, Microseismic monitoring: Inside and out: Journal of Petroleum
Technology, 61, no. 11, 80–85, https://doi.org/10.2118/118537-JPT.

INSIGHT

Special Section: Microseismic monitoring January 2024 The Leading Edge 15

 Detection of microseismic events in continuous DAS data
using convolutional neural networks
N. Boitz1 and S. Shapiro1
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

https://doi.org/10.1190/tle43010016.1

Abstract One important application for arrival-time picking was proposed

Detection and localization of microseismicity is an inevitable by Zhu and Beroza (2018), who trained a convolutional neural
task in monitoring fluid injections into subsurface rocks during network (CNN) called PhaseNet by using tens of thousands of
hydraulic stimulations. Traditionally, downhole geophones placed waveforms and arrival times handpicked by analysts. Zhu and
into boreholes or large surface geophone networks have been used Beroza (2018) adapted the U-Net architecture, proposed by
for this, but in recent years, the use of distributed acoustic sensing Ronneberger et al. (2015) for biomedical image segmentation,
(DAS) via fiber-optic cables placed into boreholes has become a and changed the layer dimensions accordingly to their desired
common technique. However, DAS registrations still have lower inputs and outputs. PhaseNet was tested on several cases of tectonic
signal-to-noise ratios than geophones; i.e., they cannot detect seismicity and even on some induced seismicity case studies (e.g.,
small-magnitude events. In this work, we develop and train a Wang et al., 2020), providing reasonable results.
convolutional neural network capable of detecting microseismic All of the aforementioned methods explicitly (Ross and Ben-
events in continuous DAS recordings incorporating arrival-time Zion, 2014) or inherently (Zhu and Beroza, 2018) use polarization
information from geophones. The network is trained on DAS and information to distinguish P from S arrivals and thus only work
geophone data from the Utah FORGE enhanced geothermal for three-component microseismic data. In contrast, microseismic
system project for which we are able to significantly shift the data acquired by distributed acoustic sensing (DAS) are only sensitive
DOI:10.1190/leedff.2024.43.issue-1

detection threshold toward smaller magnitude events. Although to deformations along the cable. A fiber in a vertical borehole is
the number of microseismic events (approximately 150) used for only sensitive to vertical deformations, and each cable segment
training is small, the tested network performance is high and therefore can be regarded as a one-component vertical receiver.
provides a complete event catalog down to magnitude MW = –1.6, Furthermore, DAS recordings are measured in terms of strain or
a notable improvement over previous studies. Using a short record- strain rate and not displacement or velocity as geophones are.
ing period of several hours for training, such a network might be Thus, apart from the numerous advantages of DAS, the
used for long-term, real-time monitoring of geothermal sites. approaches of Ross and Ben-Zion (2014) and Zhu and Beroza
Although the network is explicitly trained for the geometry of (2018) are not applicable to pick seismic arrivals in DAS data.
the data set used, the philosophy and network architecture can During DAS acquisition, optic signal from the cable is averaged
be adapted for similar case studies where long-term seismic moni- over the so-called gauge length, which is usually in the range of
toring is required (e.g., CO2 sequestration). 10 m in microseismic studies. This gauge interval is then typically
shifted by 1 m to obtain the next DAS trace; i.e., neighbored
Introduction traces share 90% of data and therefore should have a high waveform
Seismic event detection in continuous (geophone) recordings similarity. This characteristic is used by Porras et al. (2022), who
is usually the first step in analyzing tectonic processes or, in the proposed semblance-based detection methods that provided
case of microseismic monitoring, the success of reservoirs stimula- reasonable results for visible events.
tions using seismicity. For injection-induced seismicity, in which events are expected
In recent decades, several approaches for automatic detection to occur close to the injection point, moveouts for individual
of seismicity and P-wave arrival picking have been developed that events should be very similar. This characteristic was used by
often provide reasonable results, such as the short-term average Tegtow et al. (2023), who tested stacking-based approaches using
to long-term average ratio (Allen, 1978) or the envelope function moveout-corrected data sections of DAS to detect events.
(Baer and Kradolfer, 1987). However, these methods often reach Although all these methods work reasonably well, they have
their limits when it comes to S waves because S-wave arrivals are in common that they only work perfectly down to certain event
often within the P-wave coda. A method to distinguish P- from magnitudes. Signatures of these events are mostly visible by eye
S-wave arrivals using polarization analysis has been proposed by in the data, which means that the methods are at best as good as
Ross and Ben-Zion (2014). However, the accuracy of these methods an analyst. Thus, besides the amazing data density of DAS registra-
is below that of experienced analysts. tions, we have to deal with the lower data quality and the larger
The high amount of available data has increased the interest data amount.
in and importance of machine learning algorithms in the micro- To overcome these problems, we propose a machine learning-
seismic community for various applications (Anikiev et al., 2023). based method that can detect and pick visible and invisible arrivals

Manuscript received 31 August 2023; accepted 23 October 2023.

1
Freie Universität Berlin, Berlin, Germany. E-mail: boitz@geophysik.fu-berlin.de; shapiro@geophysik.fu-berlin.de.

16 The Leading Edge January 2024 Special Section: Microseismic monitoring

 reliably detect events down to magni-
tudes of MW = –1.6 without suffering
from too many false detections. Such
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

small events are invisible in the data and

cannot be reliably detected by stacking-
based algorithms.
Event detection with the network is
quick (processing 1 minute of data takes
only several seconds), and thus it can be
used for real-time monitoring. Another
Figure 1. Geometry (a) map view and (b) side view of the Utah FORGE 2019 stimulation. The green line is the trajectory of remarkable feature of the network is that
the injection borehole (58A-32). The blue line is the trajectory of the monitoring borehole (78A-32) equipped with DAS and it seems sufficient to train the network
12 3C geophones (red dots). The red crosses are locations of microseismic events provided by the geophone contractor. on a small number of events (about 150).
(c) Histogram of earthquake magnitudes observed in the three hours of stimulation.
Having a fiber as a permanent, long-
lasting observation tool, the network
can be used for long-term monitoring
of geothermal stimulations.

Data
In this work, we use continuous
microseismic data registered at the Utah
FORGE EGS project (Moore et al.,
2019). The analyzed data comprise three
DOI:10.1190/leedff.2024.43.issue-1

hours of DAS and geophone recordings

registered during two stimulation cycles
on 27 April 2019. For the 2019 stimula-
tion tests, two deep wells were drilled
— one for fluid injection and one for
downhole monitoring using geophones
and DAS (see map and side view in
Figure 1). The entire monitoring bore-
hole down to a depth of 1 km is equipped
with a fiber; geophones are only installed
in the lower part.
Both types of data are acquired in
the same borehole and synchronized in
time; i.e., microseismic arrivals have the
same arrival times with an accuracy in
the frame of milliseconds. An example
of a single microseismic event recorded
by geophones and DAS is shown in
Figures 2a and 2b.
For the corresponding time period,
Figure 2. (a) Geophone recordings (vertical component) of strong microseimic event (MW = –0.7). Red and blue squares geophone (Pankow, 2019) and DAS
indicate P and S arrivals. (b) DAS recordings of the same event as in panel (a) superimposed by the arrival-time picks from the
geophones. Because both types of data are acquired in the same borehole, arrival times coincide accurately. (c) A smaller- (Moore et al., 2019) event catalogs are
magnitude event clearly visible and pickable in the geophone recordings but invisible in (d) the corresponding DAS section. available that consist of 187 and 30
events, respectively. This large difference
in DAS data. We construct a CNN of a U-Net-like style can be explained by the higher signal-to-noise ratio (S/N) of
(Ronneberger et al., 2015) that is trained on DAS data from the geophone recordings. Consecutive studies (Lellouch et al., 2020;
Utah FORGE enhanced geothermal system (EGS) project (Moore Tegtow et al., 2023) were able to increase the number of events
et al., 2019) and is able to distinguish P arrivals and S arrivals detected by DAS using different stacking approaches and spectral
from noise. analysis; however, catalogs are only complete down to magnitude
The main difference between our approach and existing MW > –1.4.
approaches is that we also incorporate arrival-time information For illustration, Figures 2c and 2d show a microseismic event
from geophones that are installed in the same monitoring borehole of magnitude MW = –1.5 that is clearly identifiable in the geophone
as the fiber for training the network. This enables the network to recordings but invisible in the DAS recording. In the following

Special Section: Microseismic monitoring January 2024 The Leading Edge 17

 processing, we take advantage of the
fact that DAS and geophone recordings
are synchronized in time. This means
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

that we know where to expect arrivals

in the DAS data, although we might
not see them by eye and cannot enhance
them by waveform stacking. This knowl-
edge is used to train a CNN.

Method
Architecture of the CNN. The net-
work architecture is inspired by the
U-Net (Ronneberger et al., 2015) and
PhaseNet (Zhu and Beroza, 2018)
architectures that both contain a con-
traction phase with several consecutive
convolutions, stride and rectified linear
unit (ReLu) operations, and an expan-
sion phase with several deconvolutions.
The final model output is computed
using the softmax function. The input
into the model is a DAS “picture”
(5000 × 400) that consists of 5000 time
DOI:10.1190/leedff.2024.43.issue-1

samples registered at the 400 lowermost

channels of the DAS cable. The output
of the model has the size (5000 × 3) that
corresponds to the probability of P, S,
and noise arrivals at the bottom of the
cable for each of the 5000 time samples.
The detailed structure of the CNN is
shown in Figure 3a.
The major difference between the
U-Net/PhaseNet architecture and the Figure 3. (a) Visualization of the network architecture and (b) illustration of the DAS waveform labeling process. The input into
CNN proposed in this paper are the the network is a DAS recording of 400 traces with 5000 time samples (upper part of [b]). The model output is the probability of
dimensions of input and output. U-Net P, S, and noise arrival for each time sample (lower part of [b]).
and PhaseNet have equally sized input
and output dimensions. For the CNN proposed here, the second By combining the picking information from the geophone data
dimension of the output is significantly smaller. We obtain this and the magnitudes from the catalog, we accurately know when and
dimension reduction by contracting the data along both the time how large arrivals are expected to be within the DAS data. Because
and channel dimensions during the contraction phase. This is geophone and DAS data are synchronized in time, we can transfer
done to obtain the desired output dimensions and to keep the the knowledge from the the geophone data with a higher S/N to
model as small and simple as possible. In contrast, U-Net contracts DAS data. Practically, we can assign P and S arrivals to DAS data,
the data along the time axis while increasing data complexity where arrivals can be seen by eye (e.g., Figures 2a and 2b) as well as
(along the second data axis) during the contraction phase. for events where arrivals are invisible in DAS (e.g., Figure 2d).
Data labeling for the CNN. For training the CNN, or respec- The procedure of the DAS data labeling process is shown in
tively computing the weights in each layer, the network needs to Figure 3b. In contrast to the geophone data, DAS data do not show
be provided with DAS pictures (5000 × 400) and the correct single P and S first arrivals but instead show several multiples of
corresponding labels (5000 × 3). To obtain microseismic arrivals, both waves. For the labeling process, we decided to label not only
we use the pretrained PhaseNet (Zhu and Beroza, 2018) to pick the first arrivals of P and S arrivals but also corresponding windows
phases in the continuous geophone data. If P and S arrivals at in which the multiples arrive at the DAS cable. In particular, we
more than six geophones are detected within a reasonable time assign a P-arrival probability of 100% to the interval between the
interval, an event is declared to be detected. The obtained arrival first P arrival and the first S arrival. Furthermore, we assign an
times are checked against the provided geophone event catalog S-arrival probability of 100% to an interval of 1.5 times the length
provided by the contractor. Because events in the magnitude of the P-S window after the first S pick. The rest of the section is
catalog only have a source time accuracy of 1 s, catalog times labeled as 100% noise. To obtain smoother results, the transitions
cannot be directly used for training. between the classes are modeled with sigmoid functions.

18 The Leading Edge January 2024 Special Section: Microseismic monitoring

 The labeling method was particularly designed for the FORGE domain of at least 150 time samples. For individual source mecha-
geometry, but it can be adapted easily for scenarios where geo- nisms, for instance, if one of the nodal planes of the source
phones and DAS are not installed in the same borehole. If, for mechanism is aligned with the fiber (which is not the case in this
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

instance, earthquake origin times and a velocity model are known, data set), this criterion might not work, and different detection
one can compute arrival times at the registering fiber and use criteria would have to be introduced.
these for network training. Additionally, we compare predicted labels and the ground
CNN output smoothing. The outputs of the CNN are probabili- truth. Here, four different scenarios are possible. In the case that
ties of P, S, and noise for each time sample of the DAS data, which the arrival domains of P and S waves in the predicted labels for
sum up to one. These probabilities show high-frequency temporal some time overlap with the ground truth, the event is called a
fluctuations, so we must define criteria to declare an arrival as true positive (TP). Because we also trained the network on noise,
detected. For this, we smooth the raw output of the CNN according it is possible that a whole data section only contains noise. If in
to the following procedure: we define a moving window of 100 such a case the network predicts only noise, we call this a true
samples length. If more than 50 samples (50%) have a probability negative (TN) example because the network correctly identifies
higher than 50%, we assign a value of 100% probability for the that no event is within the data. If the data section contains an
respective wave type to the central sample of the moving window. event (i.e., the given labels for P and S arrivals are nonzero) and
Then we shift the time window by 10 samples. the network predicts only noise for this time period, we call this
The continuous smoothed CNN output will be used later to a false negative (FN) example because the network is not able to
detect events. For comparison, raw CNN outputs as well as the detect the event although it should be in the data. The fourth
smoothed curves are shown by the shaded and solid curves, possibility is that the network predicts P followed by S arrivals
respectively, in Figures 4 and 5. for times when the ground truth probability for P and S arrivals
This smoothing procedure is an optional processing step that is zero; i.e., the data should not contain an event. Such a scenario
lowers the number of false detections in the low-magnitude range. is referred to as false positive (FP).
From continuous CNN output to event detection. After prob- These four classes are typically used to evaluate model per-
DOI:10.1190/leedff.2024.43.issue-1

ability smoothing, the final scope of this work is to detect micro- formances in machine learning. Final model performances are
seismic events in the DAS data using the smoothed probability usually quantified using precision (P) and recall (R) that are
curves. Thus, we must define criteria when an earthquake is defined as
declared to be detected. For this particular data set, we define an
TP
event as detected if the smoothed CNN output contains a P-wave ​P = _ _​_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ​_ (1)
arrival domain of at least 150 time samples, followed by an S-arrival TP + FP

Figure 4. Four data examples from the training data. Each of the subfigures contains the corresponding data section (input into the network), the labels for training (upper left), and the
labels predicted by the network after training. Panel (a) shows a large, visible event, (b) shows an invisible event, and (c) and (d) show different types of noise. After training, the network is
able to accurately predict microseismic arrivals in the training set and is insensitive to different types of noise.

Special Section: Microseismic monitoring January 2024 The Leading Edge 19

 and and training data. As the ground truth, we accept the labels assigned
during the procedure explained here. Subsequently, we train the
TP
​R = _
​ ​. (2) CNN on the training data. We evaluate its performance using the
TP + FN
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

training data as well on the previously unseen testing data.

Later, we will use these classes and metrics to evaluate the Results
model performance. Training data. After training, the CNN is able to reproduce
Split into training and testing data. For training a neural the arrival characteristics of all 149 training events of arbitrary
network, the network is provided with some training data and magnitude. Figure 4 shows four different DAS data sections,
the respective labels assigned to the data (the so-called ground the respective labels for training (upper left), and the predictions
truth). To evaluate network performance, the network is later of the network (lower left). Figure 4a shows an example of a
provided with testing data that it has not seen before during large-magnitude event. Given and predicted labels perfectly
training. The network predictions then can be compared to the coincide. The same is true for the small-magnitude event
known ground truth for the testing data. (MW = –1.7) in Figure 4b. Here, the network was able to learn
Usually, a data set is split into 80% training data and 20% features that are invisible to humans but that apparently exist in
testing data. This split is usually done randomly. Because the data the data. It is also of particular importance to train the network
set consists of 840 DAS pictures that contain only 187 events, on different types of noise, which the network can identify
most of the pictures only contain noise. A random split might perfectly in the training data. Typical noise patterns for this data
provide almost pure noise examples in the testing data, which set are short, simultaneous energy peaks on all channels (Figure 4c)
would make the evaluation of the model performance difficult. and pumping noise from the surface that is especially visible at
Furthermore, we are interested in how the model performs on the upper part of the cable (Figure 4d).
events with different magnitudes. Thus, we use the following Testing data. The test set contains 34 events of which 26 have
methodology for test/train split: Events are grouped by magnitude, magnitudes above MW = –1.65, which we set as the desired
DOI:10.1190/leedff.2024.43.issue-1

and from these groups randomly 80% of the events are selected detection threshold. Figure 5 illustrates the detection of four
for training, 20% for testing. For instance, from the 15 events in different microseismic events of different magnitudes in the testing
the interval between M W = –1.3 and M W = –1.4, we use 12 for data. Because all examples are from the testing data, the ground
training and three for testing. In this way, magnitudes in both truth (upper left part of each subfigure in Figure 5) is unknown
data sets are correspondingly balanced to the 80/20 split. to the network, but we can use it for comparison with the CNN
Additionally, we add sections that contain only noise in testing predictions. Figure 5a shows the largest event of magnitude

Figure 5. Four data examples from the testing data. Each of the subfigures contains the corresponding data section (input into the network), the labels predicted by the network, and the
ground truth, which is unknown to the network (upper left). Panel (a) shows a large, visible event, (b) shows an intermediate magnitude event, (c) shows a small, barely visible event, and
(d) shows an invisible event. All events are detected by both P and S arrivals.

20 The Leading Edge January 2024 Special Section: Microseismic monitoring

 MW = –0.7 in the testing data set. It shows that both P- and were not detected by the algorithm (FN ), and two additional
S-wave domains are accurately determined by the network. In events were detected by the network, although the geophone
particular, this example shows that the network learned correct data do not contain an event (FP).
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

features. It provides a significantly longer S-wave arrival domain A further investigation of the two FN events revealed that
than the ground truth, which was assigned in a quite naive way these are invisible to the eye and have magnitudes very close to
(see definition earlier). A visual inspection shows that at later the desired detection threshold of MW = –1.65. A closer look at
times (approximately sample 1750), multiples of S waves are seen the two FP examples shows no visible arrivals in the data section,
in the data. The origin of these multiples is not entirely clear, as which either means that an event indeed occurred that was even
they are not observed in the geophone recordings. too small to be detected by the geophone data, or more likely that
The second example shows a microseismic event with a medium the network fails for these data sections. The 24 detected events
magnitude of MW = –1.2 (Figure 5b). In contrast to other events, contain all events down to magnitude MW = –1.63 within the
only a single P- and a few S-wave arrivals are visible in the data. testing data, which is a significant improvement over previous
The network accurately predicts onset times for both wave types studies. Below that magnitude, the network is not able to reliably
and a slightly shorter S-wave arrival domain compared to the detect events.
given labels. This might be caused by the few S-wave multiples Both precision and recall are estimated to be 92%, which is
and is again an indication that the network learned reasonable relatively high (typical values between 85% and 95%) and indicates
features of S waves. that the network also performs well on previously unseen data.
The third example event (Figure 5c) of magnitude MW = –1.54
is on the edge of what is detectable by the eye. Only the S-wave Conclusions and outlook
arrival can be clearly identified; P-wave arrivals might be visible In this work, we presented an approach for arrival-time
between channels 200 and 300 at times between samples 3700 and event detection in continuous DAS data. We trained a
and 3800. The network accurately picks the S-wave onset, but it CNN using DAS data and arrival picks from geophone data.
is not able to accurately determine the first P onset (which is also Because the S/N of the geophone data is higher, the network
DOI:10.1190/leedff.2024.43.issue-1

invisible to the human eye) for this event. Apparently, the network could also be trained on events invisible in DAS. We applied
here picks some of the P multiples at sample 3700 as a first arrival. this approach to three hours of DAS data from the Utah
Nonetheless, the network correctly identifies features of P and FORGE EGS project.
S waves in the data. After training, the network is able to reliably detect events in
The fourth example (Figure 5d) shows the smallest event previously unseen data down to magnitudes MW = –1.6 that are
in the testing data set (MW = –1.64). Here, no microseismic often invisible in the data to the human eye. The presented method
event is visible by eye in the data. Also stacking methods fail is quick and therefore can be used for real-time monitoring.
to detect this event. Nevertheless, the CNN located both P- and Training is performed on a small data set. Already after approxi-
S-arrival domains at more or less correct times, as revealed by mately 150 microseismic events, the CNN learned enough to
the ground truth. reliably detect events. Although, the network is only trained for
By also training the CNN on comparable small events, it the FORGE geometry, the network architecture can be easily
apparently learned features that are invisible to the human eye. adapted for similar case studies. For instance, one could install
This example shows the great potential of the network and the geophones in a borehole temporarily for a few hours of network
approach of training the network on such features that are invisible training, retrieve them afterward, and only use the fiber for
to the huyman eye. long-term monitoring. Thereby, one would benefit from the lower
Statistical evaluation of test events. The event classification cost of DAS cables and its better resistance against borehole
(see definitions noted earlier) for the testing data is shown in conditions (e.g., high temperatures).
Table 1. Because only a few events are in the data, the majority Currently, we are testing the same network on DAS data
of data sections only contain noise. Most of this noise was acquired at FORGE in 2022 in the same well and an additional
correctly classified by the network, resulting in a majority of well to analyze the stability of the CNN on varying noise condi-
true negative classifications (approximately 80%). Twenty-four tions and different P and S moveouts. First results are promising.
out of 26 events with magnitudes greater than MW = –1.65 are Furthermore, we are expanding the network to pick several separate
found by the algorithm and labeled as true positive. Two events arrivals along the cable, which will permit, in the case of multiple
wells, to also locate events.
Table 1. Evaluation of CNN performance on the test data. Most data examples were correctly identified as noise. Twenty-four
events were found correctly. Acknowledgments
We thank the sponsors of the
Class Explanation # of examples PHASE-AC consortium at Freie
True negative (TN) Data contain no event — network predicts no event 171 Universität Berlin for supporting the
True positive (TP) Data contain an event — network predicts an event 24 research presented in this paper. We
False positive (FP) Data contain no event — network predicts an event 2
also thank an anonymous reviewer for
comments that improved the original
False negative (FN) Data contain an event — network predicts no event 2
manuscript.

Special Section: Microseismic monitoring January 2024 The Leading Edge 21

 Data and materials availability Pankow, K., 2019, Utah FORGE seismicity associated with the 2019
Both geophone and DAS data and corresponding event cata- well 58-32 stimulation: U.S. Department of Energy, https://doi.
logs used in this study are freely available and can be obtained org/10.15121/1872101.
from the geothermal data repository (https://gdr.openei.org). Porras, J., F. Grigoli, E. Stucchi, K. Tuinstra, A. Tognarelli, F. Lanza,
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

M. Aleardi, A. Mazzotti, and S. Wiemer, 2022, A semblance based

microseismic event detector for DAS data: 24th General Assembly,
Corresponding author: boitz@geophysik.fu-berlin.de EGU, https://doi.org/10.5194/egusphere-egu22-3160.
Ronneberger, O., P. Fischer, and T. Brox, 2015, U-Net: Convolutional
References networks for biomedical image segmentation, in N. Navab, J. Hornegger,
Allen, R. V., 1978, Automatic earthquake recognition and timing from W. M. Wells, and A. F. Frangi, eds., Medical Image Computing and
single traces: Bulletin of the Seismological Society of America, 68, Computer-Assisted Intervention — MICCAI 2015, Lecture Notes
no. 5, 1521–1532, https://doi.org/10.1785/BSSA0680051521. in Computer Science, vol. 9351: Springer, 234–241, https://doi.
Anikiev, D., C. Birnie, U. bin Waheed, T. Alkhalifah, C. Gu, D. J. org/10.1007/978-3-319-24574-4_28.
Verschuur, and L. Eisner, 2023, Machine learning in microseismic Ross, Z. E., and Y. Ben-Zion, 2014, Automatic picking of direct P, S
monitoring: Earth-Science Reviews, 239, 104371, https://doi. seismic phases and fault zone head waves: Geophysical Journal
org/10.1016/j.earscirev.2023.104371. International, 199, 368–381.
Baer, M., and U. Kradolfer, 1987, An automatic phase picker for local Tegtow, W., N. Boitz, and S. Shapiro, 2023, Detection of microearth-
and teleseismic events: Bulletin of the Seismological Society of quakes triggered by hydraulic fracturing using time- and frequency-
America, 77, no. 4, 1437–1445, https://doi.org/10.1785/ domain DAS-trace stacking techniques: Presented at the 28th IUGG
BSSA0770041437. General Assembly.
Lellouch, A., N. J. Lindsey, W. L. Ellsworth, and B. L. Biondi, 2020, Wang, R., B. Schmandt, M. Zhang, M. Glasgow, E. Kiser, S. Rysanek,
Comparison between distributed acoustic sensing and geophones: and R. Stairs, 2020, Injection-induced earthquakes on complex fault
Downhole microseismic monitoring of the forge geothermal experi- zones of the Raton Basin illuminated by machine-learning phase
ment: Seismological Research Letters, 91, no. 6, 3256–3268, https:// picker and dense nodal array: Geophysical Research Letters, 47,
doi.org/10.1785/0220200149. no. 14, e2020GL088168, https://doi.org/10.1029/2020GL088168.
Moore, J., S. Simmons, J. McLennan, C. Jones, G. Skowron, Zhu, W., and G. C. Beroza, 2018, Phasenet: A deep-neural-network-
DOI:10.1190/leedff.2024.43.issue-1

P. Wannamaker, G. Nash, et al., 2019, Utah FORGE: Phase 2C based seismic arrival-time picking method: Geophysical Journal
topical report: U.S. Department of Energy, https://doi. International, 216, no. 1, 261–273, https://doi.org/10.1093/gji/
org/10.15121/1578287. ggy423.

Call for papers: Frontiers in CALL FOR

electromagnetic geophysics PAPERS
This special section invites technical papers focusing on EM data
analysis using both physics-based and data-driven techniques Special Section:
and their applications to yield high-resolution resistivity structures Understanding the subsurface helium system to
for resource exploration, environmental studies, and geological
modeling. advance helium exploration and development
The editors of Interpretation invite papers on the topic, “Understanding
Application of interest includes but is not limited to: the subsurface helium system to advance helium exploration
• EM data processing, simulation, inversion, uncertainty and development.” Suggested topics include, but are not limited to:
estimation, and interpretation • Helium exploration and appraisal case studies
• Machine-learning applications on EM data analysis • Regional helium system studies
• Integrated analysis and interpretation with other geoscientific • Redevelopment/continuing development
data (e.g., seismic, potential field, rock physics, geochemistry, of existing helium fields
and outcrop measurements) • Evaluating carbonate reservoirs for
• Field case studies highlighting the value of information extracted potential helium accumulations
from EM data • Tools and workflows for evaluating legacy
Manuscript submission deadline: ....................1 July 2024 inert gas identifications for helium
Peer review complete: .....................................1 December 2024 • Understanding the nature of helium
Publication of issue: ........................................May-June 2025 generation, migration, and accumulation
in the subsurface
Special section editors: Image courtesy of the Saskatchewan
• Comparisons between hydrocarbon and/ Geological Survey and Advanced
Colin Farquharson, cgfarquh@mun.ca or CO2 -associated helium systems and Hydrocarbon Stratigraphy
Sihong Wu, swu36@central.uh.edu “green” helium systems
Sizhuang Deng, szhdeng@gmail.com
Submission deadline: 1 FEB 2024 ❘ Publication of issue: NOV 2024
Xiaolong Wei, xwei2@stanford.edu
https://library.seg.org/page/inteio/interpretation-understanding-
Muhammad Rizwan Asif, rizwanasif@ece.au.dk subsurface-helium-system
Giulio Vignoli, gvignoli@unica.it
Special section editors: Christopher Smith, Sean Brennan, Jacinda
https://library.seg.org/page/gpysa7/geophysics-frontiers-in- Brown, Jake Cammack, Vsevolod Egorov, Duncan Mackenzie, Tim
electromagnetic-geophysics Rynott, David Seneshen, Steven Tedesco, and Melinda Yurkowski

22 The Leading Edge January 2024 Special Section: Microseismic monitoring

 DIGITALIZATION IN
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

GEOSCIENCES SYMPOSIUM
Shaping Innovation for
a Sustainable Future

21-24 OCTOBER 2024

CALL FOR AL KHOBAR, SAUDI ARABIA

ABSTRACTS IS
NOW OPEN
DEADLINE - 1st MAY 2024
EXCELLENCE IN ASIA
Energising Now and for the Future

www.geo4event.com REGISTER TODAY

DOI:10.1190/leedff.2024.43.issue-1

HOSTED BY P R I N C I PA L S P O N S O R ORGANISING SOCIETIES

2024
Middle East & Africa Events
12–14 Feb – Dhahran, Saudi Arabia 24–26 Sep – Al Khobar, Saudi Arabia 19–21 Nov – Al Khobar, Saudi Arabia
International Petroleum Technology SEG | SPWLA Workshop: Deep Reservoir Imaging and
Conference (IPTC) Seismic Petrophysics Characterization
Registration open Save the date Save the date

23–25 Apr – Muscat, Oman 21–24 Oct – Al Khobar, Saudi Arabia 10–12 Dec – Kuwait
Next Generation Geophysical Land and AAPG/EAGE/SEG: GEO 4.0 – Addressing the Challenges in
Shallow Water Acquisition Digitalization in Geosciences Symposium Processing Data for Land and Shallow
Call for abstracts closes 19 Jan Call for abstracts opens 4 Jan Water in the Middle East
Save the date
21–23 May – Abu Dhabi, UAE 5–8 Nov – Muscat, Oman
Recent Advances in Geophysical Challenges & Advances in Velocity
Reservoir Characterization and Model Building
Monitoring of CO2 Sequestration in Save the date
Carbonate Reservoirs
Call for abstracts closes 23 Feb

Current information as of 11 December 2023. Please check seg.org/events for up-to-date information.

Special Section: Microseismic monitoring January 2024 The Leading Edge 23

 Real-time passive seismic monitoring using DAS —
Today’s solutions and remaining challenges
Takashi Mizuno1 and Joël Le Calvez1
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

https://doi.org/10.1190/tle43010024.1

Abstract connection between the field and office. For example, in the area
Distributed acoustic sensing (DAS) has been widely studied, of earthquake seismology, a nationwide permanent borehole
and it has been applied in several areas of seismology. Passive station network has been maintained in Japan (Ito and Kuwahara,
seismic monitoring is one of the areas in which the application 1997; Obara, 2003). It is connected to real-time processing
of DAS is considered. Common challenges of DAS in geophysics infrastructure to improve detectability of small natural earth-
include its unique response, redundant but relatively low signal- quakes. A similar concept has been implemented for induced
to-noise ratio measurement, and high throughput of data from seismic monitoring of oil and gas reservoirs (e.g., López-Comino
the acquisition system. Here, we introduce our recent attempts et al., 2018). Recently, distributed acoustic sensing (DAS) has
to address these challenges in the implementation of real-time been considered as an acquisition system for passive seismic
DAS passive seismic monitoring. Survey design is a critical step monitoring. Considering versatile deployment options and long
in evaluating the capability and limitation of the sensor network, aperture of the seismic array, DAS-based networks will bring
and it is the same in DAS. The survey design algorithm is updated more opportunities for passive seismic monitoring. There have
considering DAS response, and we observe the unique nature of been several successful case studies (e.g., Webster et al., 2013;
a DAS network compared to a geophone network. Considering Karrenbach et al., 2017; Molteni et al., 2017). However, we
real-time monitoring, the large volume of DAS data creates a should note that deploying DAS is just one of the steps toward
bottleneck if we simply apply an existing real-time processing real-time DAS passive seismic monitoring. Due to the unique
DOI:10.1190/leedff.2024.43.issue-1

model. To accomplish real-time processing, we distributed com- nature of DAS monitoring systems (e.g., high throughput of
putation between a well site and processing center. At the well site, data from the acquisition system, single-component [1C] strain
the traditional signal enhancement, as well as deep learning-aided mea­surement, and relatively low signal-to-noise ratio [S/N] but
signature detection, are performed. The number of traces is redundant measurement), a conventional geophone-based work-
reduced, while information is enhanced. Then, the processing flow is not applicable. The objective of this article is to introduce
result is transmitted to the processing center to complete event- our recent attempts to address each of these aspects in order to
location and magnitude computations. We review the technology perform real-time passive seismic monitoring by leveraging a
and discuss remaining challenges in passive seismic monitoring DAS system.
while leveraging DAS-acquired data. Figure 1 shows the conventional workflow of real-time passive
seismic monitoring. Usually, the project should involve survey
Introduction design, during which the detectability of earthquakes and the
Monitoring of reservoir seismicity was first introduced for uncertainty of hypocenter estimates are assessed. Because DAS
waste fluid injection (e.g., Collins and Young, 2000) and geother- measures the 1C strain parallel to the borehole axis, the amplitude
mal development (e.g., Majer and McEvilly, 1979; Denlinger and of the waveform is different from what is measured by a 3C
Bufe, 1982; Eberhart-Phillips and Oppenheimer, 1984). Since geophone. Also, the noise level of a DAS system is different from
the late 1990s, the technology has been employed for hydraulic that observed with a geophone. Therefore, we need to rerun the
fracturing in tight and unconventional reservoirs (e.g., Warpinski survey design for a DAS acquisition system while considering the
et al., 1998a, 1998b). Recently, the technology has been applied nature of its response characteristics and noise. To locate events,
in CO2 injection monitoring (Kaven et al., 2015). a migration-based method is preferred due to its robustness and
Most of the time, a borehole seismic sensor is considered for the nature of the automatic operation for real-time application
passive seismic monitoring due to its superior sensitivity and (e.g., Mizuno et al., 2021). In the geophone monitoring case, the
noise level in downhole environments compared to surface field acquisition system sends all channels to the processing center
stations. Rutledge and Phillips (2003) studied earthquakes to perform the migration-based event-location process. This is
induced by hydraulic stimulation in the Carthage Cotton Valley not feasible in the DAS case due to the high throughput of data
gas field in East Texas. In California, the San Andreas Fault from the acquisition system, which is one to two orders higher
Observatory at Depth (SAFOD) project was conducted for than that of the geophone monitoring system. To overcome this,
in-situ monitoring of seismicity occurring at the San Andreas we can consider two solutions. The first is an edge computation
Fault (e.g., Zoback et al., 2005). For many decades, three- solution to complete all processing at the well site by having an
component (3C) geophone systems have generally been used, additional computation resource at the well site. The second is
and real-time monitoring has been performed using an Internet distributed computation from the well site to the office by using

Manuscript received 7 September 2023; accepted 23 October 2023.

1
SLB, Houston, Texas, USA. E-mail: tmizuno@slb.com; jcalvez2@slb.com.

24 The Leading Edge January 2024 Special Section: Microseismic monitoring

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 1. Processing workflow for real-time passive seismic monitoring. The workflow was established for 3C geophone-based measurements, and we need to upgrade it for a
DAS-based system. When using a DAS-based monitoring system, we need to introduce new technologies in each aspect due to characteristics of DAS. This article focuses on survey
design and processing.
DOI:10.1190/leedff.2024.43.issue-1

the existing network infrastructure. Each option has advantages A DAS system presents different characteristics in its ability
and disadvantages. We developed a new system that falls in the to detect microseismic activity in a 3D space compared to a
second category, and we introduce it in a case study later in this geophone-based system. To understand the difference between
article. Finally, we discuss how the magnitude of each earthquake a 3C geophone array and a DAS array, it is useful to compare
is estimated from DAS data. The magnitude (moment magnitude the detectability of each sensing point between a geophone and
[MW]) is defined by using a seismic moment of data that is com- a DAS system. Figure 2a shows the minimum magnitude maps
puted from a displacement spectrum. We need to revise the of a single 3C geophone and a single DAS measurement point.
magnitude computation algorithm to handle the strain waveform While the minimum detectable magnitude decreases with dis-
acquired from the DAS system to estimate the seismic moment tance in all directions when considering a 3C geophone, this is
from DAS data. not the case for a DAS-based system. The DAS system presents
a “shadow” in the direction normal to the axis of the optical
Passive seismic survey design for DAS: Detectability fiber. The nature of DAS response explains the shadow: DAS
The main objectives of the survey design are to (1) evaluate response becomes zero for seismic waves, which incidents normal
the limitation of sensitivity in terms of the magnitude of the to the fiber because apparent slowness becomes zero. We need
passively monitored seismic event; (2) infer the uncertainty of a dense DAS system to avoid a shadow effect when compared
attributes of each event, which includes uncertainty in hypocenter to a similar 3C geophone array. Figure 2b shows the minimum
determination; and (3) optimize sensor distribution to maximize magnitude detected by a 3C geophone network and a DAS
sensitivity while minimizing the uncertainty of the various network for a 330 m long receiver array. The 3C geophone array
attributes extracted from the measurements. The survey design assumes 30 m spacing between sensors (typical commercial
is critically important because deployment limitations often exist acquisition geometry), while the DAS system assumes 10 m
due to technical or financial constraints. Raymer and Leslie spacing (typical commercial spatial sampling setup). The DAS
(2011) developed an algorithm to address the first and second array never shows a detection shadow. Also, the DAS array
objectives for geophones. The algorithm estimates S/N as the shows a more homogeneous distribution of detectability com-
quality of data computed from an amplitude derived from ray pared to the geophone array because of dense sampling (i.e.,
theory and a given noise model. Then, it computes the limitation more receivers for the same interval). Figure 3 shows the variation
of the receiver network in terms of lower threshold of event of minimum magnitude detectable by a DAS array in terms of
magnitude. The invertibility of the event location and MW is the distance from the well in one of the real examples. We also
computed from singular value decomposition of the data kernel plot the magnitude of the event detected (details are given in
of the forward modeling problem. Mizuno and Le Calvez (2023a) the following section). The minimum magnitude predicted by
extend the work by Raymer and Leslie (2011) for DAS-based the survey design reasonably explains the lower bound of mag-
monitoring, considering the DAS response as a 1C strain nitude of the events. We should note that the accuracy of the
response. In this article, we focus on the detectability estimate survey design depends on the assumption of the noise model
for a DAS sensor array. and focal mechanism.

Special Section: Microseismic monitoring January 2024 The Leading Edge 25

 Distributed computing system
for real-time passive seismic
monitoring: QC, detection, location,
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

and magnitude estimation

Real-time processing for passive
seismic/earthquake networks has been
done for the last several decades in
academia and industry. Key components
of the system are the sensor, network,
and processing system. Traditionally,
the regional to local seismicity sensing
network consists of tens of stations, and
each station has a single component or
three components. Each station is tele-
metered using a telephone line or
Internet connection. Adding only one
DAS-based sensor network to the exist-
ing monitoring infrastructure strains
the overall system due to high through-
put (two to three orders more) of data
from the acquisition system. The data
volume changes with the length of the
borehole (interrogating length of the
DOI:10.1190/leedff.2024.43.issue-1

monitoring fiber) and the chosen spatial

and temporal sampling rate. For exam-
ple, assuming the monitoring of a total
wavefield by 2500 channels of a DAS
system using a 1 kHz sampling rate, the
data rate becomes 10 MB/s, 864 GB/
day, 26 TB/month. Two of the chal-
lenges that we have is how to QC the
acquired data and how to perform the Figure 2. Comparison of minimum magnitude detected by the 3C geophone (left) and the DAS receiver (right), assuming
processing of such a large volume of (a) single and (b) multiple receivers. The monitoring well is vertical (green trajectory), and the receiver position is presented
as a disk. With a single receiver, the 3C geophone shows isotropic variation of the minimum magnitude distribution in the 3D
data. First, a large volume of data is
space. Anisotropic distribution is presented by the DAS system. In multiple receivers (12 receivers), the DAS network shows a
hard to review by relying on human eyes homogeneous distribution by canceling the anisotropy effect from each receiver.
only. Therefore, it is difficult to QC.
Second, real-time processing is desired with automation of the
process because it is hard to assign experts 24 hours a day and
seven days a week, particularly for industry applications. This
automation should not be a black box because the reliability of
processing results requires verification by a qualified geoscientist.
We need to QC large data volumes and make the real-time
automated process as transparent as possible. To achieve this, we
have two options. The first is edge computing to complete process-
ing at the well site. The second is distributing the computing
between the edge (well site) and the processing center. The system
that we developed uses the second option. The practical advantage
of the distributed option is the possibility of integrating the data
from different wells, which fits small- to large-scale monitoring
projects. In addition, a high-performance computing facility is
not needed at the well site, which is preferable considering the
need for a small footprint on any well site. This is a key advantage
in a long-term monitoring scenario. Figure 3. The magnitude-distance distribution for the microseismicity (blue dots) and
spatial distribution of the minimum magnitude estimated from the DAS microseismic
In our system, the components are (1) computation of opti- survey design (red dotted line). The distance of the event is measured from the closest
mized data by edge computation (well site), (2) automated geophys- sensing point of DAS to the event. For this computation, 10−9 strain and a homogenous
ics processing for the data stream from the well site with human radiation pattern are assumed.

26 The Leading Edge January 2024 Special Section: Microseismic monitoring

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 4. Example of (a) raw seismic data from the DAS interrogator box and (b) optimized data continuously delivered from well site edge computation. The raw data have 1 kHz sampling
and include data from the shallow to the deeper section of the borehole. The optimized data include enhanced seismic data (depth trimming and signal enhancements are applied) and
the mask trace provided by the deep learning phase detection for raw seismic data. Although information is enriched compared to the original raw seismic data, the volume of the data is
significantly decreased. This enables real-time data QC and processing at the computation center.

can be done at the computing center if no computation facility is

available at the well site.
Figure 5 shows an example of event locations from the mask
traces. Event location is estimated by using migration-based
methods. Note that constraints in the depth range are imposed
(close to the depth where the fiber-optic cable is deployed) because
DOI:10.1190/leedff.2024.43.issue-1

single-well monitoring has limited constraints in terms of 3D

event location in this example.
In passive seismic monitoring, source parameters have been
estimated from 3C geophone recordings for the last several
decades. Among those parameters, seismic moment and MW
are the most critical. We developed an algorithm to compute
MW from DAS data by modifying the geophone-based algorithm
(Mizuno and Le Calvez, 2023b). With the event location esti-
mated from the migration-based method described earlier, the
algorithm applies a timing window for the P and S signal part
Figure 5. Map view of the event location estimated from the migration-based automatic of the DAS data to extract the P and S signals. Then, we invert
location algorithm, showing approximately 150 events from 20 minutes of continuous the signal moment and corner frequency of the displacement
acquisition. Note that because this DAS is deployed only in a single borehole (red
trajectory) and data at the horizontal section are processed (DAS sensing points are
spectrum for the P and S wave of each event, assuming an
presented as red disks), events are assumed in the same depth as the borehole. Therefore, omega-squared spectrum model (Brune, 1970; Boatright, 1978).
only distance from the borehole can be reliable in this display. We estimate the displacement spectrum from DAS data through
data conversion of DAS to a 1C geophone. Considering the
intervention at the office, and (3) a data link connection between wavenumber response of DAS data, one may consider the conver-
the well site and the office. The optimized data generated by sion to take place in the wavenumber domain. However, this
edge computation (Figure 4) include signal enhancement to requires consistency in data quality along the entire array, which
increase S/N, deep learning-based seismic signature identification is quite difficult to achieve. Hence, we apply a trace-by-trace
(Hill, 2022), and depth windowing to focus on the zone of domain conversion (e.g., Mizuno et al., 2019) for the P and S
interest. The data are referred to as “optimized data” because data by using modeled apparent slowness inferred from the
they are ready for QC and process in real time. The signature location that is estimated by the migration-based event-location
of the DAS data is enhanced by the signal processing step, which algorithm. We assume an infinitesimal strain model because
includes local stacking, and the data are annotated by a machine seismic moment determination requires modeling for the low-
learning algorithm to help interpretation. The “mask” from deep frequency side where the gauge length effect is minimal. We
learning represents the probability of the phase arrival of the first test the method with a synthetic data set. We generate the
passive seismic signal, and computation is completed in the microseismic synthetic waveforms by using ray theory (Leaney
acquisition box (Hill, 2022). This is like adding a subtitle to a et al., 2019) and compute the DAS response by using finite
television broadcast in the field of geophysics. In addition, the differentiation of neighboring geophones (Mizuno et al., 2020).
data volume is significantly decreased and is therefore viewable We generated synthetic data for the events of an MW of −2. The
by geoscientists. This volume reduction helps in real-time data inversion estimated that MW = −2 ± 0.05 for the 3C synthetic
streaming to the office. Thus, real-time automated processing geophone data (Figure 6). For the 1C DAS synthetic data

Special Section: Microseismic monitoring January 2024 The Leading Edge 27

 generated from the 3C synthetic geo-
phone data, the inversion led to
MW = −2.2 ~ −2.3 ± 0.1 (Figure 6). We
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

interpret the difference of MW between

the two inversions as being linked to
the fact that DAS is a 1C measurement
while the geophone is a 3C measure-
ment. Having a single component
underestimates the total seismic
moment at the source and the value.
Although, the difference could be in
the practically acceptable range. One
may consider the difference between Figure 6. Comparison of M W estimated from 3C geophone data and 1C DAS data with (a) a long aperture array with 60 ft gauge
the two as a correction factor for the length and (b) a small aperture array with 20 ft gauge length.
1C magnitude estimate to the 3C MW.
The real data set was acquired using
an optical fiber cable deployed in a
horizontal well, as shown in Figure 5.
We obtained MW between −1.2 and
−0.2. Based on the following three
characteristics, we conclude that the
algorithm estimates MW reasonably
well — although there could be a bias
DOI:10.1190/leedff.2024.43.issue-1

associated with the 1C measurement.

The first characteristic is the minimum
magnitude is comparable to what was Figure 7. (a) DAS waveform and (b) and (c) displacement spectrum-fitting example for the MW = −0.25 event recorded by the single DAS
inferred from the prejob survey design, cable horizontally deployed in the borehole. The displacement spectrum is estimated after DAS to 1C geophone conversion.
considering the noise level of DAS data
(Figure 3). The second characteristic is the displacement spectrum
is reasonably estimated (Figure 7). The third characteristic is
the b-value of Gutenberg-Richter law is about 2, which is seen
for the hydraulic fracturing events (Figure 8). We conclude that
the existing geophone-based algorithm can be applied to MW
inversion for DAS with trace-by-trace strain to particle velocity
conversion. Additional cross validation with geophone-derived
magnitude is crucial to validate the method.

Conclusions
Real-time passive monitoring using a DAS-based acquisition
system presents three challenges: (1) unique response, (2) redun-
dant but relatively low S/N measurement, and (3) high throughput
of data from the acquisition system. The solution introduced in
this article has been tested in select projects, and some of the
figures presented are from those experiments. In addition to Figure 8. MW versus earthquake cumulative number distribution. The b-value is estimated
these challenges, we note that assessment of the coupling of the between 1.78 and 2.68 in this example.
optical fiber to the formation is a major challenge to be studied.
Also, further automation of data QC and better deployment Data and materials availability
technology will improve the robustness of real-time DAS passive Data associated with this research are confidential and
seismic monitoring. cannot be released.

Acknowledgments Corresponding author: tmizuno@slb.com

We thank our SLB and Sintela colleagues for discussions
throughout the technology development cycle. We are grateful to References
CrownQuest for their approval of the presentation of the data in Boatwright, J., 1978, Detailed spectral analysis of two small New York
the article. We are also grateful to Nepomuk Boitz and three anony- state earthquakes: Bulletin of the Seismological Society of America,
mous reviewers for their reviews that improved this manuscript. 68, no. 4, 1117–1131, https://doi.org/10.1785/BSSA0680041117.

28 The Leading Edge January 2024 Special Section: Microseismic monitoring

 Brune, J., 1970, Tectonic stress and the spectra of seismic shear waves Molteni, D., M. J. Williams, and C. Wilson, 2017, Detecting micro-
from earthquakes: Journal of Geophysical Research, 75, no. 26, seismicity using distributed vibration: First Break, 35, no. 4, 51–55,
4997–5009, https://doi.org/10.1029/JB075i026p04997. https://doi.org/10.3997/1365-2397.35.4.87841.
Collins, D. S., and R. P. Young, 2000, Lithological controls on seismicity Obara, K., 2003, Hi-net: High sensitivity seismograph network, Japan,
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

in granite rocks: Bulletin of the Seismological Society of America, in T. Takanami and G. Kitagawa, eds., Methods and applications
90, no. 3, 709–723, https://doi.org/10.1785/0119990142. of signal processing in seismic network operations: Springer, https://
Denlinger, R. P., and C. G. Bufe, 1982, Reservoir conditions related doi.org/10.1007/BFb0117698.
to induced seismicity at the Geysers steam reservoir, northern Raymer, D. G., and H. D. Leslie, 2011, Microseismic network design
California: Bulletin of the Seismological Society of America, 72, — Estimating event detection: 73rd Conference and Exhibition, EAGE,
no. 4, 1317–1327, https://doi.org/10.1785/BSSA0720041317. Extended Abstracts, https://doi.org/10.3997/2214-4609.20149027.
Eberhart-Phillips, D., and D. H. Oppenheimer, 1984, Induced seismicity Rutledge, J. T., and W. S. Phillips, 2003, Hydraulic stimulation of natural
in the Geysers geothermal area, California: Journal of Geophysical fractures as revealed by induced microearthquakes, Carthage Cotton
Research, 89, no. B2, 1191–1207, https://doi.org/10.1029/ Valley gas field, East Texas: Geophysics, 68, no. 2, 441–452, https://
JB089iB02p01191. doi.org/10.1190/1.1567214.
Hill, D., 2022, Microseismic detection using DAS with ML for per- Warpinski, N. R., P. T. Branagan, R. E. Peterson, and S. L. Wolhart,
manent CCS monitoring: Presented at the Role of Geophysics in 1998a, An interpretation of M-Site hydraulic fracture diagnostic results:
Carbon Capture and Sequestration Workshop, SEG. Rocky Mountain Regional/Low-Permeability Reservoirs Symposium,
Ito, H., and Y. Kuwahara, 1997, Multi depth level seismic observation SPE, Extended Abstracts, https://doi.org/10.2118/39950-MS.
in the active fault drilling borehole at Nojima Hirabayashi: Presented Warpinski, N. R., P. T. Branagan, R. E. Peterson, S. L. Wolhart, and
at the Fall Meeting, Seismological Society of Japan. J. E. Uhl, 1998b, Mapping hydraulic fracture growth and geometry
Karrenbach M., A. Ridge, S. Cole, K. Boone, D. Kahn, J. Rich, K. using microseismic events detected by a wireline retrievable accel-
Silver, and D. Langton, 2017, DAS microseismic monitoring and erometer array: Gas Technology Symposium, SPE, Extended
integration with strain measurements in hydraulic fracture profiling: Abstracts, https://doi.org/10.2118/40014-MS.
Unconventional Resources Technology Conference, Extended Webster, P., J. Wall, C. Perkins, and M. Molenaar, 2013, Micro-seismic
Abstracts, https://doi.org/10.15530/URTEC-2017-2670716. detection using distributed acoustic sensing: 83rd Annual International
Kaven, J. O., S. H. Hickman, A. F. McGarr, and W. L. Ellsworth, Meeting, SEG, Expanded Abstracts, 2459–2463, https://doi.
2015, Surface monitoring of microseismicity at the Decatur, Illinois, org/10.1190/segam2013-0182.1.
DOI:10.1190/leedff.2024.43.issue-1

CO 2 Sequestration Demonstration Site: Seismological Research Zoback, M. D., S. Hickman, and W. Ellsworth, 2005, Overview of SAFOD
Letters, 86, no. 4, 1096–1101, https://doi.org/10.1785/0220150062. phases 1 and 2: Drilling, sampling and measurements in the San Andreas
Leaney, S., T. Mizuno, E. Velez, T. Cuny, and M. Perez, 2019, Fault Zone at seismogenic depth: Presented at the Fall Meeting, AGU.
Anisotropic model building, DAS simulation and imaging: 89 th
Annual International Meeting, SEG, Expanded Abstracts, 999–1003,
https://doi.org/10.1190/segam2019-3215957.1.
López-Comino, J. A., S. Cesca, J. Jarosławski, N. Montcoudiol, S. Heimann,
T. Dham, S. Lasocki, A. Gunning, P. Capuano, and W. L. Ellsworth,
2018, Induced seismicity response of hydraulic fracturing: Results of
a multidisciplinary monitoring at the Wysin site, Poland: Scientific
Reports 8, https://doi.org/10.1038/s41598-018-26970-9.
Majer, E. L., and T. V. McEvilly, 1979, Seismological investigations at
the Geysers geothermal field: Geophysics, 44, no. 2, 246–269, https://
doi.org/10.1190/1.1440965.
Mizuno, T., and J. Le Calvez, 2023a, Quantitative evaluation of DAS
passive seismic monitoring: Theory and case studies: 84th Annual
Conference and Exhibition, EAGE, Extended Abstracts, https://
doi.org/10.3997/2214-4609.202310429.
Mizuno, T., and J. Le Calvez, 2023b, Moment magnitude inversion
using single-well DAS: Presented at the Third International Meeting
for Applied Geoscience & Energy, SEG/AAPG.
Mizuno, T., S. Leaney, J. Le Calvez, F. Nasser, and M. L. Khaitan, 2019,
The significance of gauge length in particle velocity estimation from
DAS data: VSP and microseismic cases: 89th Annual International
Meeting, SEG, Expanded Abstracts, 4869–4873, https://doi.
org/10.1190/segam2019-3215250.1.
Mizuno, T., J. Le Calvez, and P. Bettinelli, 2020, Development of DAS
geophysics by simulating DAS data using geophone data: Presented
at the Advances in Distributed Sensing for Geophysics Workshop,
SEG/AGU.
Mizuno, T., J. Le Calvez, and D. Raymer, 2021, Coalescence micro-
seismic mapping for distributed acoustic sensing (DAS) and geophone
hybrid array, in Y. Li, M. Karrenbach, and J. B. Ajo-Franklin, eds.,
Distributed acoustic sensing in geophysics: Methods and applications:
AGU and Wiley, https://doi.org/10.1002/9781119521808.ch12.

Special Section: Microseismic monitoring January 2024 The Leading Edge 29

 Archaeological geophysical investigation
of Uzun Rama Steppe kurgans, Goranboy
Kamal Bayramov1,2,3, Gunel Alizada1,2,3, Sarvar Mammadov1,2,3, Vusal Azimov1,2,3, Malik Abdullayev1,2,3, Clara Jodry1,2,3, and Maksim Bano1
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

https://doi.org/10.1190/tle43010030.1

Abstract materials, and past burial rituals (Lyonnet, 2009; Jalilov, 2018;
Geophysical methods can provide valuable information when Laneri et al., 2019; Poulmarc’h et al., 2019; Gasymov, 2020;
imaging man-made subsurface structures prior to archaeological Kirichenko, 2021). These reveal the obligatory change of their
excavations. An archaeological geophysical survey was conducted way of life and transition to a seminomadic lifestyle. Moreover,
on funeral chambers (also known as kurgans) from the Early ceramics, weapons, and cloth remnants that were found in
Bronze Age, in the Shadili-Uzun Rama Steppe of the Goranboy excavated kurgans show noticeable changes in religious view.
region, near the Kurakchay river gorge in Azerbaijan. This In this study, our objective is to provide pre-excavation
multimethod survey is based on a ground-penetrating radar images based on geophysics to lead research, differentiate and
profile, heat map of the total magnetic intensity, and contiguous locate the entrance and main chambers, define the thickness
profiles of electrical resistivity and seismic refraction. Brief and width of the kurgans, and possibly locate remains in burned
processing and default inversion methods enabled us to obtain mounds. These nondestructive methods have proven to be efficient
geophysical images in accordance with the lithology (a subsoil in the detection and characterization of these structures (e.g.,
comprised of alluvial-proluvial conglomerate terrasse). The shape Morgunova and Khokhlova, 2006; Polin et al., 2017; Tóth et al.,
of the kurgans was recovered mainly through electrical resistivity 2018; Green et al., 2021; Hegyi et al., 2021). Yet, to our knowl-
tomography, enabling future targeted excavations. Overall, this edge, it has been applied only once in Azerbaijan by Fassbinder
study further testifies that geophysics provides valuable informa- et al. (2015) while using the magnetic method. Thus, we imple-
DOI:10.1190/leedff.2024.43.issue-1

tion for archaeological investigations, although higher resolution mented a multimethod survey using ground-penetrating radar
could be achieved by using more advanced field methodology (GPR), magnetic, electrical resistivity tomography (ERT), and
and processing. seismic refraction tomography. Each of these methods are based
on different physical laws. The combination of all enables the
Introduction collection of different resolutions and parameters, which may
Kurgans are funeral chambers that have great historical and be correlated and compared for a better understanding of the
cultural value. They attest to the burial tradition of the nomadic archaeological sites (e.g., Herrmann and Hammer, 2019).
populations that covered a vast area between Europe and Asia
during the first thousand years Before the Common Era (BCE) Archaeological setting
(Davis-Kimball et al., 2000; Morgunova and Khokhlova, 2006; The investigated kurgans are in the Shadili-Uzun Rama
Lyonnet, 2009; Demkin et al., 2014; Tóth et al., 2018; Poulmarc’h Steppe of the Goranboy region, near the Kurakchay river gorge,
et al., 2019). To improve the efficiency of the excavation process, in the Lesser Caucasus foothills of Azerbaijan. The steppe
geophysical methods have been applied widely and effectively presents a flat relief in which kurgans are easily identifiable. The
for many years. They provide clear and useful images of the area is located on an alluvial-proluvial terrasse of conglomerate
archaeological targets hidden underground such as kurgans, that consists of pebbles mixed with clay and sand over a layer
walls, ditches, and anthropogenic or natural cavities (Wynns of sandstones that was formed during the Cenozoic Period
et al., 1986; Kvamme, 2001; Gaffney, 2008; El-Qady and (Sosson et al., 2010; Alizadeh et al., 2016). In Figure 1, we can
Metwaly, 2019). Geophysics quickly gives trustworthy and see that the surface is covered with pebbles, and there is almost
detailed information of subsurface features with minimum no vegetation. The difference of elevation between the upper
disruption in a way that is understandable by nonexperts (e.g., part of the terrace and the lower fields (cultivated land) varies
Linford, 2006; Batayneh, 2011). from 5 to 10 m (Laneri et al., 2019).
This article presents an archaeological geophysical survey Past exploration revealed that this area was dedicated to
that was performed in May 2022 to investigate Early Bronze burying for more than two millennia. It contains numerous large
Age kurgans located in Uzun Rama Steppe, Goranboy (Figure 1). kurgans of Early Bronze, which correspond to the Kura-Arexed
During the past few decades in Azerbaijan, archaeologists Period (ca. 3500–3000 BCE), and relatively smaller burials of the
performed and reported several excavation operations on kurgans Late Bronze/Early Iron ages (Jalilov, 2018; Laneri et al., 2019).
all over the region (approximately 205 excavations). This provided The main difference between these ages is the passage from
a wealth of information on their structure, construction methods, collective to individual burials. Burial rites involved burying the

Manuscript received 27 January 2023; revision received 23 May 2023; accepted 28 September 2023.
1
Université de Strasbourg, CNRS, EOST, Institut Terre et Environnement de Strasbourg, UMR 7063, Strasbourg, France. E-mail: kamal.bayramov@
ufaz.az; g.alizada@ufaz.az; sarvar.mammadov@ufaz.az; vusal.azimov@ufaz.az; malik.abdullayev@ufaz.az; cjodry@unistra.fr; maksim.bano@unistra.fr.
2
Azerbaijan State Oil and Industry University, Baku, Azerbaijan.
3
French Azerbaijani University, Baku, Azerbaijan.

30 The Leading Edge January 2024

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 1. Field pictures of the (a) seismic and (b) ERT survey, with labels of some of the equipment used. The difference in elevation between the upper part of the terrace and the lower
fields (cultivated land) varies from 5 to 10 m. (Courtesy of C. Jodry).

Data acquisition and processing

Figure 2 shows the study area with
five unexcavated kurgans that are visible
as discolored round mounds that vary
from 15 to 25 m in diameter and 1.5 to
2 m in height from ground level. Our
geophysical acquisitions (Table 1) focus
DOI:10.1190/leedff.2024.43.issue-1

on three kurgans aligned in the north-

west–southeast direction. It is worth
mentioning that it rained abundantly
in the days before the geophysical acqui-
sitions, which means that the alluvial
layer has a high water content.
Ground-penetrating radar. GPR is
responsive to the variations of dielectric
properties in shallow depth (Neal,
2004). A standard GPR device consists
of transmitting and receiving antennas,
Figure 2. Georeferenced map of Uzun Rama Steppe (Goranboy, Azerbaijan) depicting the different geophysical acquisitions: the along with a central unit. The transmit-
two georadar profiles, the contiguous seismic and ERT lines, and the magnetic map survey. (Courtesy of Google). ting antenna sends a high-frequency
(10–2600 MHz) electromagnetic (EM)
Table 1. Descriptions of each of the geophysical acquisitions. pulse into the ground, which is reflected/
diffracted back as it passes through
Number discontinuity/heterogeneity in the
Number of points Length subsurface (Davis and Annan, 1989).
Survey of profiles per profile of the profiles Direction Reflections can also be scattered by
45 m north to south (250 MHz) objects/cavities within the medium and
GPR 2
40 m east to west (500 MHz) returned to the receiving antenna
(Mochales et al., 2008). Higher-
Magnetic 20 33 19 × 32 m east to west and north to south
frequency antennas transmit a signal
ERT 1 48 70.5 north to south with a physically smaller wavelength,
Seismic 1 48 70.5 north to south resulting in better resolution but lower
investigation depth. This is the opposite
bodies of the deceased in large pits, with some of their belongings for low-frequency antennas (Jol, 2009). GPR systems provide 2D
with them or in adjoining small spaces. The walls of the pits were subsurface images for a range of applications including archaeologi-
made of cobblestone, wood, and clay. According to religious tradi- cal, where it provides highly resolved information for archaeologists
tions, the burial chamber was filled with the same type of material to determine excavation sites (e.g., Leucci and Negri, 2006;
as the walls, then the structure was set on fire and completely Ortega-Ramírez et al., 2020).
burned. Indeed, previously excavated kurgans recovered The data acquisition includes two profiles using MALÅ’s two
semicharred wood, similar to coal. shielded antennas with frequencies of 250 and 500 MHz. The

January 2024 The Leading Edge 31

 first profile, from north to south (250 MHz), is 45 m long. The using the least-squares method. This inversion process was per-
second, from east to west (500 MHz), is 40 m long. In our case, formed using Geometrics’ Pickwin and Plotrefa softwares.
only the 500 MHz antenna had good results due to the relatively Magnetic method. The magnetic method investigates the
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

shallow depth of the kurgan. For every 5 cm spacing, we sent 16 spatial distribution of magnetization in the subsoil and can detect
GPR pulses and stacked them as a trace using MALÅ’s Ground buried objects or delimitate geologic structures. The measurement
Vision software. The resulting reflection profiles were processed on the field consists of recording the ambient magnetic field point
using the Radlab software package written in MATLAB and by point on a preset grid. To get only the effects of shallow sources,
developed at the Institut Terre et Environnement de Strasbourg. the magnetic anomaly of the total field intensity is then computed
The processing sequence involved time-zero shifts of first arrivals, by removing an estimation of the regional field. This magnetic
low-frequency removal or direct-current (DC) filtering, amplitude method is one of the most used geophysical techniques to identify
gain (linear and/or exponential), flat reflections filtering, and and map archaeological sites before excavation, providing valuable
frequency band-pass filtering of 150–850 MHz. insight into past cultures and civilizations (e.g., Eppelbaum, 2011;
Electrical resistivity tomography. ERT is based on detecting Fassbinder, 2017).
lateral and vertical electrical resistivity contrasts in the subsurface, We conducted measurements covering 19 m in the east–west
which are influenced by porosity, type of rock, clay, and water direction and 32 m in the north–south direction across the
content (Saas et al., 2008; Loke et al., 2013). This technique southernmost kurgan (Figure 2) by using a G-857 magnetometer
uses two electrodes to inject an electrical current into the subsoil (Geometrics’ proton precession magnetometer). The acquisition
while measuring the voltage, or potential difference, between includes 20 profiles carried out from east to west with 1 m
two other electrodes. This method provides high-resolution spacing in between them. Each profile contains 33 stations
imaging that is suitable for applications in archaeology and burial (measurement points) spaced 1 m apart. The processing was done
sites (Hesse et al., 1986; Tsokas et al., 2008; Di Maio et al., based on open-source Python to remove the temporal changes
2012; Tsokas et al., 2018). of earth’s magnetic field during the acquisition period. For this
The survey was carried out across what we assume to be three purpose, we apply a median filter by using an average value of
DOI:10.1190/leedff.2024.43.issue-1

kurgans in a north to south direction (Figure 2). These measure- 50,000 nT. The data are represented as a 2D heat map by inter-
ments were done with an IRIS Instruments’ Syscal Pro resistivity polating the parallel data profiles.
meter using 48 electrodes with 1.5 m spacing. In order to have
good vertical resolution and higher penetration depth, two elec- Results and discussion
trode configurations were employed: Wenner-Schlumberger (WS) Ground-penetrating radar. Figure 3a shows the GPR profile,
and dipole-dipole (DD) arrays (e.g., Dahlin et al., 2004). from 10 to 25 m, in the east–west direction (Figure 2), acquired
As an initial procedure, we eliminated noisy data points (e.g., with the 500 MHz shielded antenna. We can see multiple dif-
outlier voltage and injection points) by using IRIS Instruments’ fractions in the near subsurface (0–1 m depth). This enables us to
PROSYS software. Appending WS and DD data, as well as evaluate the velocity of EM waves at 0.1 m/ns based on the aperture
including topography, the apparent resistivity data are then inverted of the hyperbolas (e.g., Annan, 2004).
using the Geotomo Res2dinv software. We used a traditional In this instance, we interpret the diffractions as remains of
least-squares approach with default parameters (deGroot-Hedlin the unburned part of the kurgans, artifacts of the deceased, wooden
and Constable, 1990; Sasaki, 1992; Loke et al., 2003) and evaluated beam, or part of the wall. The EM wave velocity suggests a
the quality of our model based on root mean square (rms) (Gupta subsurface layer that is mainly comprised of moderate moist clay,
et al., 1997). The final model rms achieves a value of 0.96%, which which also corresponds to the filling of a kurgan.
is considered as very well defined. Note that we have omitted the GPR north–south profile
Seismic refraction. A seismic refraction profile includes obtained with a 250 MHz antenna, as there is significant signal
recording seismic waves and analyzing first-arrival times from attenuation and no reflections are discernible. This may be due to
P-wave fronts created by a man-made source. This technique low resolution, higher clay content in this direction, or higher
tracks spatial variations of mechanical properties (velocity, density, water content in the ground from rain the day before the acquisition
and elasticity) to image subsurface geometries (Steeples, 2000; (Schneidhofer et al., 2022).
Mineo et al., 2015). It has been used in the past for archaeology Electrical resistivity tomography. Figure 3b shows the inverted
surveys but often in association with another method such as resistivity profile in logarithmic scale. The inversion shows a
electrical resistivity (e.g., Leucci et al., 2007; Cardarelli et al., two-layer subsurface with a sharp limit in between. First, we have
2010; Selim et al., 2014). a medium resistivity layer (500–600 Ω.m) from the surface down
The seismic survey followed the same configuration as the to 6 m depth, then a very conductive layer (greater than 10 Ω.m).
electrical profile, with a spread of 48 geophones spaced every The maximum investigation depth is 14 m. The first layer contains
1.5 m. The shots were fired with a manual hammer of 4.41 kg and three areas of lower resistivity (30–130 Ω.m) that are limited in
a shot spacing of 3 m, starting with offsets of 1.5 m from the first thickness and length. They lie from 0–7 m, 13–28 m, and 51–70.5 m
geophone and ending 1.5 m after the last geophone. The data along the profile from north to south at 0–4 m depth.
acquisition parameters were set with a stack limit of five and a We interpret these three spots as kurgans because they are
sample interval of 0.250 ms. P-wave arrivals times were handpicked marked by rounded and colorful mounds in place. The lower-
on time-distance plots, concatenated, and subsequently inverted resistivity values are explained by a more conductive soil such as

32 The Leading Edge January 2024

 clays or sandy clays (e.g., Palacky, 1988; Saad et al., 2012), which P-wave velocities from 300 m/s at the top to 1000 m/s at the
fills the kurgans or forms its walls as shown by Jalilov (2018). bottom. The second layer shows a discrepancy of velocities along
Additionally, the higher-resistivity values for the surrounding the profile. The southern part of the profile, from X = 0 m to
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

correspond to a fluvial environment such as dry sand and gravel. X = 30 m, has a P-wave velocity of 1100 m/s. The northern part
Seismic refraction. Figure 3c shows the P-wave inversion of the profile, from X = 30 m to the end, displays a P-wave velocity
tomography with a maximum investigation depth of 10 m. We of 1800 m/s.
can see two layers of varying thicknesses and velocities. The first The velocity values are typical of weathered, more aerated,
upper layer, from the surface to approximately 2.5 m depth, shows and vegetal subsoils or uncompacted filled cavities (Leucci et al.,
a variable thickness that follows the topography and has calculated 2007). Again, we can interpret the geophysical image based on
DOI:10.1190/leedff.2024.43.issue-1

Figure 3. (a) GPR reflection profile along the east−west direction acquired by a 500 MHz antenna. (b) Inverted ERT. (c) Inverted P-wave tomography. (d) Heat map of the total magnetic intensity.

January 2024 The Leading Edge 33

 the surface markers and assume that the low-velocity layer on kurgan. All of this further testifies that geophysics provides
the surface highlights the presence of the kurgans. The higher- valuable information in archaeological investigations.
velocity layer also likely corresponds to a more compacted dry
Acknowledgments
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

sand. Although both data sets have been subjected to least-squares

inversion, the seismic model is much less detailed than the The authors thank French-Azerbaijani University (under the
electrical model, and we cannot distinguish the location and Azerbaijan State Oil and Industry University) for funding the
length of the kurgans. geophysical field trip. They also express their sincere gratitude to
Magnetic method. The magnetic result shows us a large Bakhtiyar Jalilov, head of the archaeological department, and
anomaly (positive-negative dipole) from −80 to 80 nT in the Lola Huseynova, scientific researcher for the Institute of
center of the studied area (Figure 3d). The color scale highlights Archaeology and Ethnography (under the Ministry of Science
large anomalies, but we can also clearly see smaller anomalies and Education of the Republic of Azerbaijan) for providing the
scattered around. site location and information about kurgan history. They thank
The large anomaly is obviously linked to a shallow source with Christian Camerlynck, associate professor at Sorbonne Université,
a high contrast of magnetization with the surroundings. This and Bruno Gavazzi, director of the geophysical service at Enerex,
contrast could be linked to the burning process of a kurgan (if for instructing students in the field. Finally, they would like to
the temperature is high enough so magnetization starts to align make a special acknowledgement to the entire class of the Master
in the direction of the ambient magnetic field, which would create of Geosciences 2021–2022 who participated in the field.
a strong north–south dipole). In the general case, as we are in the
northern hemisphere, we would expect to have the negative pole Data and materials availability
in the north and positive pole in the south. Here, the poles are Data associated with this research are available and can be
reversed. A possible explanation for such behavior is the exhibition obtained by contacting the corresponding author.
of strong antiferromagnetism that can happen only in certain
material (notably oxides such as hematite). This would need to be Corresponding author: cjodry@unistra.fr
DOI:10.1190/leedff.2024.43.issue-1

checked carefully in samples to be confirmed, but it could inform

on the composition of the burned materials. References
The precise shape of the kurgan is still in question because Alizadeh, A. A., I. S. Guliyev, F. A. Kadirov, and L. V. Eppelbaum, 2016,
different geometries can exhibit the same magnetic anomaly. For Geosciences of Azerbaijan: Springer International Publishing, https://
doi.org/10.1007/978-3-319-27395-2.
example, a rectangular structure can give a round anomaly when
Annan, A. P., 2004, Ground penetrating radar: Principles, procedures
it is deep enough below the surface (Fassbinder, 2017). Presumably, & applications: Sensors & Software Inc.
smaller magnetic anomalies may correspond either to smaller Batayneh, A. T., 2011, Archaeogeophysics-archaeological prospection
rooms where people put the deceased’s belongings or to some — A mini review: Journal of King Saud University - Science, 23, no.
other unknown buried objects. The area has to be excavated to 1, 83–89, https://doi.org/10.1016/j.jksus.2010.06.011.
define exactly what it is. Cardarelli, E., M. Cercato, A. Cerreto, and G. Di Filippo, 2010, Electrical
This study has shown that geophysical integrated methods resistivity and seismic refraction tomography to detect buried cavities:
Geophysical Prospecting, 58, no. 4, 685–695, https://doi.
are well suited to identify the structural remains of kurgans. These
org/10.1111/j.1365-2478.2009.00854.x.
methods provide interesting results and have proven to be comple- Dahlin, T., and B. Zhou, 2004, A numerical comparison of 2D resistivity
mentary in resolving or reducing the uncertainties associated with imaging with 10 electrode arrays: Geophysical Prospecting, 52, no.
indirect evaluations. In the future, our research will focus on 3D 5, https://doi.org/10.1111/j.1365-2478.2004.00423.x.
geophysical measurements at similar sites, as well as suitable Davis, J. L., and A. P. Annan, 1989, Ground-penetrating radar for high-
inversion techniques. resolution mapping of soil and rock stratigraphy: Geophysical
Pro s p e c t i n g , 37, no. 5, 531–551, ht t p s: //doi .
org/10.1111/j.1365-2478.1989.tb02221.x.
Conclusion Davis-Kimball, J., E. M. Murphy, L. Koryakova, and L. T. Yablonsky,
This work illustrates the applications of magnetic, seismic 2000, Kurgans, ritual sites, and settlements: Eurasian Bronze and
refraction, GPR, and ERT to investigate the archaeological site Iron Age: British Archaeological Reports, https://doi.
of Shadili-Uzun Rama Steppe in Goranboy, Azerbaijan. With org/10.30861/9781841710907.
very brief processing and default inversion methods, we provide deGroot-Hedlin, C., and S. Constable, 1990, Occam’s inversion to generate
information about the subsurface lithology and geometry and smooth, two-dimensional models from magnetotelluric data:
accurately detect the kurgan locations. This first analysis confirms Geophysics, 55, no. 12, 1613–1624, https://doi.org/10.1190/1.1442813.
Demkin, V. A., V. M. Klepikov, S. N. Udaltsov, T. S. Demkina, M. V.
good correlation between the methods and makes it possible to
Eltsov, and T. E. Khomutova, 2014, New aspects of natural science
define the direction of the kurgans, in particular with ERT. This studies of archaeological burial monuments (kurgans) in the southern
should enable archaeologists to perform targeted excavations. Russian steppes: Journal of Archaeological Science, 42, no. 1, 241–249,
However, more could be achieved by using 3D mapping and https://doi.org/10.1016/j.jas.2013.10.031.
higher-resolution geophysical methods. Indeed, a higher-frequency Di Maio, R., C. Meola, M. Grimaldi, and U. Pappalardo, 2012, New
antenna can be used for GPR, as well as a smaller spacing for the insights for conservation of Villa Imperiale (Pompeii, Italy) through
seismic refraction, to get a better resolution of the first layer. We nondestructive exploration: International Journal of Architectural
can also use a magnetic gradiometer to get the exact shape of the

34 The Leading Edge January 2024

 Heritage, 6, no. 5, 562–578, https://doi.org/10.1080/15583058.2011 Linford, N., 2006, The application of geophysical methods to archaeologi-
.593392. cal prospection: Reports on Progress in Physics, 69, 2205–2257, https://
El-Qady, G., and M. Metwaly, 2019, Archaeogeophysics: State of the doi.org/10.1088/0034-4885/69/7/R04.
art and case studies: Springer: https://doi.org/10.1007/ Loke, M. H., I. Acworth, and T. Dahlin, 2003, A comparison of smooth
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

978-3-319-78861-6. and blocky inversion methods in 2D electrical imaging surveys:

Eppelbaum, L. V., 2011, Study of magnetic anomalies over archaeological Exploration Geophysics, 34, no. 3, 182–187, https://doi.org/10.1071/
targets in urban environments: Physics and Chemistry of the Earth, EG03182.
Parts A/B/C, 36, no. 16, 1318–1330, https://doi.org/10.1016/j. Loke, M. H., J. E. Chambers, D. F. Rucker, O. Kuras, and P. B. Wilkinson,
pce.2011.02.005. 2013, Recent developments in the direct-current geoelectrical imaging
Fassbinder, J. W. E., 2017, Magnetometry for archaeology, in A. S. Gilbert, method: Journal of Applied Geophysics, 95, 135–156, https://doi.
ed., Encyclopedia of geoarchaeology: Springer, 499–514, https://doi. org/10.1016/j.jappgeo.2013.02.017.
org/10.1007/978-1-4020-4409-0_169. Lyonnet, B., 2009, The Early Bronze Age in Azerbaijan in the light of
Fassbinder, J. W. E., A. Gass, I. Hofmann, A. Belinskij, and H. Parzinger, recent discoveries: Paléorient, 40, no. 2, 115–130, https://doi.
2015, Early Iron Age kurgans from the North-Caucasus: Archaeologia org/10.3406/paleo.2014.5638.
Polona, 53, 280–284. Mineo, S., G. Pappalardo, F. Rapisarda, A. Cubito, and G. Di Maria,
Gaffney, C., 2008, Detecting trends in the prediction of the buried past: 2015, Integrated geostructural, seismic and infrared thermography
A review of geophysical techniques in archaeology: Archaeometry, 50, surveys for the study of an unstable rock slope in the Peloritani Chain
no. 2, 313–336, https://doi.org/10.1111/j.1475-4754.2008.00388.x. (NE Sicily): Engineering Geology, 195, 225–235, https://doi.
Gasymov, P. P., 2020, Cultural attribution of Early Bronze Age tombs org/10.1016/j.enggeo.2015.06.010.
under kurgans in Azerbaijan: Archaeology, Ethnology & Anthropology Mochales, T., A. M. Casas, E. L. Pueyo, O. Pueyo, M. T. Román, A.
of Eurasia, 48, no. 2, 22–28, https:// doi.org/10.17746/1563- Pocoví, M. A. Soriano, and D. Ansón, 2008, Detection of underground
0110.2020.48.2.022-028. cavities by combining gravity, magnetic and ground penetrating radar
Green, S., C. Green, and J. Schuldenrein, 2021, Archaeology as a public surveys: A case study from the Zaragoza area, NE Spain: Environmental
good: Archaeologies, 17, 1–18, https://doi.org/10.1007/ G eolog y, 53, 10 67–1077, ht t ps: //doi.org /10.10 07/
s11759-021-09414-1. s00254-007-0733-7.
Gupta, P. K., S. Niwas, and V. K. Gaur, 1997, Straightforward inversion Morgunova, N. L., and O. S. Khokhlova, 2006, Kurgans and nomads:
DOI:10.1190/leedff.2024.43.issue-1

of vertical electrical sounding data: Geophysics, 62, no. 3, 713–1003, New investigations of mound burials in the southern Urals: Antiquity,
https://doi.org/10.1190/1.1444187. 80, no. 308, 303–317, https://doi.org/10.1017/S0003598X00093637.
Hegyi, A., D. Diaconescu, P. Urdea, A. Sarris, M. Pisz, and A. Onaca, Neal, A., 2004, Ground-penetrating radar and its use in sedimentology:
2021, Using geophysics to characterize a prehistoric burial mound in Principles, problems and progress: Earth-Science Reviews, 66, no.
Romania: Remote Sensing, 13, no. 5, https://doi.org/10.3390/ 3-4, 261–330, https://doi.org/10.1016/j.earscirev.2004.01.004.
rs13050842. Ortega-Ramírez, J., M. Bano, M. T. Cordero-Arce, L. A. Villa-Alvarado,
Herrmann, J. T., and E. L. Hammer, 2019, Archaeo-geophysical survey and C. C. Fraga, 2020, Application of non-invasive geophysical
of Bronze and Iron Age fortress landscapes of the South Caucasus: methods (GPR and ERT) to locate the ancient foundations of the
Journal of Archaeological Science: Reports, 24, 663–676, https://doi. first cathedral of Puebla, Mexico: A case study: Journal of Applied
org/10.17746/1563-0110.2020.48.2.022-028. Geophysics, 174, https://doi.org/10.1016/j.jappgeo.2020.103958.
Hesse, A., A. Jolivet, and A. Tabbagh, 1986, New prospects in shallow Palacky, G. J., 1988, Resistivity characteristics of geologic targets, in M.
depth electrical surveying for archaeological and pedological applica- N. Nabighian, ed., Electromagnetic methods in applied geophysics,
tions: Geophysics, 51, no. 3, 585–594, https://doi.org/10.1190/1.1442113. volume 1: Theory, 52–129, https://doi.org/10.1190/1.9781560802631.
Jalilov, B., 2018, The collective burial kurgans of Uzun-Rama: Türkiye ch3.
Bilimler Akademisi Arkeoloji Dergisi, 93–106. Polin, S., M. Daragan, and K. Bondar, 2017, New investigations of
Jol, H. M., 2009, Ground penetrating radar: Theory and applications: Scythian kurgans and their periphery in the lower Dnieper region:
Elsevier Science. Non-destructive measurements and archaeological proof, in S. V.
Kirichenko, D. A., 2021, A new burial of the late phase of Kura-Araxes Pankova and J. Simpson, eds., Masters of the steppe: The impact of
cultural tradition/early kurgan tradition from Azerbaijan: Journal of the Scythians and Later Nomad Societies of Eurasia: Archaeopress.
Archaeology, Anthropology and Interdisciplinary Studies, 3, no. 1, Poulmarc’h, M., N. Laneri, and E. Hasanov, 2019, Innovative approach
105–117. to the research of ethnographic-archeological heritage in Ganja based
Kvamme, K. L., 2001, Current practices in archaeogeophysics, in P. Goldberg, on materials of kurgans: Theoretical & Applied Science, 77, no. 9,
V. T. Holliday, and C. R. Ferring, eds., Earth sciences and archaeology, 341–345, https://doi.org/10.15863/TAS.2019.09.77.60.
353–384, https://doi.org/10.1007/978-1-4615-1183-0_13. Sasaki, Y., 1992, Resolution of resistivity tomography inferred from
Laneri, N., B. Jalilov, L. Crescioli, G. Guarducci, J. Kneisel, M. numerical simulation: Geophysical Prospecting, 40, no. 4, 453–463,
Poulmarc’h, A. Ricci, and S. Valentini, 2019, The first season of the https://doi.org/10.1111/j.1365-2478.1992.tb00536.x.
Azero-Italian Ganja Region Kurgan Archaeological Project in western Saad, R., M. Nawawi, and E. T. Mohamad, 2012, Groundwater detection
Azerbaijan: Ancient Near Eastern Studies: 56, 135–162, https://doi. in alluvium using 2-D electrical resistivity tomography (ERT): The
org/10.2143/ANES.56.0.3286814. Electronic Journal of Geotechnical Engineering, 17, 369–376.
Leucci, G., and S. Negri, 2006, Use of ground penetrating radar to map Sass, O., R. Bell, and T. Glade, 2008, Comparison of GPR, 2D-resistivity
subsurface archaeological features in an urban area: Journal of and traditional techniques for the subsurface exploration of the
Archaeological Science, 33, no. 4, 502–512, https://doi.org/10.1016/j. Öschingen landslide, Swabian Alb (Germany): Geomorphology, 93,
jas.2005.09.006. no. 1–2, 89–103, https://doi.org/10.1016/j.geomorph.2006.12.019.
Leucci, G., F. Greco, L. De Giorgi, and R. Mauceri, 2007, Three- Schneidhofer, P., C. Tonning, R. Cannell, E. Nau, A. Hinterleitner, G. J.
dimensional image of seismic refraction tomography and electrical Verhoeven, L. Gustavsen, K. Paasche, W. Neubauer, and T. Gansum,
resistivity tomography survey in the castle of Occhiolà (Sicily, Italy): 2022, The influence of environmental factors on the quality of GPR
Journal of Archaeological Science, 34, no. 2, 233–242, https://doi. data: The Borre Monitoring Project: Remote Sensing, 14, https://doi.
org/10.1016/j.jas.2006.04.010. org/10.3390/rs14143289.

January 2024 The Leading Edge 35

 2024
Selim, E. I., A. A. Basheer, G. Elqady, and M. A. Hafez, 2014, Shallow
seismic refraction, two-dimensional electrical resistivity imaging,
and ground penetrating radar for imaging the ancient monuments
at the western shore of Old Luxor City, Egypt: Archaeological
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Discovery, 2, no. 2, 31–43, https://doi.org/10.4236/ad.2014.22005.

Sosson, M., Y. Rolland, C. Müller, T. Danelian, R. Melkonyan, S. Kekelia,
S. Adamia, et al., 2010, Subductions, obduction and collision in the
Lesser Caucasus (Armenia, Azerbaijan, Georgia), new insights, in
Asia-Pacific Events
M. Sosson, N. Kaymakci, R. A. Stephenson, F. Bergerat, V. Starostenko,
eds., Sedimentary basin tectonics from the Black Sea and Caucasus
to the Arabian Platform, https://doi.org/10.1144/SP340.14. 27 Feb–1 Mar – Kuala Lumpur, Malaysia
Steeples, D. W., 2000, A review of shallow seismic methods: Annali di Offshore Technology Conference Asia (OTC Asia)
Geofisica, 43, no. 6, 1021–1044.
Tóth, C. A., Z. Prónay, M. Braun, P. Nagy, M. Pethe, P. Tildy, B. Buró, 19–21 Mar – Mumbai, India
T. Kertész, R. W. McIntosh, and M. Molnár, 2018, Geoarchaeological Role of Geosciences in Carbon Storage
study of szálka and vajda kurgans (Great Hungarian Plain) based on
radiocarbon and geophysical analyses: Radiocarbon, 60, no. 5, 7–9 Apr – Chengdu, China
1425–1437, https://doi.org/10.1017/RDC.2018.102. 1st Tarim Ultra-Deep Oil and Gas Exploration
Tsokas, G. N., P. I. Tsourlos, G. Vargemezis, and M. Novack, 2008, Technology Workshop
Non-destructive electrical resistivity tomography for indoor investiga-
tion: The case of Kapnikarea Church in Athens: Archaeological 23–24 Apr – Kuala Lumpur, Malaysia
Prospection, 15, no. 1, 47–61, https://doi.org/10.1002/arp.321. Borehole Geophysics:
Tsokas, G. N., P. I. Tsourlos, J.-H. Kim, M.-J. Yi, G. Vargemezis, New Developments for Global Energy
M. Lefantzis, E. Fikos, and K. Peristeri, 2018, ERT imaging of the
interior of the huge tumulus of Kastas in Amphipolis (northern 19–22 May – Shenzen, China
Greece): Archaeological Prospection, 25, no. 4, 347–361, https://doi. GEM 2024 Shenzhen: International Workshop
org/10.1002/arp.1718.
DOI:10.1190/leedff.2024.43.issue-1

on Gravity, Electrical, and Magnetic Methods

Wynn, J. C., 1986, A review of geophysical methods used in archaeology:
Geoarchaeology, 1, no. 3, 245–257, https://doi.org/10.1002/ and Their Applications
gea.3340010302. Current information as of 16 Nov 2023. Please check
seg.org/events for up-to-date information and more events.

STUDENT DEADLINES

SEG/Chevron
Student Student Scholarships
Field Leadership Chapter For the
Camps Symposium Reports 2024–2025 year
Application deadline: Application deadline: Deadline: 31 Jan Application deadline:
1 Feb 19 Feb for ranking 1 Mar

seg.org/programs/student-programs

36 The Leading Edge January 2024

 Characterization of anisotropy in basin-scale subsurface
using teleseismic receiver function analysis
Yiran Li1 and Alex Nikulin1
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

https://doi.org/10.1190/tle43010037.1

Abstract Victor et al., 2020). The intended applications will not only
Teleseismic receiver function (RF) analysis offers a passive- contribute to petroleum exploration but also may benefit other
source analogue to detect impedance boundaries using the con- aspects of the current energy landscape that rely on knowledge
verted body waves generated by earthquakes. While the technique of subsurface seismic properties, such as geothermal and CO 2
traditionally has targeted deep earth structures such as the Moho storage operations.
and transition zones, there is growing interest in assessing its Another passive-source seismic technique that has been
applicability in basin-scale seismic characterization, ultimately advocated for industry integration, and that is perhaps the most
aimed for onshore commercial integration as a cost-effective promising analogue to conventional multichannel reflection imag-
complement to existing active-source seismic surveys. Specifically, ing, is teleseismic receiver function (RF) analysis. Teleseismic RF
the conventional broadband seismometers used in global obser- analysis was first employed in the 1960s to extract upgoing con-
vational seismic experiments are not only logistically simple in verted body waves originating from acoustic impedance boundaries
terms of data acquisition, but they also record ground motions in within the crust and upper mantle (Phinney, 1964; Langston,
three mutually orthogonal time series, enabling effective detection 1977). Following those seminal works, various flavors of processing
of shear waves and directional variations of observed signals. philosophy and algorithms were developed to accommodate
Here, we perform teleseismic RF analysis to detect shear-wave observational requirements for structures of different scales and
anisotropy and related symmetry axes orientations in a basin depths. Traditional targets include the continental Moho (e.g.,
DOI:10.1190/leedff.2024.43.issue-1

setting, using open-source seismic data recorded at 55 closely Li et al., 2018), subduction zones (e.g., Nikulin et al., 2019),
spaced seismic stations in the LaBarge Passive Seismic Experiment lithosphere-asthenosphere boundaries (e.g., Li et al., 2021), and
deployed in Wyoming between November 2008 and June 2009. transition zones (e.g., Lawrence and Shearer, 2006). Only recently
We find that the strengths and geometry of the observed anisotropy have they been evaluated for applications in basin settings (e.g.,
are variable along the array. Significantly, not only can anisotropy Leahy et al., 2012) in light of potential adaptation in industry-
effectively delineate subsurface fault interfaces, it can also sub- related subsurface exploration. While consensus is varied on the
stantiate and reveal additional interpretable signals that are recoverability of shallow subsurface structures through this
otherwise disregarded. The estimated fast axes orientations com- method, the majority of previous works have focused on character-
pare favorably with the complex fracture systems documented in izing the base of sedimentary basins at regional scales (Agrawal
the region. Finally, we show that P-to-S amplitude variations et al., 2022; Homman et al., 2022); relatively few have attempted
with P incidence are systematic and modelable using existing to characterize the seismic properties within the sedimentary
computational tools, offering an opportunity to develop an analysis interval. Several factors may contribute to this underutilization,
technique similar to amplitude variation with offset with the including the difficulty of associating a specific seismic interface
products of RF analysis. to those that are known to exist in sedimentary intervals, inevitably
due to the poor vertical resolution (on the order of a few hundred
Introduction meters) compared to the standards achieved by active-source
Passive-source seismic imaging techniques have garnered seismic surveys (e.g., Subašić et al., 2019).
interest from the energy sector in recent decades as an alternate Despite the limitations, one way in which teleseismic RF
mode of subsurface characterization that may mitigate both analysis could add value to the current capabilities of single-
cost-related and logistical limitations suffered by standard active- component seismic surveys is through its ability to effectively
source seismic surveys. Several initiatives in both public and characterize shear-wave anisotropy. Typical teleseismic broadband
private sectors have funded experiments to assess the applicability instrumentations utilize three mutually orthogonal sensors that
of existing passive-source techniques for the purpose of char- can record ground motion in 3D, enabling reliable observations
acterizing the shallow subsurface, with hopes of retrieving of both P and S waves as well as anisotropy-related shear-wave
salient geophysical constraints at minimal added cost. For birefringence. Specifically, the presence of anisotropy generates
example, ambient noise (up to 10 Hz) recorded on terrestrial a characteristic back azimuthal dependence in both the amplitude
broadband seismometers has successfully constructed 3D VS and the polarity of the observed RF signals (Shiomi and Park,
models at basin scale through seismic interferometry techniques 2008), which have been widely observed in various tectonic
(e.g., Curtis et al., 2006) in various energy exploration contexts context, ranging from those associated with the serpentinization
(e.g., Greater Geneva Basin, Planès et al., 2020; Parnaiba Basin, of plate contact in subduction zones (Nikulin et al., 2009) to

Manuscript received 2 June 2023; revision received 30 August 2023; accepted 10 October 2023.
1
Binghamton University, Department of Earth Sciences, Binghamton, New York, USA. E-mail: yli426@binghamton.edu; anikulin@binghamton.edu.

January 2024 The Leading Edge 37

 crustal flow within orogenic plateaus (Xie et al., 2020). The Methodology
theoretical framework for predicting the characteristic amplitude RF time series are estimated by deconvolving the vertical
patterns has been derived assuming transverse isotropy of various component seismogram from the horizontal component seismo-
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

(vertical, tilted, or horizontal) symmetry axes orientations (Levin grams. Because the vertical component seismogram is presumed
and Park, 1997), with forward-modeling tools (e.g., Chen et al., to contain the signatures of seismic source and the teleseismic
2021) that make it possible for the observed patterns to be linked path that waves follow, deconvolving it from the horizontal com-
to rock properties of interest in the sedimentary context, such as ponent seismograms detangles those effects and highlights the
fracture orientations. With both existing predictive and obser- energy that originates from the earth beneath the receiver, such
vational frameworks, teleseismic RF analysis holds potential to as P-to-S conversions generated by impedance boundaries, as well
offer anisotropic constraints as a first-look survey technique in as their associated multiples (Figure 2a). Among various techniques
frontier exploration scenarios where three-component seismom- developed for RF analysis (e.g., Ammon, 1991; Ligorría and
eters are available. Ammon, 1999), we use the multitaper spectral correlation (MTC)
In this study, we characterize anisotropy within the sedimen- method of Park and Levin (2000) to estimate the RFs. The MTC
tary basin by utilizing open-source broadband seismic data col- technique is particularly useful for high-frequency analysis because
lected by the LaBarge Passive Seismic Experiment (LPSE) in its spectral estimation is relatively leakage-free, and it has means
Wyoming, deployed between November 2008 and June 2009 as to distinguish noise and weigh the individual RF time series prior
a part of an experimental initiative funded by ExxonMobil to to stacking. The technique thus allows for reliable usage of high-
assess the potential of ambient noise and teleseismic RF techniques frequency signals where contamination by noise is a concern.
for characterizing shallow subsurface (Saltzer et al., 2011), par- In an ideal case, RF analysis assumes a 1D horizontally
ticularly the seismic velocity constraints from low-frequency stratified earth, where the polarization of converted S waves from
(5–10 Hz) components that are lacking in conventional active- near-vertical incidence are mostly captured in the horizontal
source data. The project has demonstrated the capacity of RF component seismograms oriented parallel to the ground.
techniques to delineate salient sedimentary interfaces beneath Furthermore, the P-to-S phase is expected to reside in the plane
DOI:10.1190/leedff.2024.43.issue-1

the seismic array by employing closely (250 m) spaced interstation that contains the raypath, thus restricted in R-component seis-
intervals that allowed overlapping lateral sampling regions. mograms (Figure 2b). In the presence of anisotropy or dipped
Specifically, strong impedance (greater than 20% change in boundary, however, the scattering of the converted phase will
velocity) associated with local lithostratigraphy and subsurface distribute some energy to the T component in the plane orthogonal
structure has enabled detection of a known fault interface, as well to the raypath. The characteristic directional pattern in the ampli-
as the broad transition from the overlying sedimentary units to tude and polarity is thus seen in the T component, controlled by
the Precambrian granitic basement. Anisotropy beneath the array, the orientation and the dip of the anisotropy symmetry axes.
as observable from the transverse (T) component RF time series Specifically, the T-component amplitude and polarity can vary
(orthogonal to the radial [R] component parallel to the ground), with either a two-lobed or a four-lobed pattern (Figure 3).
has been briefly reported by Subašić et al. (2019) but not character- Assuming a transversely isotropic system, a horizontal symmetry
ized in detail along the array. In this study, we demonstrate axis yields a four-lobed pattern, while a dipped symmetry axis
effective delineation of fault interface through anisotropy. We results in a mix of two- and four-lobed patterns (Park and Levin,
also show the systematic P-to-S amplitude variation as related to 2016). R-component amplitudes experience variation with similar
P incidence, with hopes of further unlocking the potential of this patterns but are directionally shifted by 90° or 45°.
teleseismic technique for onshore commercial integration. The relative contributions of isotropic and anisotropic compo-
nents in the observed RFs can be quanti-
fied by fitting the directional patterns
of R and T amplitudes to a scaled sum
of harmonic functions cos(k * BAZ) and
sin(k * BAZ), where BAZ is the direc-
tion from which the waves arrive at a
site; k = 0 corresponds to the isotropic
component, and k = 1, 2 corresponds to
two- and four-lobed anisotropic com-
ponents, respectively. We use the har-
monic decomposition analysis by Park
and Levin (2016) to perform the fitting
operation simultaneously for both R and
T RF spectra. Both R and T components
have spectral portions that do not follow
Figure 1. Summary of broadband seismic data used in this study. (a) The distribution of broadband seismic stations in the LPSE the lobed directional patterns. This may
in relation to local topography and relevant geologic features known in the study area (Ryan et al., 2009). (b) A representative be due to factors such as 3D velocity
distribution of seismic events used in this study, based on those used at site L28. heterogeneities and noise in the seismic

38 The Leading Edge January 2024

 record. Such spectral portions are assigned to the “unmodeled” approximately 4000 event-station pairs, with approximately 60
components for the purpose of assessing the reliability of interpreted events on average per site. In addition to clearly visible RF signals,
signals (Figure 2f). If both two- and four-lobed amplitudes are our directional coverage is near complete with overlapping (50%)
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

low, the signal does not show directional dependence, and is thus stacking bins, enabling detailed characterization of variation in
considered isotropic. Here, we evaluate the presence of either type anisotropy across the array.
of lobed amplitude pattern by examining the relative strength of
modeled and unmodeled signals as given by the vector lengths of Results
those harmonic components (e.g., Xie et al., 2020). For the two- The limited vertical resolution due to the low teleseismic fre-
lobed signals, we also estimate apparent symmetry axes orientations quency means that the observed RF signals likely reflect broad
using the approach taken in Olugboji and Park (2016) to compare lithostratigraphic geometry marked by boundaries with significant
with the known regional fracture orientations. contrast in seismic properties. The inspection of frequency depen-
To reduce signal variabilities associated with different seismic dence in this study (Figure 4) suggests that most sites do not show
sources, we created a template list of clear P arrivals (MW 5–7.3) significant changes in waveforms beyond the Fmax of approximately
based on site L28 (Figure 1a) and retrieved the same event sets 10 Hz (corresponding to lower-frequency cutoff of 5 Hz due to
for all sites along the LPSE array. The P arrivals were then tapering), in agreement with the relative estimates of teleseismic
automatically determined and picked based on a reference global signal and noise spectra reported in Leahy et al. (2012). In particular,
model (IASP91, Kennett and Engdahl, 1991) to ensure consistent the first-order features emerge by Fmax of 5 Hz and maintain their
preprocessing procedures for all sites. The events were curated general characteristics through higher-frequency cutoffs. Beyond
while ensuring reasonable directional and epicentral coverages to Fmax of 10 Hz, some signal waveforms show additional complexifica-
effectively capture signal patterns associated with anisotropy and tion in manners shared between adjacent sites (e.g., sites L25 and
variations in incidence angle (Figure 1b). In total, we retrieved L26), possibly indicative of finer interfaces that may be further
DOI:10.1190/leedff.2024.43.issue-1

Figure 2. Schematics showing key components of RF workflow and data products used in this study. (a) Teleseismic P-to-S conversion across an impedance boundary. (b) Relationship between
ENZ and RTZ coordinate systems relative to the orientation of S-wave incidence. (c) Example of a single RF trace showing primary P arrival (0 s delay) and later signals with polarity corresponding
to positive (red) and negative (blue) impedance contrast. (d) Example of RF signals and their multiples in R and T epicentral gathers from a simple isotropic subsurface model, calculated at Fmax
of 10 Hz. (e) Example of R and T back azimuthal gathers based on the same isotropic subsurface model used in (e). (f) Example of harmonic time series, with different time series representing
components of overall RF signal explainable by isotropic impedance, anisotropy, as well as noise or small-scale heterogeneities.

January 2024 The Leading Edge 39

 anisotropy-induced P-to-S amplitude
variations with P incidence, which can
significantly enhance the applicability of
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

teleseismic RF analysis targeted at sedi-

mentary environment. With all observ-
ables combined, here we can observe
complex, geologically significant signals
despite the limited frequency range.
Because the constant harmonics
(Figures 5a and 5d) of our RF analysis
represent the proportion of signal ampli-
tudes that do not vary with incidence
direction, their characteristics should
roughly resemble those of the
R-component time series described in
Leahy et al. (2012). Despite different
data set and quality control standards
employed in this study, signal charac-
teristics and spatial geometry that we
observe in the first 2 s of P-to-S delay
are consistent with the reports of Leahy
et al. (2012). Specifically, we find both
the negative (red) signal to the west of
DOI:10.1190/leedff.2024.43.issue-1

the array (signal 1, Figure 5), interpreted

to be the interface between the Paleozoic
Figure 3. Synthetic BAZ gathers showing the directional polarity and amplitude patterns expected for different anisotropic scenarios, carbonate overthrust and the underlying
calculated at Fmax of 6 Hz. Panels (a) and (b) assume a single-layer model with impedance decreasing from the top anisotropic Mesozoic siliciclastic, as well as the
layer to the bottom isotropic half-space. (a) Two-lobe amplitude pattern, resulting from dipped symmetry axes, oriented north-
northeast–south-southwest (30° and 210° BAZ). (b) Four-lobe amplitude pattern, resulting from horizontal symmetry axes, oriented
positive (blue) signal found at all sites
north-northeast–south-southwest (30° and 210° BAZ). The black arrows highlight the directions where signal polarity change. along the array (between 0.6 and 1.0 s
delay), interpreted to be the Paleozoic
carbonate overlying the Precambrian basement. We also observe
similar frequency response associated with those signals, including
the lack of significant reduction in the signal width beyond Fmax
of 10 Hz.
Furthermore, we note systematic variation in the signal
amplitudes and waveform associated with the observed signals.
For the negative signal (signal 1, Figure 5) in particular, the
amplitudes observed in our constant harmonics are particularly
strong between stations L07 and L15 (segment B, Figure 5), and
significantly weakens elsewhere, toward the proposed surficial
trace of Hogsback Fault (stations L16–L22, segment C, Figure 5),
as well as toward the western end of the array (L01–L06,
segment A, Figure 5). This distinct amplitude pattern is observed
Figure 4. The constant harmonic components for sites L23–L33, showing progression in the resulting regardless of varying Fmax , suggesting lateral variation in subsurface
waveforms with increasing analysis signal content. The time series colors correspond to Fmax values, seismic properties across the fault interface or within the overthrust
from 2 (red) to 11 Hz (green). carbonate itself. The waveforms of the deeper positive signal also
show systematic lateral variations and frequency response relative
isolated using more sophisticated filtering schemes. Previous for- to the surficial Hogsback Fault trace. Specifically, the signals near
ward-modeling studies that have examined the vertical resolution the fault trace (L16–L22) generally show weaker amplitudes and
of teleseismic RF analysis have also shown that gradual impedance more broad and complex waveforms compared to those observed
transitions, such as those produced by burial compaction, can result at stations elsewhere.
in signals with characteristic wavelengths that do not decrease Directionally variant signal components, as seen from col-
regardless of increased analysis frequency (Levin et al., 2016). To lective amplitudes of two-lobed (Figures 5b and 5e) and four-lobed
supplement the limited resolution of the teleseismic RF technique, amplitude patterns (Figures 5c and 5f), are equally as complex,
we demonstrate here that considerable subsurface properties can showing both continuous and localized anisotropic presence across
be inferred by examining additional signal characteristics, such as both transects. As demonstrated in Figure 3, we find that two-lobe

40 The Leading Edge January 2024

 amplitudes are generally stronger than four-lobe amplitudes across While forward modeling is a preferred approach to constrain
the array, consistent with the dipped seismic structures expected the precise geometry of the observed anisotropy, inspection of
in the compressional tectonic environment of the study area. For amplitude and polarity patterns in back azimuthal gathers offers
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

both transects A and B, the overall spatial distribution of anisot- a first-order assessment of potential variations in anisotropy. In
ropy is concentrated to the west of the array, in the hanging wall particular, the orientation of symmetry axes holds systematic
side of the Hogsback Fault. In fact, the negative signal observed relationship with the incidence directions where the reversals in
in the constant harmonics is associated with a strong, continuous signal polarity are observed. Based on the broad differences in
two-lobe signal that terminates exactly near the proposed surficial directional signal patterns, we find variable anisotropy at different
trace of the Hogsback Fault (Figures 5b and 5e). While signal locations and depths across the array, even along a seemingly
contributions from the dipped interface are likely, locally observed continuous anisotropic signal. In particular, the two-lobe signal
coincident four-lobe amplitude (bottom row panels, Figure 6) associated with the carbonate overthrust shows systematic variation
also suggests anisotropic contributions. Furthermore, the distribu- of anisotropy that can be broadly distinguished into three segments
tion of anisotropy is not only limited to the proximity of Hogsback based on salient characteristics in directional signal patterns
Fault. We also observe a deeper two-lobe signal that coincides (Figure 6). In segment A, the main reversals in polarity are
with the top of underlying Paleozoic carbonates (signal 2, Figure 5), observed at back azimuths of 90° and 210° in the T-component
as well as an even deeper lobed signal that is likely associated with gather, and the associated negative (red) signal in the R-component
the Precambrian crystalline basement (signal 3, Figure 5). In gather are observed in back azimuthal range 165°–345° (Figure 6a).
contrast to signal 1, signals 2 and 3 show stronger four-lobe The T-component gather in segment B shows polarity reversals
amplitude variations in the observed signal, suggesting a different at similar back azimuths. However, the negative R-component
sense of anisotropy than those associated with the shallower signals are observed between wider back azimuthal range 90°–345°,
Hogsback Fault. The overall magnitude of anisotropy weakens with distinctly stronger amplitude in the two-lobe harmonic
significantly toward the center of the array and strengthens again component (Figure 6b). Segment C is distinctly different from
to the east. Critically, the presence of anisotropy also enables both segments A and B, showing much weaker two-lobe harmonic
DOI:10.1190/leedff.2024.43.issue-1

identification of additional seismic structures that are otherwise amplitude and different T-component amplitude patterns
left hidden. For example, we note a potential west-dipping struc- (Figure 6c). The estimated fast axes orientations are generally
ture associated with the positive (blue) signal in constant harmonic consistent along the Hogsback Fault interface, mostly orienting
(signal 4, Figure 5) based on a coincident two-lobe signal observed northeast to southwest, except for a few sites closest to the fault
between sites L25 and L33. (L42 and L43) that show north to south orientation (Figure 8).

Figure 5. (a) and (d) Constant, (b) and (e) two-lobe, and (c) and (f) four-lobe harmonic components along transects A and B, calculated at Fmax of 6 Hz. The anisotropic amplitudes in panels (b), (e),
(c), and (f) reflect the difference between corresponding modeled and unmodeled amplitudes in two-and four-lobe vector lengths, in order to indicate presence and absence of anisotropy along the
transects. Annotated RF signals and observations are elaborated in the text. Red inverted triangles indicate the proposed locations of Hogsback thrust fault trace along each transect.

January 2024 The Leading Edge 41

 Finally, we also show that P-to-S amplitude varies systemati- that subsurface seismic properties are homogenous within the
cally with P incidence and can be modeled using synthetic gathers sampling width of the incidence rays. Furthermore, because the
generated by available reflectivity algorithms (Figure 7). Epicentral directional variation of signal amplitudes are expected in aniso-
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

gathers routinely used in RF analysis are equivalent to common- tropic scenarios, only events from limited BAZ range should be
midpoint gathers in exploration seismology, combining traces of used to observe the amplitude variation due to P incidence
varying P incidence. Because the locations of conversion vary (Figure 7). Reassuringly, our epicentral gathers show systematic
depending on the epicentral distance (Figure 7c), effective observa- variations in P-to-S amplitudes that can be effectively simulated
tion of the P-to-S amplitude variation must rely on the assumption using an anisotropic model while an isotropic model fails to do
so, further stressing the importance of
integrating anisotropy into subsurface
interpretations (Figure 7d). While
detailed modeling is warranted to better
understand the effects of both VS imped-
ance and anisotropy on the resulting
synthetic P-to-S amplitude curve, we
note that our model in Figure 7d only
utilizes a 15% decrease in isotropic VS
from the overlying carbonates to the
siliciclastic beneath and relatively strong
shear-wave anisotropy (40%) to simulate
the carbonate in the top layer. We also
find that most of the curvature and
radial amplitudes can be fitted by simply
DOI:10.1190/leedff.2024.43.issue-1

altering the strength of anisotropy.

Discussion
The harmonic decomposition of
teleseismic RF analysis has successfully
constrained anisotropy in various basin
exploration contexts, with observed
Figure 6. Examples of BAZ gathers showing characteristic directional dependence of T-component amplitude and polarity. The anisotropy linked to critical subsurface
sites are located at different surface locations along the Paleozoic carbonate overthrust, showing varying signal patterns as properties such as porosity and fracture
well as overall anisotropic strengths relative to the Hogsback thrust fault. orientations (e.g., Licciardi and Piana
Agostinetti, 2017). As such, the observed
anisotropy beneath LPSE can be further
compared with known structural char-
acteristics in the region to infer potential
fracture patterns or spatial variations in
velocity. While forward modeling was
not carried out here, the strong two-lobe
signals indicate that most of the observed
anisotropy can be associated to a trans-
verse isotropic system with tilted sym-
metry axes, potentially explainable by
a combination of physical factors such
as dipped impedance interface associ-
ated with the Hogsback thrust, the
sheared material along the fault if any,
as well as the fracture systems known
in the area. In addition, the lateral
variation in the strength of both iso-
Figure 7. Demonstration of variation in P-to-S amplitude as a function of P incidence, determined by the epicentral distances of seismic tropic and anisotropic amplitudes
events. (a) Source distribution (top) and radial epicentral gather (bottom) for site L04, calculated at Fmax of 6 Hz with events originating along signal 1 is also intriguing, as it
from BAZ range 135°–150°. (b) Source distribution and radial epicentral gather for L15, calculated using the same delimited BAZ range
as (a). (c) Schematic showing the general relationship between P incidence and source epicentral distance. (d) Observed (circle and suggests complexity across the fault
triangles) and modeled (solid and dashed lines) variations in normalized P-to-S amplitude as a function of epicentral distances. The interface that could stem from com-
amplitudes of negative RF signals highlighted by the green lines in panels (a) and (b) are used to examine the dependence on P incidence. positional heterogeneity or small-scale

42 The Leading Edge January 2024

 anisotropy-causing structures that can reside in either side of
the fault. If the isotropic impedance contrast between the Paleozoic
carbonate and the Mesozoic siliciclastic is presumed to be the same
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

across the fault, it is possible that the apparent variation in constant

harmonic amplitudes along the fault interface may in fact be due
to the variations in the geometry or the strength of the anisotropy.
We also note corresponding lateral variations in the strength of
two-lobe harmonics, with the strongest amplitudes observed in
segment B. The overall signal width is also wider for both constant
and two-lobe harmonics along segment B, potentially indicating
a wider zone of shear deformation compared to those along seg-
ments A and C. With strong spatial correlation to the fault, the
observed anisotropy is also markedly weaker to the east of the
proposed surficial fault trace (Figure 6d), demonstrating that
anisotropic signals are much more reliable at delineating the
geometry of subsurface fault interface than isotropic signals.
The anisotropy beneath LaBarge array is not only observable
from RF analysis but also from azimuthally varying VS in ambient Figure 8. Apparent orientations of tilted anisotropy symmetry axes as estimated from the k = 1
noise tomography that has been constructed using signals of harmonics.
similar frequency range (below 10 Hz), thus proving to be a
nonnegligible seismic characteristic in the underlying sedimentary predominantly north–south-trending J1 fractures near the surficial
interval. In terms of symmetry axes orientations, Subašić et al. fault trace in areas southward from our study area, which is
(2019) noted from the strong sin(BAZ) component harmonics consistent with the axis orientations we observe at stations L42
DOI:10.1190/leedff.2024.43.issue-1

that the predominant anisotropic symmetry axes, likely oriented and L43 (Figure 8). In addition, other notable signals observed
east to west, are consistent with what is expected from the regional in our transects (e.g., signal 4) are also compatible with local
structural trends. Our observed fast axes orientations are locally geologic structures, particularly compared to both west- and
in agreement with this view of Subašić et al. (2019), though we east-dipping faults that have been interpreted from the reflection
also find notable lateral variations in the estimated axes orientations seismic profiles in the region (e.g., Greenhalgh et al., 2015). This
on the hanging wall side of the Hogsback thrust. As noted in the illustrates that the usage of additional signal attributes, such as
results section, the estimated symmetry axes orientations are anisotropy, can help substantiate and reveal additional interpretable
variable but not random, as several adjacent stations share similar structures that would have otherwise remained disregarded.
axes orientations. For example, while the stations closest to the Necessarily, several complications can arise when combining
surficial fault trace are oriented north to south, few sites to the teleseismic RF analysis with other common data sets that are
west (e.g., L18–L20) show distinctly different but identical orienta- utilized for subsurface characterization in industry settings. As
tions bearing approximately 40° from the north. The axes orienta- both passive-source and active-source seismic surveys utilize very
tions along segment B are markedly more variable than both different frequency ranges, the relationship between the observed
segments A and C, which could reflect generally more complex RF signals and the complex velocity structures imaged in refraction
axes geometries along the ramp. It is noteworthy that the approach or reflection profiles would be best established through detailed
by Olugboji and Park (2016) assumes a fast symmetry axes, and forward modeling. For example, while the basin feature that is
additional analysis is warranted to determine whether the patterns most plausibly captured by the RF analysis is the sediment-
in BAZ gather may better correspond to those associated with basement transition, compactions in the basal sedimentary intervals
the tilted slow symmetry axes. Nonetheless, we find that segment B can introduce uncertainties in the estimates of basement depths
exhibits overall different signal characteristics than those from by imitating the velocities of the underlying crystalline crust.
segments A and C. Furthermore, the observed axes orientations Other uncertainties may also be examined and minimized to
are compatible with the fracture orientations documented in the bring the observations closer to the prediction to enable effective
region. Specifically, three sets of fractures (J1, J2, and J3) have forward modeling. In the case of variations in P-to-S amplitudes
been observed and interpreted in the context of both extensional with P incidence, the minor deviations between the observed and
and compressional structural regimes associated with past tectonic predicted anisotropic amplitude trends (Figure 7d) may be due
events in the area (e.g., Lorenz and Laubach, 1994). Northeast– in part to the deviations of the actual raypaths, and by extension
southwest-oriented fractures are noted for both J2 and J3 fracture the incidence angles, from those predicted by a global reference
sets, the origins of which have been linked to the primary Eocene velocity model as done in this study. For future work, more accurate
compressional regime and subsequent mechanical shear to local measure of incidence angles should be performed with ray tracing
stress, respectively. The general axes orientations, as well as the using local subsurface velocity models. Despite many uncertainties,
small-scale variations that we observe here, are compatible with where subsurface structures and sense of impedance change is
the descriptions of fractures in the regions around the Hogsback known with confidence, signals observed in the teleseismic RF
thrust. For example, Lorenz and Laubach (1994) report analysis can offer valuable anisotropic constraints, which can

January 2024 The Leading Edge 43

 inform both geomechanic characterizations and local velocity Seismological Society of America, 67, no. 3, 713–724, https://doi.
models. Furthermore, the upgoing nature of the teleseismic P-to-S org/10.1785/BSSA0670030713.
waves could also complement the resolution loss at the depth Lawrence, J. F., and P. M. Shearer, 2006, A global study of transition
zone thickness using receiver functions: Journal of Geophysical
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

suffered by active-source surveys due to attenuation. All in all,

Research. Solid Earth, 111, no. B6, B06307, https://doi.
given the drastically low cost of survey logistics, instrumentation, org/10.1029/2005JB003973.
and data processing, the utilization of teleseismic RF analysis Leahy, G. M., R. L. Saltzer, and J. Schmedes, 2012, Imaging the
presents a potential economic advantage, especially in frontier shallow crust with teleseismic receiver functions: Geophysical
exploration contexts where first-order characterization of the basin Journal International, 191, no. 2, 627–636, https://doi.
(e.g., basin depth, presence of anisotropy) is needed. org/10.1111/j.1365-246X.2012.05615.x.
Levin, V., J. A. VanTongeren, and A. Servali, 2016, How sharp is the
sharp Archean Moho? Example from eastern Superior Province:
Acknowledgments Geophysical Research Letters, 43, no. 5, 1928–1933, https://doi.
This work was funded by the American Chemical Society PRF org/10.1002/2016GL067729.
grant 59945-DNI8 and The Graduate School of Binghamton Levin, V., and J. Park, 1997, P-SH conversions in a flat-layered medium
University for the first author. The analysis was completed using with anisotropy of arbitrary orientation: Geophysical Journal
Recfunk21 code (https://github.com/ruseismology/recfunk21), International, 131, 253–266, https://doi.org/10.1111/j.1365-
and images were generated using Generic Mapping Tool (Wessel 246X.1997.tb01220.x.
et al., 2013), as well as open-source Python packages. We thank Li, C., H. Gao, M. L. Williams, and V. Levin, 2018, Crustal thickness
variation in the northern Appalachian Mountains: Implications for
Xiaoran Chen and Jeffrey Pietras for helpful discussions. All seismic
the geometry of 3-D tectonic boundaries within the crust:
waveforms analyzed here are accessible through IRIS DMC Geophysical Research Letters, 45, no. 12, 6061–6070, https://doi.
(https://ds.iris.edu/ds/nodes/dmc/). We gratefully acknowledge org/10.1029/2018GL078777.
the three anonymous reviewers for constructive feedback that Li, Y., V. Levin, A. Nikulin, and X. Chen, 2021, Systematic mapping
greatly improved this work, as well as the editorial board for of upper mantle seismic discontinuities beneath Northeastern North
assistance throughout this manuscript preparation process. America: Geochemistry, Geophysics, Geosystems, 22, no. 7,
DOI:10.1190/leedff.2024.43.issue-1

e2021GC009710, https://doi.org/10.1029/2021GC009710.
Licciardi, A., and N. Piana Agostinetti, 2017, Sedimentary basin explora-
Data and materials availability tion with receiver functions: Seismic structure and anisotropy of the
Data associated with this research are available and can be Dublin Basin (Ireland): Geophysics, 82, no. 4, KS41–KS55, https://
accessed via https://ds.iris.edu/ds/nodes/dmc/. doi.org/10.1190/geo2016-0471.1.
Ligorría, P., and C. J. Ammon, 1999, Iterative deconvolution and receiver-
Corresponding author: yli426@binghamton.edu function estimation: Bulletin of the Seismological Society of America,
89, no. 5, 1395–1400, https://doi.org/10.1785/BSSA0890051395.
References Lorenz, J. C., and S. E. Laubach, 1994, Description and interpretation
Agrawal, S., C. M. Eakin, and J. O’Donnell, 2022, Characterizing the of natural fracture patterns in sandstone of the Frontier Formation
cover across South Australia: A simple passive-seismic method for along Hogsback, southwestern Wyoming: Gas Research Institute
estimating sedimentary thickness: Geophysical Journal International, Report, 94-0020.
231, no. 3, 1850–1864, https://doi.org/10.1093/gji/ggac294. Nikulin, A., J. R. Bourke, J. R. Domino, and J. Park, 2019, Tracing
Ammon, C. J., 1991, The isolation of receiver effects from teleseismic geophysical indicators of fluid-induced serpentinization in the Pampean
waveforms: Bulletin of the Seismological Society of America, 81, no. flat slab of Central Chile: Geochemistry, Geophysics, Geosystems,
6, 2504–2510, https://doi.org/10.1785/BSSA0810062504. 20, no. 9, 4408–4425, https://doi.org/10.1029/2019GC008491.
Chen, X., J. Park, and V. Levin, 2021, Anisotropic layering and seismic Nikulin, A., V. Levin, and J. Park, 2009, Receiver function study of the
body waves: Deformation gradients, initial S-polarizations, and Cascadia megathrust: Evidence for localized serpentinization:
converted-wave birefringence: Pure and Applied Geophysics, 178, Geochemistry, Geophysics, Geosystems, 10, no. 7, 2009GC002376,
no. 6, 2001–2023, https://doi.org/10.1007/s00024-021-02755-6. https://doi.org/10.1029/2009GC002376.
Curtis, A., P. Gerstoft, H. Sato, R. Snieder, and K. Wapenaar, 2006, Olugboji, T., and J. Park, 2016, Crustal anisotropy beneath Pacific
Seismic interferometry — Turning noise into signal: The Leading Ocean-Islands from harmonic decomposition of receiver functions:
Edge, 25, no. 9, 1082–1092, https://doi.org/10.1190/1.2349814. Geochemistry, Geophysics, Geosystems, 17, no. 3, 810–832, https://
Greenhalgh, S. R., J. H. McBride, J. M. Bartley, R. W. Keach II, B. B. doi.org/10.1002/2015GC006166.
Britt, and B. J. Kowallis, 2015, Along-strike variability of thrust Park, J., and V. Levin, 2000, Receiver functions from multiple-taper
fault vergence: Interpretation, 3, no. 3, SX1–SX12, https://doi. spectral correlation estimates: Bulletin of the Seismological Society
org/10.1190/INT-2014-0182.1. of America, 90, no. 6, 1507–1520, https://doi.org/10.1785/0119990122.
Homman, K., A. Nyblade, K. Schmid, R. Anthony, and K. Carter, 2022, Park, J., and V. Levin, 2016, Anisotropic shear zones revealed by backa-
Basement structure of the Appalachian Basin in Pennsylvania: zimuthal harmonics of teleseismic receiver functions: Geophysical
Tectonophysics, 837, no. June, 229451, https://doi.org/10.1016/j. Journal International, 207, no. 2, 1216–1243, https://doi.org/10.1093/
tecto.2022.229451. gji/ggw323.
Kennett, B. L. N., and E. R. Engdahl, 1991, Traveltimes for global Phinney, R. A., 1964, Structure of the earth’s crust from spectral behavior
earthquake location and phase identification: Geophysical Journal of long-period body waves: Journal of Geophysical Research, 69, no.
International, 105, no. 2, 429–465, https://doi.org/10.1111/j.1365- 14, 2997–3017, https://doi.org/10.1029/JZ069i014p02997.
246X.1991.tb06724.x. Planès, T., A. Obermann, V. Antunes, and M. Lupi, 2020, Ambient-noise
Langston, C. A., 1977, Corvallis, Oregon, crustal and upper mantle tomography of the Greater Geneva Basin in a geothermal exploration
receiver structure from teleseismic P and S waves: Bulletin of the context: Geophysical Journal International, 220, no. 1, 370–383,
https://doi.org/10.1093/gji/ggz457.

44 The Leading Edge January 2024

 Ryan, W. B., S. M. Carbotte, J. O. Coplan, S. O’Hara, A. Melkonian,
UNDER THE PATRONAGE OF HIS ROYAL HIGHNESS
R. Arko, R. A. Weissel, et al., 2009, Global multi‐resolution topog- PRINCE SAUD BIN NAYEF BIN ABDULAZIZ AL-SAUD,
raphy synthesis: Geochemistry, Geophysics, Geosystems, 10, no. 3, GOVERNOR OF THE EASTERN PROVINCE
Q03014, https://doi.org/10.1029/2008GC002332.
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Saltzer, R., G. Leahy, J. Schmedes, J. Roth, and E. Rumpfhuber, 2011,

Earthquakes — A naturally occurring source of low-frequency data:
81st Annual International Meeting, SEG, Expanded Abstracts,
3689–3692, https://doi.org/10.1190/1.3627967.
Shiomi, K., and J. Park, 2008, Structural features of the subducting slab
beneath the Kii Peninsula, Central Japan: Seismic evidence of slab 12-14 FEBRUARY 2024 | DHAHRAN EXPO | KINGDOM OF SAUDI ARABIA
segmentation, dehydration, and anisotropy: Journal of Geophysical
Research: Solid Earth, 113, no. B10, B10318, https://doi.
inventing solutions > leading the transition
org/10.1029/2007JB005535.
Subašić, S., N. Piana Agostinetti, and C. J. Bean, 2019, Estimating
lateral and vertical resolution in receiver function data for shallow
crust exploration: Geophysical Journal International, 218, no. 3,
2045–2053, https://doi.org/10.1093/gji/ggz262.
Victor, T., J. Julià, N. J. White, and V. Rodríguez-Tribaldos, 2020, Joint 20,000+ 250+ 5+ 500+
inversion of high-frequency receiver functions and surface-wave Energy Exhibitors Executive Technical
dispersion: Case study in the Parnaíba Basin of Northeast Brazil: Professionals Sessions Presentations
Bulletin of the Seismological Society of America, 110, no. 3, 1372–
1386, https://doi.org/10.1785/0120190203.
Wessel, P., W. H. F. Smith, R. Scharroo, J. Luis, and F. Wobbe, 2013,
SECURE YOUR SEAT
Generic mapping tools: Improved version released: Eos, 94, no. 45, REGISTER TODAY
409–410, https://doi.org/10.1002/2013EO450001. go.iptcnet.org/register_now
Xie, Z., V. Levin, and Q. Wu, 2020, Crustal anisotropy beneath north-
eastern Tibetan Plateau from the harmonic decomposition of receiver
DOI:10.1190/leedff.2024.43.issue-1

functions: Geophysical Journal International, 220, no. 3, 1585–1603, sponsoring societies

https://doi.org/10.1093/gji/ggz526.

7–9 Feb – Houston, TX, USA 30 Apr–2 May – Virtual

NAPE Summit Week Summit on Geophysical Detection of
Registration open Explosive Remnants of War: Solving

20
25–28 Feb – Vancouver, British Current Challenges of Unexploded
Columbia, Canada Ordnance (UXO) and Demining
ASCE Geo-Congress 2024 Call for abstracts closes 15 Feb
Registration open 6–9 May – Houston, TX, USA
Offshore Technology Conference (OTC)

24
7 Mar –Denver, CO, USA
Denver Geophysical Society’s Registration opens mid-Jan
Annual 3D Seismic Symposium 4–5 Jun – Fort Worth, TX, USA
Registration will open in Jan SPE/SEG Workshop: Injection
11–13 Mar – Houston, TX, USA Induced Seismicity
Carbon, Capture, Utilization, and Early bird registration closes 7 May
Storage (CCUS)

Events
17–19 Jun – Houston, TX, USA
Registration open Unconventional Resources
27–29 Mar – Norman, OK, USA Technology Conference (URTeC)

in the
SEG/SPE/SPWLA Workshop: From Registration opens soon
Measurement to Theory: Adventures
24–27 Jun – Virtual
in Rock Physics, Petrophysics, and

Americas
Net-Zero Emissions Workshop
Engineering
Call for abstracts opens 19 Jan
Early bird registration closes 1 Feb
14–19 Jul – Golden, CO, USA
Geophysical Research for
Current information as of 13 Dec 2023. Gigatonnes CO2 Storage Workshop
Please check seg.org/events for more
Call for abstracts closes 29 Feb
information and more events.

January 2024 The Leading Edge 45

 A decade of technology advancement in seismic processing:
A case study from reprocessing legacy sparse OBC data
in Kashagan Field
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Jaewoo Park1, Craig Hyslop1, Ares Ouzounis1, Alecia Wawrzynski1, Tetyana Vdovina1, Sunwoong Lee1, Katja Hasner 2, William D. Ibanez2,
Steve Schreuder3, Zharas Tapalov 3, Assembek Gabdullin3, and Nurzhan Yergaliyev 3
https://doi.org/10.1190/tle43010046.1

Abstract fold with 25 × 25 m common-depth-point bin). Among several

Ceaseless advancements in seismic data processing technology vintages of legacy seismic processing for this sparse OBC data,
and workflow enable the extraction of increasing amounts of the 2010–2011 reprocessing work has been used as a benchmark
subsurface information essential for derisking and optimizing due to the extensive effort involved and the inclusion of the latest
natural resource development. The Kashagan oil field, one of the technologies at that time, including denoise, complex velocity
world’s largest carbonate reservoirs, is faced with significant devel- model building (VMB), and depth imaging (e.g., Kirchhoff
opment optimization challenges due to a combination of complex migration, wave equation migration). Although an image improve-
geology and suboptimal seismic data coverage. The latest processing ment was quite evident in the final seismic image, the challenges
technologies were applied to the 2001–2002 vintage sparse ocean- from the complex geology and sparse OBC data persisted, resulting
bottom cable data and produced step-change improvements over in nongeologic geometries of parts of the carbonate platform and
the 2010–2011 legacy processing and imaging results. Such sig- poor image focusing and reflector continuity at the reservoir level
nificant image uplift is representative of technology advancements (Figure 1).
over the span of a decade fully leveraging the newly preprocessed Complex geology at the Kashagan Field poses extreme chal-
DOI:10.1190/leedff.2024.43.issue-1

input data, including (1) an integrated model building workflow, lenges even for the most modern seismic processing technology.
(2) adaptive application of full-waveform inversion, (3) geologically Shallow water (approximately 4–7 m) creates significant multiple
constrained reflection tomography, and (4) least-squares imaging. noise, reverberating immediately from the shot record start time
The new seismic results improved the structural and stratigraphic (Figure 1) (Anderson et al., 2022). There are two distinct layers
imaging of multiple layers of shallow carbonates, mitigated fault of carbonate rock with sharp velocity contrast against the sur-
shadow and other complex overburden effects, increased the resolu- rounding clastics: the Cretaceous-age carbonate immediately
tion of subsalt carbonate reservoir features such as karst and internal below water bottom, and the Jurassic-age carbonate below
fractures, improved well marker depth misfit, and ultimately Cretaceous clastic sediments. These carbonate layers, apart from
influenced the placement of upcoming wells and reservoir develop- generating strong peg-leg multiples, are also displaced by numerous
ment plans. Further improvements in seismic imaging would be large faults with associated fault shadows and poor image focusing
feasible with a modern acquisition of more densely sampled seismic below. Beneath the Jurassic carbonate, there are salt diapirs of
data, which would allow the full potential of the latest seismic various sizes and geometries, in some cases almost as shallow as
processing technologies and workflows to be unlocked. the seafloor. These salt structures are interspersed with minibasins
of Cambrian-age clastic sediments, the velocities of which can
Introduction be faster than that of the adjacent salt. The target reservoirs are
The Kashagan Field is one of the largest giant oil fields dis- situated below this complex overburden of salt and minibasins
covered in modern exploration history (Pettingill, 2001). The field and contain various carbonate platform facies — a layered flat
is located in the northeast part of the Caspian Sea, south of Atyaru platform interior, a less compacted rim facies, and slope facies
in Kazakhstan (Figure 1). After its initial discovery in 2000, only — all of which display various degrees of nonmatrix features:
a few 3D seismic data acquisition surveys were conducted due to karst and fractures (Figure 1).
severe environmental and regulatory challenges (e.g., the area is The impact of these geologic challenges on the imaging is
ice bound for five months annually, waters are ultra-shallow, and aggravated by the relatively outdated sparse OBC data, which
zero-discharge regulations are in place offshore). The most exten- were acquired with a cross-spread layout of shot and receiver lines,
sive seismic survey that exists over the field was acquired in orthogonally intersecting each other. The maximum absolute
2001–2002 using 3D two-component (3D2C) ocean-bottom cable offset is 4500 and 3000–5400 m in inline and crossline directions,
(OBC) with sparse acquisition geometry and low fold (60 nominal respectively. The sparse source and receiver line spacings are 450

Manuscript received 15 September 2023; accepted 9 November 2023.

1
ExxonMobil Technology and Engineering Company, Houston, Texas, USA. E-mail: jaewoo.park@exxonmobil.com; craig.hyslop@exxonmobil.com;
ares.ouzounis@exxonmobil.com; alecia.wawrzynski@exxonmobil.com; tetyana.vdovina@exxonmobil.com; sunwoong.lee@exxonmobil.com.
2
ExxonMobil Upstream Company, Houston, Texas, USA. E-mail: katja.hasner@exxonmobil.com; william.d.ibanez@exxonmobil.com.
3
North Caspian Operating Company, Atyrau, Kazakhstan. E-mail: steve.r.schreuder@gmail.com; zharas.tapalov@ncoc.kz; assembek.gabdullin@ncoc.kz;
nurzhan.yergaliyev@ncoc.kz.

46 The Leading Edge January 2024

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 1. Legacy Kirchhoff PSDM image overlain by the corresponding velocity model.
DOI:10.1190/leedff.2024.43.issue-1

and 600 m, resulting in uneven offset distribution and missing to address geophysical challenges from the shallow to the deep.
near-offset traces, essential for data regularization and denoise Third, we show a joint tomography example to demonstrate how
work. The acquisition footprint due to the relatively low fold is to leverage geologic understanding into the model building process,
sufficiently strong to imprint on the primary reflection signal particularly in areas less constrained by seismic data only. Finally,
down to the reservoir level. The data gap from the sparse acquisition we explain how least-squares (LS) imaging in both Kirchhoff
is especially severe at shallow depths, creating significant chal- migration and reverse time migration (RTM) helps further
lenges in mitigating multiples generated from the shallow water enhance image fidelity affected by sparse acquisition, uneven
and carbonate layers (Anderson et al., 2022). illumination, and migration swing.
Recognizing the need for an improvement over the 2010–2011
legacy seismic image, the North Caspian Operating Company Integrated model building workflow
(NCOC) in 2019 initiated a reprocessing effort of the 3D2C The project timeline of Kashagan West spanned from early
full-field seismic survey, and awarded ExxonMobil a contract to 2021 to mid 2022, amid the COVID-19 pandemic. Throughout
lead the in-house VMB and imaging effort, in collaboration with the project, weekly technical updates and monthly partner meet-
preprocessing work done by PGSK. The work was performed in ings were held to monitor and review progress and to ensure
two stages covering two halves of the field: the Kashagan East technical alignment among the various stakeholders located in
project was completed successfully in 2020, demonstrating a several different countries around the globe. Online meeting tools
step-change improvement in the final seismic image and leading (Zoom, Microsoft Teams) enabled us to overcome physical barriers
to the Kashagan West project being initiated to achieve the same and collaborate without any pandemic-related restrictions. Both
quality of improvement. Given that the previous field-wide pro- virtual and physical collaboration among the integrated teams of
cessing and imaging effort used the same 3D2C OBC raw input geophysicists and geologists at ExxonMobil, NCOC, and the
data, the uplift in the latest image with respect to the decade-old preprocessing team at PGSK was the key element to integration
legacy result illustrates the advancement in seismic processing of the various data sets (surface seismic, vertical seismic profile,
technology and workflows, providing an insight to how much well log), geologic interpretations, and insights into the VMB
additional subsurface information can be garnered without acquir- process. For example, the velocity and anisotropic models for the
ing a new and expensive seismic survey. Jurassic carbonate layer and related faults started from the fit-for-
Among the many updated technologies and workflows applied purpose interpretation of both the top and base of the carbonate
in this project, this paper focuses on the advancement of VMB layer based on the depth image, migrated from the smooth back-
and imaging technology, in particular, the four critical elements ground model (Figures 2a and 2b). A representative constant
that contributed significantly to the outcome. First, we emphasize carbonate velocity (approximately 3830 m/s), derived from the
the significance of integrating the processing technologies, work- well data, was injected between the top and base of the Jurassic
flow, and geologic interpretation in a more efficient way. Second, carbonate. Because the correct depth of the hard carbonate bound-
we highlight the high-resolution model update, mostly driven by ary was constrained only at the sparse well locations, a moderate
variable types of full-waveform inversion (FWI), which is adapted model smoothing was applied to generate the starting model for

January 2024 The Leading Edge 47

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms
DOI:10.1190/leedff.2024.43.issue-1

Figure 2. Integrated VMB workflow for the shallow Jurassic carbonate and fault. (a) Kirchhoff PSDM stack overlain by the smooth background velocity. (b) Fit-for-purpose Jurassic
carbonate interpretation. (c) eFWI starting model with smoothed carbonate layer inserted. (d) Updated eFWI velocity and corresponding Kirchhoff PSDM stack. Velocity review was
carried out on the sparsely migrated PSDM stack (a, c, and d), while the fit-for-purpose interpretation was done on the dense stack (b). TUC: top of Upper Cretaceous, BUC: base of Upper
Cretaceous, TUJ: top of Upper Jurassic, BUJ: base of Upper Jurassic, TPT_U: top of Permian-Triassic Unconformity, TOS_Scen1: top of salt scenario 1.

elastic FWI (eFWI) (Figure 2c). The velocity update from eFWI unable to achieve this goal in practice due to multiple unyielding
successfully captured the geometries and velocities of the hard factors: limited usable offset for reflection tomography (e.g., less
geologic boundaries, in particular the base of Cretaceous carbonate, than 2 km offset down to the base of Cretaceous clastics), strong
the top and base of Jurassic carbonate, as well as related faults velocity contrasts and variation beyond the resolving power of
(Figure 2d). A similar workflow was implemented for salt VMB, refraction tomography (e.g., faulted Cretaceous and Jurassic car-
in which salt overhangs were interpreted using a simultaneous bonate layers), and severe multiple noise caused by the shallow
multi-Z horizon picking approach (Mattson et al., 2020), and water depths. Utilizing the FWI via an eFWI technology applica-
further updated by eFWI. This semiautomated VMB workflow tion was the solution to these challenges.
led to a superior seismic image at most places compared with the After the source-receiver reciprocity was applied, approxi-
legacy approach in which the geologic boundary was interpreted mately 40,000 shots were selected with minimum preprocessing
with strenuous effort and multiple iterations of trial and error applied, including swell noise and surface wave attenuation. The
(Figure 1). Moreover, this interpretation-guided inversion-based source wavelet was estimated in frequency domain and adapted
approach does not require the conventional model building practice to account for low signal-to-noise ratio (S/N) in this land-like
(the sequential interpretation from the top to the base of the survey. Considering the strong elastic effects due to the presence
boundary of salt overhang and carbonate layer), significantly of property contrasts around carbonate and salt lithologies, as
reducing the cycle time in the depth model building. well as the steeply dipping geologic layers, full elastic tilted
transverse isotropy physics was adopted to properly simulate
Adaptive application of FWI complex phase and amplitude changes along offset (Wang et al.,
To achieve a high-quality seismic image at reservoir depths, 2021). A free-surface boundary condition was used to capture
it is critical to first build an accurate overburden model that ghost and multiple effects in the field data.
captures the complexity with sufficiently high resolution. Depending on the updated target depth and its surrounding
Conventional model building methods such as reflection and challenges, three different types of FWI were deployed sequen-
refraction tomography, or even manual update approaches, are tially (Figure 3). The background model, constructed using

48 The Leading Edge January 2024

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 3. Adaptive application of FWI technology and the corresponding maximum update depth. The Kirchhoff PSDM stack is overlain by the final velocity model in which the high-
resolution features are derived from FWI updates. (For comparison, see same location as the legacy result in Figure 1.)
DOI:10.1190/leedff.2024.43.issue-1

conventional VMB tools, was not sufficiently accurate to use as to lack of reflectivity or the location being beyond the maximum
the FWI starting model for the very shallow part of the model penetration depth of diving waves. These observations would
(less than approximately 1 km depth) due to the complex nature typically correlate with a geologically implausible geometry of
of the Cretaceous carbonate layer and the presence of salt domes the carbonate platform, implying potentially incorrect velocities
and faults. Moreover, poor seismic S/N under 5 Hz challenged of overlying salt structures or minibasins (Figure 4a).
a reliable low-frequency update from the conventional waveform- Based on our geologic understanding of carbonate platform
based FWI approach. Thus, traveltime-based FWI (TT-FWI) evolution and extensive well data in the Kashagan Field docu-
was performed to minimize the traveltime difference between menting the stratigraphic relationships between salt and the
simulated and observed data, a method less prone to any potential underlying carbonate platform, we developed a notional model-
cycle skipping errors. The updated model from TT-FWI was able based base of salt (BoS) horizon, juxtaposed to the platform top,
to capture the shallow Cretaceous carbonate layer, faults, and within the platform interior, and utilized it as an additional
strong velocity contrast associated with shallow salt diapirs. geologic constraint into tomographic inversion. The tomographic
Once the complexity of the very shallow region was resolved, objective function simultaneously minimized both the seismic
waveform-based eFWI, which minimizes the canonical cross- reflection residual moveout error and the depth discrepancy
correlation objective function, was applied to resolve in high between the ideal and interpreted BoS horizon. Due to the use
resolution the velocity contrast between Jurassic carbonates and of the model-based geologic constraint, i.e., the ideal BoS
surrounding clastics, numerous faults, and complex salt diapirs. horizon, reflected into the model update, careful post-tomography
After the shallow part of the velocity model down to the Jurassic migration QC was undertaken to better understand the impact
carbonate was updated, additional record length of reflection data and benefit of the different minimization schemes tested in the
was utilized to increase the maximum depth of the eFWI velocity objective function.
update down to the base of the minibasins. Figure 3 shows that The use of residual moveout error as the sole input to reflection
the combined update from the adaptive FWI applications matches tomography for updating the overburden velocity stops short of
well with the complex variation of Kashagan geology, but also mitigating unrealistically collapsed top carbonate platform geom-
improves image focusing and reflector continuity down to the etry (Figure 4b). Horizon-constrained tomography was found to
reservoir level. resolve this geometry issue at BoS and top carbonate but caused
structural undulation and defocusing at shallower depths because
Geologically constrained reflection tomography the BoS horizon constraint alone does not have vertical resolving
Even after several iterations of reflection tomography for the power (Figure 4c). A joint tomography approach provided the
background model update, we found that stack focusing did not most optimal result — gather flatness was maintained, the col-
improve much in certain areas where the residual moveout in lapsed geometry of the top carbonate was mitigated, and stack
common-image gather (CIG) was relatively small and insensitive focusing improved without introducing structural wobble and
to any velocity update. An FWI update was also ineffective due defocused reflectivity in the shallower intervals (Figure 4d).

January 2024 The Leading Edge 49

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms
DOI:10.1190/leedff.2024.43.issue-1

Figure 4. Kirchhoff PSDM stack migrated with (a) the starting velocity before tomographic update, (b) reflection tomography only, (c) horizon tomography only, and (d) joint tomography
velocity update. The vertical dotted lines indicate well locations. The unrealistic morphology of the platform interior top and shallow reflectors (indicated by the red arrows) is mostly
healed by the joint tomography update.

After successfully updating the overburden velocity above the to delineate geologic details such as faults, the Cretaceous clastic
carbonate platform interior, we applied the same joint tomography sediment, and the shallow postsalt leads (Okere and Toothill,
approach to the platform slope and intraplatform sections. A 2012). LS-RTM is better suited for imaging the complex geology
smoothed version of the BoS horizon and of a deeper mid-Devonian containing high velocity contrasts and steep dips, such as
reflector horizon was used as the geologic constraint to update Cretaceous and Jurassic carbonate, fault shadow, salt diapirs, and
velocity above the platform slope and the intraplatform, respectively, the deep carbonate platform.
and achieved similar success: flatter CIG’s, improved stack focusing, The workflow of LS-Kirchhoff includes four steps: (1) true-
and more geologically plausible geometry of key horizons. amplitude Kirchhoff migration, (2) Kirchhoff demigration and
remigration, (3) estimation of the Hessian via nonstationary
Least-squares imaging matching filter (NMF) process (Brytik et al., 2022), and
The complex geology of the Kashagan Field poses a significant (4) application of the inverse Hessian to Kirchhoff gathers.
challenge for VMB, but also for seismic imaging, including Figure 5a illustrates the outstanding issues in the conventional
amplitude fidelity and S/N due to uneven illumination and scat- Kirchhoff image: scattering noise from the shallow salt diapirs
tering noise. Moreover, the decades-old sparse OBC data are not propagating through the whole depth, unbalanced illumination
sufficient in fold and S/N, leading to strong acquisition footprint at the target reservoir reflectors, and low S/N. LS-Kirchhoff
and residual noise even after aggressive denoise during preprocess- imaging significantly mitigated these issues, improving the image
ing (Anderson et al., 2022). LS migration technology in both fidelity and rendering it more suitable for geologic interpretation
Kirchhoff prestack depth migration (PSDM) and RTM was (Figure 5b)
leveraged to mitigate these issues. LS-Kirchhoff was used for the Unlike LS-Kirchhoff, LS-RTM is performed via a combina-
highest-resolution image, especially at shallower depths, helping tion of image-domain and data-domain updates to achieve the

50 The Leading Edge January 2024

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Figure 5. (a) Conventional Kirchhoff PSDM and (b) LS-Kirchhoff PSDM stack.

optimal image quality at all depth levels.

DOI:10.1190/leedff.2024.43.issue-1

First, the image-domain update was

carried out using the NMF approach
to improve the overall amplitude bal-
ance and S/N. Then, three iterations of
data-domain updates were performed
to further refine high-resolution fea-
tures, such as sharp fault boundaries
and shallow carbonate layers. Figure 6
shows the image uplift from LS-RTM
compared to the conventional RTM,
delineating the fault plane and enhanc-
ing amplitude balance, bandwidth, and
resolution of the reflectivity, important
for the correct interpretation of reservoir
facies (e.g., rim, karst, internal platform
bedding) and geologic modeling.

Results and discussion

In addition to the four critical ele-
ments that have been discussed here,
other important factors contributed to
the success of this project: increased
emphasis on postprocessing steps,
extensive scenario testing and interac-
tive VMB for minibasins, as well as
eff icient col laboration among
ExxonMobil, NCOC, and partners
during the iterative scenario tests and
review. These integrated VMB and
imaging efforts, along with denoise
work in preprocessing (Anderson et al.,
2022), led to a step-change outcome
Figure 6. (a) Conventional RTM and (b) LS-RTM stack. The dotted vertical line indicates a well location. compared with the legacy data,

January 2024 The Leading Edge 51

 including more geologically plausible
platform geometry, improved continu-
ity and focusing of primary reflectors,
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

reduced noise and multiple content,

and improved resolution in the reser-
voir interval.
For example, the new LS-Kirchhoff
stack clearly demonstrates overall image
uplift throughout the entire section
(Figure 7): the sharp delineation of
faults and Cretaceous and Jurassic car-
bonate layers, reduced fault shadow
effect, better focusing of the top and
flank of salt structures and minibasin
reflectors, the flat and continuous car-
bonate platform top, and a better
description of carbonate facies change
(rim, interior, slope, and paleo-platform
beddings) and karst geobodies.
Moreover, the new image is more con-
sistent with well data (sonic, checkshot,
well markers); e.g., most well marker
misfit errors are less than 0.6% with
DOI:10.1190/leedff.2024.43.issue-1

depth and less than 0.25% at the target

reservoir. Similarly, the new LS-RTM Figure 7. Vertical slice of (a) legacy Kirchhoff PSDM and (b) new LS-Kirchhoff PSDM stack.
stack shows improved image focusing
and noise reduction, enabling a clearer
delineation of the boundary between
platform interior and rim facies, higher
concentration of karst features at the
rim and slope, fracture networks at the
platform interior, and faults in the res-
ervoir section. Many such features were
not present in the legacy wave equation
migration stack (Figure 8).
Although continued advances in
seismic processing and imaging can be
anticipated in the future, further sig-
nificant improvements of this seismic
data would be challenged and/or limited
due to the sparsely sampled, low S/N
nature of the legacy data. In a parallel
testing effort, we reprocessed a small
high-density 3D four-component pilot
survey acquired over the southeastern
portion of the field (20 times source and
receiver effort, 5 times fold with respect
to the legacy data, significantly better
distribution of near-offset traces).
Preliminary images using higher-
density data provided additional
improvement in S/N and seismic resolu-
tion, increasing the interpretability of
karst features and connectivity of
fracture networks, critical for geologic
model building, well placement, and Figure 8. Horizontal slice of (a) legacy wave equation migration and (b) new LS-RTM stack.

52 The Leading Edge January 2024

 field development planning (ExxonMobil internal report). The References
imaging uplift is due not only to increased fold and S/N, but also Anderson, D., A. Svetlichnyy, V. Zhelanov, J. Oukili, S. Schreuder,
to an improved ability to apply the modern high-end technology Y. Taikulakov, I. Chikichev, and A. Baumstein, 2022, 3D demultiple
techniques dramatically improve the imaging of a giant Kashagan
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

and workflows such as broadband wavelet treatment, advanced

reservoir in the ultra-shallow North Caspian Sea: First Break, 40,
demultiple application (due to the availability of near offsets),
no. 9, 51–57, https://doi.org/10.3997/1365-2397.f b2022074.
improved 5D data interpolation and binning, higher-frequency Brytik, V., G. Palacharla, R. Bansal, D. Snyder, X. Li, Y. H. Cha,
FWI, and reliable tomographic velocity model updates. P. Routh, I. Dura-Gomez, D. Pavlov, and C. Marcinkovich, 2022,
A recipe for practical iterative LSRTM with synthetic and real data
Conclusions examples from Brazil: Second International Meeting for Applied
Both the Kashagan East and West reprocessing projects, Geoscience & Energy, SEG/AAPG, Expanded Abstracts, 2689–2693,
executed between 2020 and 2022, utilized 3D OBC input data https://doi.org/10.1190/image2022-3751103.1.
Mattson, A. G., M. Royhan Gani, T. Roesler, N. D. Gani, and J. T.
that were two decades old. The projects aimed to image the deep
Ford, 2020, 3D mapping of intruding salt bodies in the subsurface
carbonate reservoir underneath the complex geology. Compared of the Gulf of Mexico using 3D seismic data: Results in Geophysical
with the 2011 legacy reprocessing effort, the latest result shows Sciences, 1–4, 100004, https://doi.org/10.1016/j.ringps.2020.100004.
significant imaging uplift and represents the amount of additional Okere, D., and S. Toothill, 2012, New insights into hydrocarbon plays
subsurface information that can be further extracted from the in the Caspian Sea, Kazakhstan: Petroleum Geoscience, 18, 253–268,
old vintage seismic data by leveraging modern seismic processing https://doi.org/10.1144/1354-079311-045.
technology and workflows. Moreover, it is believed that the key Pettingill, H. S., 2001, Giant field discoveries of the 1990s: The Leading
Edge, 20, no. 7, 698–704, https://doi.org/10.1190/1.1487280.
to the successful and timely outcome cannot be attributed solely
Wang, H., O. Burtz, P. Routh, D. Wang, J. Violet, R. Lu, and S. Lazaratos,
to technology advancements, but also to agile application of the 2021, Anisotropic 3D elastic full-wavefield inversion to directly estimate
technologies and seamless project execution. For instance, the elastic properties and its role in interpretation: The Leading Edge, 40,
majority of the VMB effort was carried out in parallel with the no. 4, 277–286, https://doi.org/10.1190/tle40040277.1.
vendor-based preprocessing work. The early-stage shallow velocity
DOI:10.1190/leedff.2024.43.issue-1

update was driven by TT-FWI and eFWI using raw input data
with minimal preprocessing, while reflection tomography used
intermediate fast-track preprocessed data with heavy multiples CALL FOR
noise still present, leveraging additional in-house preprocessing PAPERS
capability and interpretation support. As a result, the final
LS-migration using the final preprocessed input was executed
immediately after the fully preprocessed input data became
Special Section: Mobile Shales
The main objective of this special section is to collect new geologic
available, making it possible to to provide the stunning image
and geophysical observations over a range of mobile-shale structures
uplift on time. Continuous processing technology and workflow in sedimentary basins. Contributions dealing with seismic processing,
improvements, in tandem with any future higher-density data seismic interpretation and delineation, logging data, gravity modeling,
acquisition, are expected to unlock the full potential of seismic field examples, and mechanical properties of mobile-shale structures
imaging at one of the world’s largest oil fields and help derisk in any tectonic setting are welcomed. We plan to collect contributions
upcoming business and technical decisions, such as well placement from both academia and industry that explore modeling tools and
and reservoir management and development planning. workflows to improve seismic imaging of mobile shales and to unravel
their origin and evolution.
Acknowledgments Potential topics of interest include but are not limited to:
We thank ExxonMobil Technology and Engineering Company • Seismic imaging and interpretation
(EMTEC), North Caspian Operating Company (NCOC), and our • Seismic processing
partners (KMGK, Shell, ENI, Total, CNPC, and INPEX) for • Physical properties of mobile shales
• Modeling the origin and evolution of mobile shales
permission to publish this work. We also thank Carey Marcinkovich,
Shawn Dewberry, Arjun Srinivasan, Haiyong Quan, Hong Zhao, Submission deadline: 1 FEB 2024
and Ivan Chikichev for their foundational work, and we acknowledge Publication of issue: NOV 2024
Gboyega Ayeni and Alex Martinez for the support of this project. https://library.seg.org/page/inteio/interpretation-mobile-shales
Special section editors:
Data and materials availability Juan I. Soto, Dallas B. Dunlap, Michael R. Hudec, Chris K. Morley,
Data associated with this research are confidential and cannot Mark R. P. Tingay, and Lesli J. Wood
be released.
Interpretation, copublished by SEG and AAPG, aims to advance the practice of
subsurface interpretation.
Corresponding author: jaewoo.park@exxonmobil.com

January 2024 The Leading Edge 53

 Hey members!
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

did you know ?

DOI:10.1190/leedff.2024.43.issue-1

Your SEG membership entitles you to purchase member subscriptions

to SEG Library journals at significant discounts.
Bundled options allow you to select only the publications you need and to save
on Geophysics, Interpretation, and Journal of Environmental
and Engineering Geophysics single-article prices.

Visit seg.org/benefits for options and pricing.

 Visit seg.org and be sure to log in with your e-mail address as your username.
After renewing or joining, visit seg.org/benefits and scroll to the bottom of the page.
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

1 View the available subscriptions, and click to select

the bundle you want to add.

Then (1) click the plus sign to increase the quantity

2
DOI:10.1190/leedff.2024.43.issue-1

to 1, (2) add it to your cart, and (3) view your cart.

3 Click to checkout.

Enter your payment information.

Click to continue.
4

5 Click Pay Now.

 IMAGE ’23:
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Bigger, better
T he Third International Meeting for Applied Geoscience
and Energy (IMAGE ’23), which took place 28 August to
1 September 2023, saw the greatest number of attendees, exhibi-
Forum, with integrating geology and geophysics as its theme;
(2) the FWI Case Studies 1 session with presentations on
improving imaging of subsalt reservoirs through elastic full-
tors, and program sessions of any IMAGE to date. waveform inversion, improving reservoir imaging through
More than 7300 attendees were present at the George R. long-offset ocean-bottom node data, and resolving geologic
Brown Convention Center in Houston, Texas, for the third complexity with legacy streamer surveys; and (3) the Seismic
combined annual meeting of SEG and the American Association Processing 2: Imaging Case Study session with presentations
of Petroleum Geologists (AAPG). That number was a 23% increase on improving seismic resolution by reprocessing multiple surveys,
over the attendance of IMAGE ’22 and a 43% increase over that the latest advancements in the Suriname shallow waters, reimag-
of the first IMAGE, which included both in-person and virtual ing narrow-azimuth streamer data in the northeastern Gulf of
attendees. Not only was this the most highly attended IMAGE Mexico, and high-resolution imaging in the Campos Basin
yet, its attendance was also higher than any SEG Annual Meeting using legacy seismic acquisition.
DOI:10.1190/leedff.2024.43.issue-1

since the one held in Denver, Colorado, in 2014. The evenings were equally full for those who had networking
The 7329 attendees at IMAGE ’23 represented 99 countries. on their minds. Attendees had a particularly full dance card
They were able to take part in a technical program that included Tuesday evening with honors and awards ceremonies taking
1107 accepted abstracts divided into 123 oral sessions, 70 poster place for both SEG and AAPG and a reception for the Near-
sessions, 24 special sessions, plus a variety of near-surface technical surface Geophysics Technical Section. That evening was capped
panel sessions, strategic panel sessions, breakfasts, luncheons, and off by the AAPG/SEG President’s Reception and the always-
other events. In between technical sessions, they had the oppor- entertaining Presidential Jam. For those who still had energy
tunity to browse the wares of 258 exhibitors — also a record to spare — and even for those who didn’t —
number for IMAGE and a number not approached since the SEG Wednesday evening activities included various
Annual Meetings of 2017 and 2018, which each had just over university alumni receptions, a Latin
250 exhibitors. America/Caribbean region reception,
By most any measure, IMAGE ’23 was a success, and that is and the annual editors’ reception. The Opening Session
backed up by postconference survey results that showed 90% of Nearly 750 participants (a mod- of IMAGE ’23 featured a
attendees rated the event good to excellent, 90% rated the technical est 8% increase over IMAGE ’22) discussion of the role of geophysics in
content good to excellent, and 92% would recommend the event stuck around on Friday and capped addressing global grand challenges.
Jonathan Arthur (left), executive director
to industry peers (IMAGE, 2023) off an eventful IMAGE week by of the American Geosciences Institute, led
IMAGE ’23 kicked off a week of activity on Sunday, the discussion and was joined by
27 August, with preconvention short courses and the first day SEG President Ken Tubman (middle)
of the SEG/Chevron Student Leadership Symposium. Monday and AAPG President
saw a continuation of preconvention courses, the second day of Claudia Hackbarth (right).
the Student Leadership Symposium, the SEG Council Meeting,
the IMAGE ’23 Opening Session, and a first chance for most
attendees to get into the exhibit hall for the Icebreaker Reception.
The Opening Session featured a keynote address by Jonathan
Arthur, executive director of the American Geosciences Institute,
on the value of geoscience in addressing global challenges. After
his individual presentation, Arthur invited SEG President
Ken Tubman and AAPG President Claudia Hackbarth on stage
to engage in a short panel discussion on the subject.
Tuesday brought with it the start of the technical program
oral, poster, and special sessions, which kept participants engaged
from 8 a.m. until 5 p.m. or after through Thursday. Among the
most popular technical sessions were (1) the Discovery Thinking

56 The Leading Edge January 2024

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

The IMAGE bookstore

occupied prime real estate on the
With nearly 260 exhibitors
concourse level of the George R. on hand and record IMAGE
Brown Convention Center. Adjacent attendance, the exhibit hall
to technical session rooms, the was continuously packed —
bookstore’s aisles were filled during despite frequent sightings of
breaks, although opportunities both a bear and a dinosaur.
could be found for more
leisurely browsing.

attending one or more of the 14 postconvention workshops that • Recent challenges, advances, and future applications of
were made available: machine learning in the geosciences
• Role of geosciences in sustainable land development
• Advances in the ground and airborne induced polarization • What is the future of seismic imaging?
methods for mineral exploration
• Clastic petrography and diagenesis With IMAGE ’23 concluded, preparation has already begun
DOI:10.1190/leedff.2024.43.issue-1

• Creating value: Assessing uncertainty and resolving seismic for the Fourth International Meeting for Applied Geoscience and
ambiguity Energy. IMAGE will once again convene in Houston at the
• Distributed fiber-optic sensing: Advances in applications and George R. Brown Convention Center, and organizers expect to
ML-based techniques see a continuation of the upward trend in attendance, exhibition
• Drone geophysics applied to legacy oil and gas wells and participation, and quantity and quality of technical program
pipelines content. IMAGE ’24 will take place 26–29 August 2024. Watch
• Elastic and multiparameter FWI: What is appropriate for https://www.imageevent.org, SEG’s social media channels, and
field data applications the pages of The Leading Edge for updates as they arise.
• Energy transition: How to use multidiscipline and multiphys-
ics to power energy transition Reference
• Harmonizing, integrating, and repurposing subsurface data IMAGE, 2023, IMAGE ’23 Post Show Report, https://irp.cdn-website.
• Insights and questions related to CCUS and reservoir char- com/3c79eb10/files/uploaded/IMG23_PostShowReport_web.pdf,
accessed 5 December 2023.
acterization utility in terms of static earth model description
based on geophysical methods
• Investigations into sampling and data density: Promises and
learnings for improving resolution You know the beat’s
• Orphaned wells: A new opportunity for an old problem bumping when Mike Forrest
hits the floor — a Presidential
Jam hall-of-fame moment to
cap off the most successful
IMAGE to date.

Kurt Marfurt received SEG’s

highest honor, the Maurice
Ewing Medal, on Tuesday night
during the SEG Honors and
Awards Ceremony. January 2024 The Leading Edge 57
 The Leading Edge
Erratum
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Erratum

I n the October 2023 article titled

“Delve deep: Seismic revolution in
acquisition and model building for
explorations” by Vigh et al., the incor-
rect caption was attached to Figure 6
on page 667. The figure with its cor-
rected caption is presented here.
DOI:10.1190/leedff.2024.43.issue-1

Figure 6. (a) FWI-derived velocity up to 14 Hz. (b) FWI-derived reflectivity from the final FWI velocity. (c) Reverse time migration
of the OBN data at 14 Hz (no enhancement). (d) Enhanced reverse time migration.

CONNECT WITH E&P

TRENDS, COMPANIES
AND PROFESSIONALS
AT NAPE SUMMIT.
Use Code “SEG” for $50 OFF!

TECHNICAL PROGRAM
EXPANDED ABSTRACTS
2024 SUMMIT WEEK NOW AVAILABLE
FEB 7-9 | Houston
imageevent.org/abstracts2023

58 The Leading Edge January 2024

 The Leading Edge
Announcements
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

SEG-Y revised for first time in six years • addition of a new stanza for XML-formatted Trace Header

T he Society of Exploration Geophysicists has released an

update to the SEG-Y seismic data-exchange standard.
Incorporating more than five years of development and feed- •
Layout with deprecation of the existing Trace Header Mapping
stanza
several corrections and clarifications in format description text
back from a large community of users, SEG-Y Revision 2.1
(SEG-Y_r2.1) provides a method of capturing and recording user Joel Allard, Victor Ancira, Stewart Levin, and Jill Lewis, led
knowledge via an XML file, which is written between the binary by chair Shawn New, guided development of this revision and
header and the trace data. This format revision ultimately renders edited its content under the auspices of the SEG Technical
data sets machine readable and suited as inputs for machine Standards Committee. SEG published the first version of the
learning and artificial intelligence applications. Notable new SEG-Y standard in 1975. Revision 1 was published in 2002, and
features of SEG-Y_r2.1 include: Revision 2 was issued in 2017. SEG-Y_r2.1 is the first revision
issued by the Society in six years.
• binary header bytes 3507-10 changed to insert new survey The updated format standard may be downloaded from the
type field SEG Technical Standards web page at https://library.seg.org/
• deprecated appendices D-3 through D-5 removed, and sub- seg-technical-standards. The full PDF can be viewed at https://
sequent appendices renumbered library.seg.org/seg-y_rev2-1.
DOI:10.1190/leedff.2024.43.issue-1

The Leading Edge

Board Report
SEG Board of Directors actions from September and October 2023
Approved minutes from the 27 August 2023 and 28 September
2023 Board meetings.

Accepted the SEG unaudited financial statements for July and

August 2023.

Approved extending the agreement to conduct joint annual meet-

ings between AAPG and SEG for years 2026–2030.

Approved new appointments for Bill Abriel, Mohammed Badri,

and Chitty Chang to serve on the SEG Foundation Board of
Directors.

Approved a three-year reappointment for Aria Abubakar to serve

on the SEAM Board of Directors and to adjust the start date of
terms for Jim Hollis, Chengbo Li, and Rich Cieslewicz to begin
immediately.

Approved a petition for affiliation from the China University of

Mining and Technology Student Chapter.

January 2024 The Leading Edge 59

 Reviews
C o o r d i n at e d by Julie Aitken
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

NASA’s Voyager Missions — Exploring the Outer Solar System and Beyond, absorbing and could arguably make a book of its own. However,
second edition, by Ben Evans, ISBN 978-3-031-07923-8, Springer, it is soon time to start looking at the targets of the two missions.
2022, 243 p., US$32.99 (print), $24.99 (e-book). Each of the planets explored (Jupiter, Saturn, Uranus, and Neptune)
is afforded a chapter of its own, often with a lot of detail of the

T he passage of time is a peculiar thing. As Einstein would tell

you, it is all relative to the observer. The Voyager missions
launched in 1977, almost half a century ago. For some of you reading
many satellites that orbit each planet. These smaller worlds are
often as fascinating as the main planet itself and present some of
the strangest and most intriguing landscapes within the solar
this, this was well before you were born. For some of us, it was part system. The author also adeptly weaves in what we have learned
of our childhood and seems not that long ago. However, these about the solar system in the years following the visits of the Voyager
launches are now closer in time to the end of the Second World spacecraft, putting into context what we learned then, what we
War than today. Yet, the two spacecraft are still sending back have learned since, and what we still do not know. During these
pioneering information from the deepest reaches of space that chapters, my only regret was that I was reading the e-book. I suspect
mankind has ever probed. The engineering, planning, and fore- my little Kindle could do no justice whatsoever to the copious
thought that allowed that to happen is astonishing and extremely illustrations (68 throughout the volume), especially of such strange
humbling. The description of these activities is probably worth the objects as the “cantaloupe” moon of Neptune called Triton.
price of this book alone. However, there is much more to discover The Voyager missions are exemplary and extraordinary pieces
in this newly updated edition of a volume first published in 2003. of science that deserve to be understood and appreciated by a wide
Ben Evans, a well-published author on all things to do with audience. This book should help do that. My only concern is it
mankind’s forays into space, opens the book with a nicely detailed sometimes uncomfortably straddles the gap between a popular-
DOI:10.1190/leedff.2024.43.issue-1

historical introduction. In particular, Evans describes the discovery science book and an academic tome. However, it is impossible to
of the outer planets (and their many satellites), which would eventu- get past the mind-boggling engineering and computer program-
ally become the target of the Voyager missions. The book really ming achievements that allow a computerized probe built in 1975
kicks into gear in the second chapter, in which the fascinating to still operate, send back information, and even be manipulated
history of how the missions came about is described. This conflation when Voyager 1 is at a distance of nearly 15 billion miles (at the
of economics, politics, science, and sheer good luck (the relative time of writing) from home.
positions of the outer planets in the mid to late 1970s meant this — John Brittan
was the best time to launch a multiplanet probe) is never less than Weybridge, UK

Use your geoscience skills

to impact a community in
need. Apply for GWB funding
at seg.org/gwb

60 The Leading Edge January 2024

 The Leading Edge
Membership
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

A pplications for Active membership have been received from

the candidates listed below. This publication does not consti-
tute election but places the names before the membership at large
Requirements for membership
Active: Eight years of professional experience practicing
or teaching geophysics or a related scientific field. Mem-
in accordance with SEG’s Bylaws, Article III, Section 5. If any bership applications and details of other types of member-
member has information bearing on the qualifications of these ship, including Associate, Student, and Corporate, may
candidates, it should be sent to the SEG president within 30 days. be obtained at https://seg.org/membership.

For Active membership Fuentes, Candela (Argentina)

Aabø, Tala Maria (Denmark) George, Gilbert M. (India)
Abakarov, Dashgin (Azerbaijan) Guo, YuTing (China)
Abeldinova, Zhazira (Italy) Gupta, Kartik (India)
Aguilar Arias, David (Mexico) Gwlavin, Mbouity Ngoma (Italy)
Akomolafe, Fauzan (Nigeria) Hernandez Casasus, Alejandro (Mexico)
Albanay, Montather (Saudi Arabia) Hipolito, Angel (Mexico)
Al-Fakih, Abdulrahman Ahmed (Saudi Arabia) Hovor, Eric (Ghana)
Alkhardawi, Mansour (Saudi Arabia) Huaigu, Tang (China)
Almadani, Saleh (Saudi Arabia) Huang, Song (China)
Almessabi, Abdulrahman (UAE) Hughes, Makenna (USA)
DOI:10.1190/leedff.2024.43.issue-1

Almkainah, Mohammed (Saudi Arabia) Ilyakova, Ekaterina (Canada)

Almoubarak, Ahmed (USA) Kashyap, Aditya Raj (India)
Al-Qadasi, Basem (Saudi Arabia) Khan, Baseem (India)
Alvarado Izarraras, Luis Gerardo (Mexico) King, Matilda (USA)
Arias Ramos, Eduardo (Mexico) Kloss, Maria Gabriela (Brazil)
Asare-Bediako, Baah (Ghana) Kohut, Amanda (Brazil)
Ateeq, Amna (Pakistan) Kumar, Mihir (India)
Audu, Faith (Nigeria) Lagos, Vanesa (Colombia)
Balooni, Satya (India) Larkin, Steven (USA)
Banerjee, Sankar (India) LaRue, Robert (Australia)
Barahona, Fabian (Mexico) Leucci, Giovanni (Italy)
Bhattacharjee, Somaabha (India) Li, Wei (China)
Bilic, Gina (Canada) Liang, Feng (China)
Büyük, Ersin (Turkey) Lofton, Hermione (USA)
Cancino, Maxwell (Mexico) Lubis, Kevin (Indonesia)
Carrington, Eseverere (Nigeria) Lund, Hugh (UK)
Chand, Somnath (India) Maciel, Maria Guadalupe (Argentina)
Cini, Estefania (Argentina) Maigana, Maryam (Nigeria)
Crichlow, Alishia (USA) Mammadov, Kanan (Azerbaijan)
Crowe, Rykley (USA) Marques, Carlos (Italy)
Danchenko, Victor (Russian Federation) Marques, Felipe (Brazil)
Daqiq, Mohammad Taqi (India) Martinez, Tila (Mexico)
De La Cruz Rodriguez, Adimaely (Mexico) Mejia Calderon, Juan (Colombia)
De La Rosa Andrade, Jonathan (Mexico) Mendoza Garcia, Emiliano (Mexico)
Dewanty, Cecilia (Indonesia) Meng, Xiaobo (China)
Di Maio, Serena (Italy) Mgbeji, Onyedikachi (Nigeria)
Di Pane, Ana (Argentina) Miller, William (USA)
Dixit, Shikha (India) Mills, Stephen (USA)
Doshi, Rakesh (Malaysia) Min, Fan (China)
Echavarri, Zoe (Argentina) Miorali, Mattia (UK)
Egger, Christopher (USA) Moaaf, Turki (Saudi Arabia)
Egunjobi, Kehinde (United States Minor Outlying Islands) More, Shivprasad (India)
Elsergany, Mohamed (Saudi Arabia) Moya, Erika (Argentina)

January 2024 The Leading Edge 61

 Mubin, Mukhriz (Malaysia) Sitton, Gwen (USA)
Nava, Aarón (Mexico) Skuce, Kelly (Canada)
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Nazaraliyev, Rashad (Azerbaijan) Spiller, Reginal (USA)

Ocañas, Ezequiel (Mexico) Thakur, Shristi (India)
Ogieva, Mathew (USA) Theobaldo Jorge, Vinicius (Brazil)
Ogunsakin, Oluwakunle (Nigeria) Thieme, Donald (USA)
Oladuji, Olaoluwa (USA) Trinidad Ricardez, Belen (Mexico)
Paolini, Luca (Brazil) Ugwu, Obinna Chukwuemeka (Italy)
Peng, Younghui (USA) Uribe Loaiza, Jhon (Colombia)
Penukula, Akhil (India) Van Wyk, Milaan (USA)
Perez Jimenez, Gustavo (Mexico) Varela Molina, Gonzalo Ezequiel (Argentina)
Persico, Raffaele (Italy) Villalpando Mota, Marlen (Mexico)
Prado Barros, Enkar David (USA) Villalpando Mota, Marlen (Italy)
Puchakhova, Ariza (Azerbaijan) Wang, Hao (China)
Rezaei, Shiba (UK) Wang, Herbert (USA)
Rivera, Emilio (Mexico) Wear, Jonathan (USA)
Robles, Jenifer (Mexico) Wright, Michael (USA)
Rodriguez, Zulema (Mexico) Yadav, Sneh (India)
Romero Landin, Deisy (Mexico) Yang, Zhentao (China)
Ruempker, Georg (Germany) Youmans, Brad (Canada)
Schuette, Jaren (USA) Zhang, Lili (China)
Seyed Sajadi, Shahrzad (Norway) Zhiqiang, Jiang (Italy)
DOI:10.1190/leedff.2024.43.issue-1

Shi, Qingmin (China) Zhou, Mei (China)

Shivhare, Ayush (India)

Join the

ARE YOU PASSIONATE ABOUT GEOSCIENCE?

Use our digital space to fulfill your ambitions.
Experience the support of a global community ignited
by a shared passion. Find opportunities, connect, and
exchange tips and tricks with people in your field.

62 The Leading Edge January 2024

 The Leading Edge
Meetings Calendar
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

See https://seg.org/events/events-calendar for the latest updates.

FEBRUARY 2024 APRIL 2024 21–23 MAY

Recent Advances in Geophysical Reservoir
7–9 FEB 7–9 APR Characterization and Monitoring of CO2
NAPE Summit Week SEG 1st Tarim Ultra-Deep Oil & Gas Exploration Sequestration in Carbonate Reservoirs
https://napeexpo.com/summit Technology Workshop https://seg.org/calendar_events/co2-
Houston, Texas, USA https://seg.org/calendar_events/2024-seg-1st- sequestration-carbonate-reservoirs-2024
tarim-ultra-deep-oil-gas-exploration-technology- Abu Dhabi, UAE
12–14 FEB workshop
International Petroleum Technology Conference
https://www.iptcnet.org
Chengdu, China
JUNE 2024
Dhahran Expo, Dammam, Saudi Arabia 23–24 APR 4–5 JUN
Borehole Geophysics: New Developments for SPE/SEG Workshop: Injection Induced
25–28 FEB Global Energy Seismicity
ASCE Geo-Congress 2024 https://seg.org/calendar_events/borehole- https://www.spe-events.org/workshop/injection-
https://www.geocongress.org geophysics-for-exploration-development-ccs induced-seismicity
Vancouver, British Columbia, Canada Kuala Lumpur, Malaysia Fort Worth, Texas, USA

27 FEB–1 MAR 23–25 APR 17–19 JUN

Offshore Technology Conference Asia Next Generation Land and Shallow Water Unconventional Resources Technology
DOI:10.1190/leedff.2024.43.issue-1

https://2024.otcasia.org Acquisition Workshop Conference (URTeC)

Kuala Lumpur, Malaysia https://seg.org/calendar_events/new-generation- https://urtec.org/2024
aquisition-2024 Houston, Texas, USA

MARCH 2024 Muscat, Oman

24–27 JUN
7 MAR 30 APR–2 MAY Net-Zero Emissions Workshop
Denver Geophysical Society’s Annual 3D Summit on Geophysical Detection of Explosive https://seg.org/calendar_events/net-zero-
Seismic Symposium Remnants of War: Solving Current Challenges of emissions-workshop
https://denvergeo.org/events/29th-3d-seismic- Unexploded Ordinance (UXO) and Demining Virtual
symposium https://seg.org/calendar_events/summit-on-
Denver, Colorado, USA geophysical-detection-of-explosive-remnants-of-
war-solving-current-challenges-of-unexploded-
JULY 2024
11–13 MAR ordnance-and-demining 3–4 JUL
Carbon Capture, Utilization, and Storage (CCUS) Virtual Advanced Geophysical Solutions for Complex
https://ccusevent.org/2024 Geological Settings
Houston, Texas, USA
MAY 2024 https://seg.org/calendar_events/advanced-
geophysical-solutions-for-complex-geological-
19–21 MAR 6–9 MAY settings
Role of Geosciences in Carbon Storage Offshore Technology Conference Kuala Lumpur, Malaysia
https://seg.org/calendar_events/role-of- https://2024.otcnet.org
geosciences-in-geological-carbon-storage Houston, Texas, USA 14–19 JUL
Mumbai, India Geophysical Research for Gigatonnes
19–22 MAY CO2 Storage Workshop
27–29 MAR GEM 2024 Shenzhen: International Workshop on https://seg.org/calendar_events/geophysical-
SEG/SPE/SPWLA Workshop: From Gravity, Electrical, and Magnetic Methods and research-toward-gigatonnes-co2-storage
Measurements to Theory: Adventures in Rock Their Applications Golden, Colorado, USA
Physics, Petrophysics, and Engineering https://seg.org/calendar_events/gem-2024
https://seg.org/calendar_events/adventures-in- Shenzhen, China
rock-physics-petrophysics-and-engineering
Norman, Oklahoma, USA

January 2024 The Leading Edge 63

 S e i s m i c S o u n d o f f — C o o r d i n at e d by Andrew Geary
Solving future challenges for deep exploration
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

J oin Chao Wang and Stephen Graf on a captivating tour through

the state of the art in deep exploration, from ocean-bottom
nodes (OBNs) to artificial intelligence (AI)-assisted interpretation.
Graf: I think the potential is there for
both. To Chao’s point, FWI and some
new technologies on land are more
Listen to the full episode at https://go.seg.org/episode-203. difficult. Which one is easier is an in-
teresting multifaceted question because
Andrew Geary: How has the meaning of deep exploration the ease of exploration depends on the
changed throughout the history of oil and gas exploration? inherent subsurface risk, above-ground
or water risk, and company objective.
Stephen Graf: A mentor of mine said that exploration takes There are a lot of factors that go into
many forms. Deep exploration is a constantly moving definition it. For example, if the goal of a company
and is always evolving. For example, the first out-of-sight land is to flip an acres position, it may be
offshore well was in 1947 in 15 ft of water in the Gulf of Mexico easier for them to explore onshore to
shelf. Not too long ago, the first subsalt canopy well was drilled quickly prove up a position. If the goal is to have steady long-term
in South Mars 200. Those are two examples where the goalpost production, it may be easier to explore offshore. The main risk
naturally keeps moving and gets pushed farther out. It also de- elements now become geologic things such as top seal. So, for
pends on which frame of reference you are using. We come to carbon sequestration, the risk is basically a top-seal risk. How
the present, where we are starting to reach into the 20,000 psi much is the CO2 going to leak off in X number of years? I think
realm. We have the technology to do that. We have new discover- the one with greater potential and the one that is easier is the
ies in the Guyana and Orange basins. So, I think it’s this con- age-old answer: it really just depends.
DOI:10.1190/leedff.2024.43.issue-1

stantly moving target. In addition, you factor in net-zero targets

and carbon capture and sequestration. That is a completely new Geary: What is one misconception that members of the
territory that can be considered in and of itself a sort of exploration public often have about deep exploration?
because we truly do not know how these large-scale sequestration
efforts are going to work. Graf: First and foremost is safety. When you go out to a modern
drillship, semisub, or land rig, they are clean and massive. They
Geary: Is exploring on land or water easier at this time? are floating cities. The commitment to safety in the industry has
Which has the greater potential? made leaps and bounds over the past decade or two. I think that
gets missed, particularly when news stories come out that paint
Chao Wang: Different environments come with their own the industry in a negative light. For the things that do come out,
complexities, advantages, and potentials. In terms of ease, I they definitely need to be addressed. However, with new AI-
think we can say land exploration may be considered easier due focused technologies, I think safety is going to continue to improve.
to easy access and low cost. However, research challenges still Furthermore, the amount of cutting-edge technology away from
remain. On land, the challenges include near-surface variability the safety side gets missed. It goes unseen. How much technologi-
and signal-to-noise ratio. These require advanced static cor- cal advancements in the oil and gas industry have transferred to
rection, accurate near-surface velocity model building, and other industries? It is truly mind boggling to think about the
simulations such as advanced noise attenuation and signal amount of technology that goes into the things that we all do in
enhancement techniques. These are very important for land. the industry. I think that gets lost. Maybe we need to do a better
For marine, it is quite different. The challenges include complex job of messaging and promoting that.
subsalt imaging, illumination, and model uncertainties. One
good technology for subsalt is a new acquisition technology Wang: Deep exploration is not only for drilling deeper wells to
such as OBN and advanced inversion algorithms such as elastic find more hydrocarbons. While depth is certainly a component,
full-waveform inversion (FWI). These are used to enhance it is not the only one. We need to combine all geology, technology,
accuracy and resolution. So, they all have different challenges and environments. So, it is not just about finding more oil and
and different potentials. gas but doing so responsibly and efficiently.

64 The Leading Edge January 2024

Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

26-29 August 2024 • Houston, Texas

CALL
FOR
DOI:10.1190/leedff.2024.43.issue-1

ABSTRACTS
SHARE YOUR KNOWLEDGE Opens
WITH A GLOBAL AUDIENCE 15 January
SEG and AAPG, in conjunction with SEPM, are
set to host the annual International Meeting for
Applied Geoscience and Energy (IMAGE), from
26–29 August 2024 in Houston, Texas at the Start planning now to share your case studies,
George R. Brown Convention Center. technological advancements, and research
discoveries with the leading assembly of applied
IMAGE ’24 is the most influential platform for geophysicists, thought leaders, and technical
energy professionals and gathers a community experts from around the world. Whether you are
of geoscientists and industry leaders that a veteran or novice speaker, we want to hear
collaborate to guide and shape our future. from you!

IMAGEevent.org
Downloaded 07/23/25 to 122.161.72.155. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/page/policies/terms

Discover Sercel
Land Services
DOI:10.1190/leedff.2024.43.issue-1

SOLVING COMPLEX CHALLENGES

WITH TECHNOLOGY-DRIVEN EXPERTISE
Sercel specializes in solving the most complex operational challenges for
seismic contractors and operators, ensuring that projects are completed
on time, within budget, and according to technical specifications.

Trust us to streamline your operations, improve operational efficiency, and

deliver unrivaled project success.

Explore our suite of land services:

• Train & launch
Start your project with confidence
• Asset optimization
Get the best from your equipment
• Operational excellence
Steer your crew performance

Nantes, France
sales.nantes@sercel.com
Houston, USA
For more information sales.houston@sercel.com
www.sercel.com