Академический Документы
Профессиональный Документы
Культура Документы
6
Special Section: Special Section:
Nonreflection seismic and inversion
of surface and guided waves
A TGS COMPANY
See the Value
Discover how Arcis
can deliver your solution with
industry-leading service and quality.
AMALGAMATED REEF
2 KM
WINNEPEGOSIS REEFS
(Structuralcurvature)
2km
Nega2ve
Posi2ve
(Amplitudecurvature)
Winnepegosisreefs Amalgamatedreef
(Amplitude curvature)
(Structural curvature)
(Structuralcurvature)
2km
Nega2ve
Posi2ve
(Amplitudecurvature)
Winnepegosisreefs Amalgamatedreef
(Structuralcurvature)
2km
Nega2ve
Posi2ve
(Amplitudecurvature)
Winnepegosisreefs Amalgamatedreef
(Structuralcurvature)
2km
Nega2ve
Posi2ve
(Amplitudecurvature)
Winnepegosisreefs Amalgamatedreef
Positive
Negative
(Structuralcurvature)
2km
Nega2ve
Posi2ve
(Amplitudecurvature)
Winnepegosisreefs Amalgamatedreef
Reservoir Analysis Adds Value
At Arcis, we offer a state-of-the-art suite of seismic reservoir
services comprising seismic attributes, AVO analysis, AVAz/
VVAz, seismic amplitude inversion, thin-bed reectivity
inversion, spectral decomposition, multi-attribute analysis
and other accurate solutions aimed at characterizing
reservoirs.
View more Reservoir Solution examples at arcis.com
LeadingEdge_Arcis_FP_ResAnl_1_05_2013.indd 1 2013-05-08 10:53 AM
A Clearer Image
www.pgs.com
Ramform Titan Built to Perform
The new Ramform Titan puts PGS further ahead as the leader in
safe, productive and effcient 3D Seismic Acquisition.
With a length of 104 meters and a 70 meter wide back deck, this
vessel increases stability and enables safe, hands-off deployment
and recovery of multiple streamers simultaneously.
Capable of towing up to 24 streamers and designed to extract
the full potential from the Geostreamer technology; the Ramform
Titan illustrates an ongoing process of enhancement aimed
towards one single purpose to collect the maximum amount of
data quickly, safely and reliably.
More effcient seismic acquisition
Larger window of operation
Repair and maintenance at sea
Optimised for GeoStreamer
(Print ISSN 1070-485X; Online ISSN 1938-3789) is published monthly by the Society of Exploration Geophysicists, 8801 S. Yale Ave.,
Tulsa, Oklahoma 74137 USA; phone 1-918-497-5500. Per iodicals postage paid at Tulsa and additional mailing ofces. Print subscriptions for members of
the Society in good standing are included in membership dues paid at the World Bank III and IV rate. Dues for Active and Associate members for 2013 vary
depending on the three-tiered dues structure based on World Bank classication of the members country of citizenship or primary work residence. Dues
are US$90 (World Bank IV countries), $48 (World Bank III countries), and $12 (World Bank I and II countries). Dues for all Student members regardless of country of citizenship or primary residence are $21 and include
online access to journals. Students may receive TLE in print by paying an additional $36. Print and online single-site subscriptions for academic institutions, public libraries, and nonmembers are as follows: $155, Domestic
(United States and its possessions); $190, Surface Freight (Canada, Mexico, Central and South America, Caribbean); and $200, Mandatory Air Freight (Europe, Asia, Middle East, Africa, and Oceania). For corporations
and government agencies, print and online single-site subscriptions are: $840, Domestic (United States and its possessions); $875, Surface Freight (Canada, Mexico, Central and South America, Caribbean); and $885,
Mandatory Air Freight (Europe, Asia, Middle East, Africa, and Oceania). Print-only subscriptions for corporations and government agencies are: $340, Domestic (United States and its possessions); $375, Surface Freight
(Canada, Mexico, Central and South America, Caribbean); and $385, Mandatory Air Freight (Europe, Asia, Middle East, Africa, and Oceania). Rates are subject to change without notice. Subscriptions to the SEG Digital
Library include subscriptions to TLE. Subscribers to GeoScienceWorld are entitled to a $30 discount off print-only subscriptions to TLE. See www.seg.org/publications/subscriptions for ordering information and details.
Single-copy price is $16 for members and $32 for nonmembers. Postage rates are available from the SEG business ofce. Advertising rates will be furnished upon request. No advertisement will be accepted for products
or services that cannot be demonstrated to be based on accepted principles of the physical sciences. 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 the ofcers or members of SEG. Unsolicited manuscripts and materials will not be returned unless accompanied by a self-addressed, stamped envelope. Copyright 2013 by the Society of Exploration
Geophysicists. Material may not be reproduced without written permission. Printed in USA.
POSTMASTER: Send changes of address to
THE LEADING EDGE
Box 702740, Tulsa, OK 74170-2740 USA
T
h
e
L
e
a
d
i
n
g
E
d
g
e
E
d
i
t
o
r
i
a
l
B
o
a
r
d Chairman
William Goodway
Apache Canada
9 Avenue Southwest
Calgary, AB T2P 3V4 Canada
Tel: +1-403-303-5958
bill.goodway@apachecorp.com
Gregory S. Baker
University of Tennessee
1412 Circle Drive
Knoxville, TN 37996, USA
Tel: +1-865-974-6003
gbaker@tennessee.edu
Ezequiel F. Gonzalez
Shell International E&P
3333 Highway 6 S.
Houston, TX 77082, USA
Tel: +1-281-544-7396
Ezequiel.gonzalez@shell.com
Shuki Ronen
CGGVeritas
shuki.ronen@gmail.com
Tad Smith
Apache Corporation
2000 Post Oak Blvd. Suite 100
Houston, TX 77056, USA
Tel: +1-713-296-6251
tad.smith@apachecorp.com
Carlos Torres-Verdin
University of Texas Department of
Petroleum and Geosystems Engineering
1 University Station, Mail Stop C0300
Austin, TX 78712-0228, USA
Tel: +1-512-471-4216
cverdin@mail.utexas.edu
Special editor
Christopher L. Liner
University of Arkansas
Department of Geosciences
346 Arkansas Avenue, STOS G18
Fayetteville, AR 72701-1201
Tel: +1-479-575-4835
liner@uark.edu
President
David J. Monk
Apache Corporation
2000 Post Oak Blvd.
Houston, TX 77056, USA
Tel: +1-713-296-6339
david.monk@apachecorp.com
President-Elect
Don W. Steeples
University of Kansas
1475 Jayhawk Blvd.
Lawrence, KS 66045, USA
Tel: +1-785-864-2730
don@ku.edu
First Vice-President
Richard D. Miller
Kansas Geological Survey
1930 Constant Ave., Campus West
Lawrence, KS 66047, USA
Tel: +1-785-864-2091
rmiller@kgs.ku.edu
Second Vice-President
Dennis A. Cooke
The University of Adelaide
Santos Building/Australian School of Petroleum
Adelaide, South Australia, 5005 Australia
Tel: +61-(0)418249130
dcooke@seg.org
Treasurer
Gary G. Servos
Ovation Data Services
14199 Westfair East Dr.
Houston, TX 77041, USA
Tel: +1-713-300-1651
garyservos@ovationdata.com
Editor
Tamas Nemeth
Chevron
6001 Bollinger Canyon Road
San Ramon, CA 94583, USA
Tel: +1-925-842-0998
tamas.nemeth@chevron.com
Past-President
Bob A. Hardage
Bureau of Economic Geology
University Station, Box X
Austin, TX 78713, USA
Tel: +1-512-471-0300
bob.hardage@beg.utexas.edu
Director at Large
Christine E. Krohn
ExxonMobil Upstream Research
3319 Mercer Street
Houston, TX 77252, USA
Tel: +1-713-461-7056
chris.e.krohn@exxonmobil.com
Director at Large
Alfred Liaw
Anadarko Petroleum Corporation
1201 Lake Robbins Dr.
The Woodlands, TX 77380, USA
Tel: +1-832-636-1225
alfred.liaw@anadarko.com
Director at Large
Elsa Jeanneth Jaimes R.
OGX Petrleo e Gs Ltda
Av Calle 116 #7-15 Of 1402
Bogot, Colombia
Tel: +1-571-587-1200
ejaimes@seg.org
Director at Large
Samir Abdelmoaty
Tel: +20100 166 6662
sabdelmoaty@seg.org
Director at Large
A. Peter Annan
Sensors & Software Inc.
1040 Stacey Court
Mississauga, ON L4W 2X8, Canada
Tel: +1-905-624-8909
apa@sensoft.com
Director at Large
Edith J. Miller
Chevron U.S.A. Inc.
1500 Louisiana St., 20-040
Houston, TX 77002, USA
Tel: +1-832-854-4556
edithjmiller@gmail.com
Chair of the Council
Mike Graul
TexSeis, Inc.
10810 Katy Freeway, Ste. 201
Houston, TX 77043, USA
Tel: +1-713-248-3562
mgraul@texseis.com
Real training for the real world. www.ou.edu/mcee
OUs Mewbourne College:
Where the classroom meets the oilfield.
Students at the University of Oklahomas Mewbourne College of Earth & Energy
arent just here to get a diplomatheyre here to prepare for successful careers in
the oil and gas industry, aided by some of the most advanced technology available.
The National Oilwell Varco Interactive Drilling and Well Control Simulator
The M-I SWACO Drilling and Completion Fluids Laboratory
The new undergraduate Petroleum Engineering Laboratory
The new Petrophysics and Frontier Shale Laboratories
598 The Leading Edge June 2013
E
d
i
t
o
r
i
a
l
C
a
l
e
n
d
a
r
Issue ... Special Section theme ........................................ Due date ............. Guest editors
2013
Jul .........Hydrogeophysics ................................................................ past due .................... Rick Miller, rmiller@kgs.ku.edu
Kamini Singha, ksingha@mines.edu
Aug .......Gravity and potential elds ................................................ past due .................... Michal Ruder, meruder@wintermoon.com
........................................................................................................................................... Robert Pawlowski, RPawlowski@Chevron.com
Sep ........ Full waveform inversion ...............................................................past due ......................Antoine Guitton, aguitton@gmail.com
Tariq Alkhalifah, tariq.alkhalifah@kaust.edu.sa
Chris Liner*, chris.liner@gmail.com
Oct ......... Geohazards ..................................................................................15 Jun 2013 ................Carlos Torres-Verdin*, cverdin@uts.cc.utexas.edu
3D VSP Jitendra Gulati, jgulati@slb.com
Robert Stewart, rrstewar@central.uh.edu
Nov .......Offshore and onshore broadband seismology .................... 15 Jul 2013 ............... Shuki Ronen*, shuki.ronen@gmail.com
........................................................................................................................................... William Goodway*, bill.goodway@apachecorp.com
Dec ........Unconventional resources technology ................................ 15 Aug 2013 ............. Tad Smith*, Tad.Smith@apachecorp.com
........................................................................................................................................... Carlos Torres-Verdin*, cverdin@uts.cc.utexas.edu
........................................................................................................................................... Ezequiel Gonzalez*, ezequiel.gonzalez@shell.com
2014
Jan ........Middle East ......................................................................... 15 Sep 2013 ............. Chris Liner*, chris.liner@gmail.com
Adel El-Emam, AEMAM@kockw.com
Said Mahrooqi, Said.S.Mahrooqi@pdo.co.om
Feb ........4D ....................................................................................... 15 Oct 2013 .............. William Goodway*, bill.goodway@apachecorp.com
........................................................................................................................................... Alan Jackson, Alan.Jackson@shell.com
Mar .......Rock physics ...................................................................... 15 Nov 2013 ............. Tad Smith*, Tad.Smith@apachecorp.com
........................................................................................................................................... Carlos Torres-Verdin*, cverdin@uts.cc.utexas.edu
........................................................................................................................................... Ezequiel Gonzalez*, ezequiel.gonzalez@shell.com
(* Current TLE Board member)
www.seg.org/meetings/pol13
Notice to authors
TLE publishes articles on all areas of applied geophysics and disciplines which
impact it. To submit a paper for possible publication in a specic issue, please
e-mail an inquiry to the appropriate guest editor for that issue. Authors are
encouraged to submit their papers at any time, regardless of whether they t the
schedule. To submit an article on an unscheduled topic, contact Dean Clark, TLE
editor, dclark@seg.org or 1-918-497-5535.
Electronic submission of articles
Electronic submissions should include the manuscript le, gures and other
graphics, a PDF of the manuscript and gures, and the authors contact
information. These les can be uploaded to an FTP site (the preferred method)
or burned to a CD and mailed to the appropriate editor. Once accepted for TLE,
the les will be opened and edited on a Mac or a PC using various software
applications. To simplify conversion, gures should be submitted in TIFF, PDF
or EPS (.tif, .pdf or .eps) le formats, with a resolution of at least 300 dpi (pixels
per inch). High-resolution images can be placed in Word or PowerPoint if
placed large on the page; these will be converted to PDF format. Once the paper
is accepted, please also mail the appropriate editor a printed color copy of the
manuscript with any gures, tables, and equations to be included. For assistance
with electronic submission, contact Tonia Gist, tgist@seg.org or 1-918-497-
5575. More details are online at http://www.seg.org/resources/publications/tle/
submission-guidelines.
Notice to lead authors
Lead authors of articles published in TLE who are not members of SEG should
apply for a one-year free membership and subscription to TLE by contacting
Member Services, fax 1-918-497-5557 or membership@seg.org. Lead or
corresponding authors also are required to sign a copyright transfer agreement,
which gives TLE permission to publish the work and details the magazines and
the authors rights. TLE staff will send a form to be signed and sent back after the
article is accepted for publication. The form can be downloaded at http://www.
seg.org/documents/10161/74670/SEG_Copyright_form.pdf.
A revolution in super-resolution
for your shale and carbonate plays
Z-Terra Inc.
1I1I1 |ar| kew, !alte 1I - Neastea, !/ II1
1.1. www.t-terra.cem
Ceatact as te4a, at sa|eszt-terra.cem.
In shale and carbonate plays,
you need every advantage
you can get. Z-Terra Difraction
Imaging gives you a crucial
technological edge with the
highest resolution imaging and
mapping of natural fractures
and subtle stratigraphic
discontinuities.
With max-resolution difraction
imaging of natural fractures
and subtle stratigraphic
discontinuities youll optimize
stimulation, well placement and
production.
The max-resolution
imaging you need to:
t Identify potential sweet spots
in unconventional shale
reservoirs and conventional
carbonate reservoirs
t Map the areal distribution of
natural fractures
t Detect and map subtle
stratigraphic changes and
discontinuities
t Output pre stack difraction
gather data to properly
image fracture amplitude as a
function of azimuth
The images are clear so is
the choice: Z-Terra Difraction
Imaging.
Data Courtesy of Seitel
Z-Terra Difraction
Imaging the edge
you need in shale and
carbonate plays.
ZTERRA
Before
After
600 The Leading Edge June 2013
O
ver the past few years, there has been much anticipation
and more than a little industry anxiety concerning the
waves of departures of seasoned workersthe silver
tsunami. Teir incoming, less-experienced replacements
were certain to arrive at their jobs without the guidance and
mentoring necessary to ensure the continuity of leadership
and quality work product required by a demanding and high-
stakes business.
In the September 2007 issue of Te Leading Edge, outgo-
ing SEG President Leon Tomsen wrote a Presidents Page
discussing, among other things, SEGs role in facilitating the
knowledge transfer from the long-term experienced Earth-
science professionals to the younger geophysicists beginning
to enter the workforce.
Two months later, in the November issue, President-elect
Larry Lines addressed the importance of developing greater
communication between industry and the institutions re-
sponsible for educating future Earth scientists. He also point-
ed to the growing SEG student membership (exceeding 7000
at that time) as an indication of a level of condence that a
steady pipeline of eager and motivated professionals could be
counted on to meet demandprovided they started their of-
cial employment better prepared than any previous genera-
tion of workers.
Now I, as a member of this change, am taking the oppor-
tunity to put forth my own observations and provide some
insight as a member of that new crew on how SEG is meeting
its goal of being a vital player in what most would agree is the
important aair of wisdom transfer.
I entered the workforce in 2006, with legions of peers,
only to see the 2008 crash drastically cut the supply of new
Earth scientists. Fortunately, the tide has changed and the
new crew is growing.
SEG indeed serves an important role in the transition of
crews. It has put programs in place, many with the help of
generous sponsors and donors, that better prepare students
for their professional careers, and that help early-career geo-
physicists ramp up faster by providing access to greater re-
sources and knowledge.
Te SEG Web site is the organizations front door, and it
now oers entry to a variety of services that benet all mem-
bers. For me, most members of the new crew and our more
global membership, organizations dont exist if they dont
have a Web site. Tis is our point of access.
Te Web site (www.seg.org) has undergone an extensive
transformation over the last couple of years through major
eorts by sta, and additional work by the purpose-built On-
line Committee. It should be a place for one-stop-shopping
Society of Exploration Geophysicists
Presidents Page
The great crew change redux
for geophysicists, providing access to journals, lecture record-
ings, opportunities for online collaboration, information on
the student chapters, and events and event registration. Much
of what you can nd there has been made possible by mul-
tiyear sustaining investments to the SEG Foundation pro-
vided by corporations that understand the importance of a
vibrant online oering: CNPC, ION, Landmark, PGS, IHS,
and Apache.
Yes, it is a work in progress as are many things worthwhile
and broad in scope. If you feel somewhat overwhelmed by
the oerings, I direct you to just a few of oerings you can
nd there:
Te SEG Wiki. Tis content started with Sheris Encyclo-
pedic Dictionary and the intent is to provide an accurate,
timely, and useful collection of information that will prove
to be an invaluable resource for working geophysicists,
educators, and students in the eld of geophysics. Tis is a
living document that will grow with input from the mem-
bership of SEG. Any SEG member can add descriptions
of his or her favorite geophysical terms and methods and
users are encouraged to read, verify, and improve the con-
tents of these pages. Visit Wiki.seg.org.
Recordings of Honorary Lectures and Distinguished Lec-
tures from the past 10 years are available to watch free of
charge. Honorary Lecture tours are region-specic, and
typically you wouldnt have a chance to see the lectures
in regions other than your own. Now you can! And if the
Distinguished Lecture tour didnt stop in your town, or
you just want to see it again, go visit http://www.seg.org/
education/online-education/online-presentations, or from
the SEG home page, see under Resources Multimedia.
Its an online Disneyland for geophysical professionals!
Te complete Continuing Education curriculum. Te cur-
riculum has grown quickly in the last couple of years and
we are working on providing more courses for early-career
geophysicists. Te curriculum consists not only of face-
to-face courses, but also of virtual classes (log on to a live
90-minute seminar), and a broad range of interactive on-
line (34 hour) classes oered through IHRDC.
Information about the SEG Foundation at http://www.seg.
org/web/foundation, which includes a description of nu-
merous programs that advance our eld and scholarship
opportunities.
Tese are just a sampling of how SEG is reaching out
to its members, whether long-standing or new to the club,
around the world. How are these and other programs work-
ing for you? Please send your feedback to me (edithjmiller@
gmail.com) or your friendly SEG sta member, Board mem-
ber, or committee chair (nd them online!). We wont know
unless you tell us.
EDITH MILLER
SEG Director at large
Te SEG Web site is the organizations front
door and it now oers entry to a variety of
services that benet all members.
Visit us at EAGE stand no. 930
slb.com/isometrix
*Mark of Schlumberger. 2013 Schlumberger. 13-se-0062
MARINE ISOMETRIC
SEISMIC TECHNOLOGY
IsoMetrix
First three IsoMetrix surveys acquired in ongoing program
Its reading between the lines that sets our technology apart.
Since its mid-2012 launch, IsoMetrix* marine isometric seismic technology has acquired surveys
in the North Sea, Barents Sea, and South Africa, with the program continuing in Canada.
IsoMetrix technology overcomes the sparse crossline sampling of conventional seismic methods.
The returning wavefeld is fully captured to deliver a 6.25m Th i f ld i f
2
grid of data, providing the clearest,
most accurate images of the subsurface ever recorded.
602 The Leading Edge June 2013
F
or some reason, unknown to me, subjects
come to mind that have no relationship to
anything going on at the time. I was talking
with a friend about sundry things when the
subject of the solid inner core of the Earth was
mentioned. Strange, what?
I wonder how someone decided on that, I asked
innocently.
He replied that they used earthquake data. Tat didnt
satisfy me. Show me the data! I cant imagine a solid core
to the Earth. I can see the molten lava bubbling out of the
ground. I vowed to look into this. To the Internet! To Wiki-
pedia! To the Core of the Earth! It turns out that a seismolo-
gist back in 1936 named Inge Lehmann deduced its presence
from observations of earthquake-generated seismic waves
that reect o the boundary of the inner core and can be de-
tected by sensitive seismographs on the Earths surface. Fairly
straightforward dont you think? A reected seismic wave!
Sensitive seismographs!
Many universities have these large cylinders with an ink
stylus that record data on paper records. Tese are the sen-
sitive seismographs? Most diagrams in text books show the
raypaths of an earthquake. Te rst arrivals are most interest-
ing because the later arrivals are often well below the signal-
to-noise levels. But wait! I know little of this subject. I think
we should think like exploration geophysicists rather than to
take a statement that a reection o of an inner core can be
detected and more importantly ascribed to an interface 3000
miles deep.
Tat reminds me of one of my rst jobs after I spent a
few years in the eld. I was given old single-fold seismic data
to map, located in the Anadarko Basin. Most of the data had
been limited to 23 s. It is clear to me that if we are going to
search for a reection from the Lehmann layer, we will need
more record time! At 20,000 ft/s, we need a recorded time of
about 1584 s (two-way time) for the 3000-mile depth. I am
not sure of the velocity. Does it get faster when the subsurface
is compressed or liquid? Velocity increases with temperature
but the change is small up where we exploration geophysicists
are working. Lets round o. It would take about half an hour
to get the reection back to the surface. But also consider the
other interfaces between the source and the detector array.
Tere are other layers in the Earth, the crust, the Moho and
the Gutenberg discontinuities, and other sundry layers.
Will the reection be a rst arrival? Tis should be easy
for an exploration geophysicist. Tese are rst principles.
Consider a plane through the center of the Earth with the
earthquake on the great circle. Concentric circles inside the
The Leading Edge
From the other side
A column by Lee Lawyer with stories about geophysics and geophysicists
outer circle are the layers of the Earth. How could we as ex-
ploration geophysicists determine that there is a solid core?
We shoot an earthquake at one point on the Earth and fol-
low the rays through the various interfaces. We have sensitive
geophones located encircling the Earth. Remember that the
molten part of the Earth wont support shear.
We could try shear-wave splitting but that is beyond my
imagination. How about converted waves? We shoot at the
surface. Tis generates P- and S-waves, etc. Te S-waves are
attenuated in the molten zone. Te P-wave plows ahead. It
encounters a solid core and a converted shear wave is created,
which follows the parent P-wave. When the S-wave gets to
the other side of the solid core, it converts back to a P-wave,
which then follows the parent P-wave to the surface. Te time
dierence of those two arrivals is a measure of the size and
physical properties of the solid core. Tat seems simple to
me. Whew!
Te little reading I have done says that the inner core is
solid iron-nickel. It can be solid at the temperature and pres-
sure at that depth. Tis is veried in the laboratory (some
lab!). Te reason the iron is at the core is dierentiation,
which has been around a long time; that is, it sank there!
Some say the inner core is growing as the Earth cools. (Oops
... back to global warming.) Pressure is interesting. At the
center of the Earth, gravity is zero (theoretically). Te people
who went to the center of the Earth in the movie should have
been oating around like they do going up to outer space.
Te corrections for a borehole gravity meter are sort of a re-
verse elevation correction or terrain correction. Calculating
the pressure is complicated by the decreasing eect of grav-
ity as you approach the core. Objects weigh less as you get
deeper into the Earth.
As exploration geophysicists, we want to see the data (but
please dont send it)! Where are the data brokers when you
need one?
To contact the Other Side, call or write L. C. (Lee) Lawyer,
Box 441449, Houston, TX 77244-1449 (e-mail LLAWYER@
prodigy.net).
604 The Leading Edge June 2013
S
EG continues to develop a rst-class oering of online
education courses and state-of-the-art online lectures to
better serve our members around the world. High-quality
online education must be available to SEG members no
matter where they live and work, and when their work
requires the most up-to-date geophysical knowledge. Our
success over the past ve years would not have been possible
without the leadership support and investment of CNPC.
As CNPC Assistant President Xu Wenrong stated in
November 2007 when announcing the CNPC investment,
SEG works as a good bridge connecting the east and west
geophysical communities, thus bringing together all ex-
perts in the oil and gas industry and geophysical societ-
ies. Te nancial support from CNPC will reinforce this
bridge, provide investment for the virtual research insti-
tute that is SEG, promote training and technical exchange
among the geophysical community, and build the educa-
tion and training of students. We are pleased that the on-
line education progress of the last ve years is making this
aspiration a reality.
In 2007, SEG Online Education was limited to the
addition of two Distinguished Lectures and one Distin-
guished Instructor Short Course (DISC) each year. Today,
with CNPC support, SEG oers the following:
Distinguished Lectures. Recordings of lectures by globally
recognized experts on current geophysical topics, ranging
from 23 per year.
Honorary Lectures. Recordings of lectures by regionally rec-
ognized experts on locally relevant topics, ranging from
67 per year.
Translations. Select Distinguished and Honorary Lec-
tures are translated to assist members in core regions of
the world. Chinese, Spanish, and Russian are available
today. Translations range from 13 per year.
Distinguished Instructor Short Courses (DISC). Each
year, SEG selects a world-renowned expert to provide
an eight-hour short course at select locations around
the world. Te course is recorded and provided online.
HRDC Introductory Courses. SEG oers more
than 30 introductory courses on a broad range of
geophysical topics.
Technical Program Recordings. Many technical pre-
sentations at SEGs annual meetings from 2010
to 2012 have were recorded and made available
on DVD. Starting with the upcoming 2013 An-
nual Meeting in Houston, they will be made avail-
able on USB. In addition, many of the 2012 pre-
sentations are online and available for individual
purchase. More information is at http://shop.seg.
org/OnlineStore/Courses/OnlineCourses/tabid/178/
Default.aspz?Category=TECH PRES AM12.
The Leading Edge
Foundation News
Advancing the online education experience
CNPC, ION leading the way
Virtual Courses. Live online short courses on emerging
technologies, with student-instructor communication,
have been oered since 2011. Students can participate live
and/or review the recorded presentation at any time, and
courses happen six times per year.
Multilesson Courses. SEG is developing long courses on
topics of fundamental importance to the geophysical
profession. Tese courses follow best practices for online
learning. Te rst such course is Geophysics 101: Seis-
mic Waves in Hydrocarbon Exploration, taught by Leon
Tomsen; the rst lesson for this course is currently avail-
able for members to download.
Te SEG Online eort has become even more of a crit-
ical factor as our organization continues to grow globally,
providing a means for the geophysical community around
the world to interact and to have ready access to a wealth
Growth of SEG Online Education oerings from 20082013.
June 2013 The Leading Edge 605
of information and knowledge. Our progress would not be
as rapid without the support of the visionary corporations
and individuals making sustaining investments through
the SEG Foundation. Te SEG Foundation thanks CNPC
and ION (see sidebar) for their investment in Online Educa-
tion, and Apache for its investment in the new SEG Wiki;
as well as IHS, Rutt Bridges, and Susie and Bob Peebler for
their passionate nancial support of SEGs online vision.
ION renews support to SEG Online Education
In 2007, ION was one of
the rst corporations to invest
in the SEG Online project,
designed to provide news and
resources to SEGs global mem-
bership. In 2012, ION renewed
its ve-year commitment of US
$400,000 and focused its con-
tinued investment on to SEG
Online Education, making the
announcement at SEG Foun-
dations 2012 Donor Luncheon
in Las Vegas.
As we truly have the mem-
bership continuing to globalize,
and as the nature of learning
To participate in SEG Online Education today, visit
www.seg.org/el earning. To learn more about the vari-
ous SEG programs or to help support these programs
through a donation to the SEG Foundation, please visit
www.seg.org/foundation.
NATALIE BLYTHE
Communications Specialist
SEG Foundation
ION Chairman Bob Peebler with Foundation Chair Tomas
Smith after announcing IONs $400,000 renewal commit-
ment to SEG Online Education.
has changed so dramatically
with new students and new
members, it is important to
have the right infrastruc-
ture to provide online sup-
port. Even in the last four
or ve years, things have
progressed in the ways
people communicate. We
need knowledge without
borders, said Bob Peebler,
ION chairman. SEG On-
line is exciting and will
continue to evolve and
grow, and we are excited to
continue our support.
606 The Leading Edge June 2013
A
t the end of January, SEAM Corporation issued the rst
request-for-bids (RFB) on numerical simulations for its
Phase II consortium Land Seismic Challenges. Tis RFB
was for synthetic 3D seismic acquisition using the SEAM
Unconventional Model, a model designed to represent
challenges of imaging and characterizing subtle features of
shale-gas reservoirs set in a realistic stratigraphic section
and overlain by buried topography typical of midcontinent
basins. SEAM is in the process of evaluating the bids received
in response to the RFB and expects to start production
simulations with the Unconventional Model in July.
Two other models are nearing completion in SEAM Phase
II. Te Arid Model places the reservoirs and stratigraphic sec-
tion of the Unconventional Model beneath a 500-m thick
near-surface region with features typical of desert areas. Te
Foothills Model, the last model to be built in Phase II, con-
tains complex fold-and-thrust structures at reservoir depths,
rapid lateral velocity variation at shallow depths, and extreme
topography at the surfaceall of which characterize the geol-
ogy of compressive tectonic zones such as the Andes Moun-
tains of South America. Tis update describes the Arid Model,
which will be the next SEAM model to enter production nu-
merical simulations.
Features of desert terrains that can disrupt seismic explora-
tion include karsts, wadis or buried river channels, sand and
unconsolidated surface sediments, outcropping bedrock, and
topography. Figures 1 and 2 illustrate the challenge of rep-
resenting desert regions in a single Earth model. Figure 1
is a shaded relief map of mountainous desert called jebels,
which contain deeply eroded plateaus and high isolated peaks,
The Leading Edge
SEAM Update
SEAM Phase II: The Arid ModelSeismic
exploration in desert terrains
creating rough topography at scales from kilometers to tens
of meters. Figure 2 is a horizontal slice through a large 3D
seismic image collected on the Saudi Arabian Peninsula; the
slice is at a depth of several hundred meters and covers a region
about 25 15 km in lateral extent. Te quasi-circular objects
stippling the image are buried karsts, caused by acidic rain-
water percolating through and dissolving limestone bedrock.
Like jebel topography, karst structures occupy a wide range of
scales. In this image, the largest ones are 12 km in diameter;
the smallest visible ones are a few tens of meters in diameter.
Smaller karsts are undoubtedly present below the resolution of
the image. Figure 2 shows an outcropping karst large enough
to swallow a eld vehicle.
Representing these desert features at their natural size and
distribution would require an Earth model extending 50100
km with individual features at the meter scale. Numerical
simulations with such a model would be well beyond the ca-
pabilities of todays largest computers using the most advanced
seismic modeling algorithms. Figure 3 shows an early design
Figure 1. Shaded relief map of jebel topography in the Eastern Province
of the Saudi Arabian peninsula (the Arabic word jebel means hill or
mountain). Tese terrains are created when plateaus are deeply eroded
by water and then shaped by winds, and are characterized by reliefs
ranging from gentle slopes to sharp scarps and isolated peaks. Te region
shown is about 50 50 km. Maximum elevation dierence is 250 m.
Figure 2. (top) A horizontal slice at a depth of several hundred meters
through a 3D seismic image collected on the Saudi Arabian Peninsula.
Te dark circular objects in the lower right of the image are about 2 km
in diameter and are limestone dissolution structures called karsts. Te
dots stippling the image are literally hundreds of karsts of smaller sizes,
just within the resolution of the image. (bottom) Photo of a karst break-
ing through the surface.
June 2013 The Leading Edge 607
Figure 3. Early design concept for the near-surface geology of the Arid
Model. Te near-surface region is 10 km 10 km 500 m and incor-
porates a series of features characteristic of desert areas, including karsts,
buried wadis, dry riverbeds, and outcropping bedrock.
Figure 4. Features of the Arid model represented as geobodies. (top)Te
deep karst eld buried at a depth of about 400 m. Te large karst is
about 2 km in diameter; the intersecting karst fairways follow the trends
of buried stream channels. (bottom) A series of braided stream channels
representing a shallow wadi in the rst 100 m of the subsurface.
Figure 5. Slices through the Arid Model. Tese renderings of the com-
pressional-wave velocity eld at 50 and 400 m show the juxtaposition of
braided stream channels with shallow and deep karst elds.
concept for the Arid Model, in which two compromises were
made for a manageable realization. Te rst was to ignore to-
pography; the second, to restrict the modeled volume to the
same size and resolution as the SEAM Unconventional Mod-
elthat is, to a region 10 10 km in lateral extent and 3.75
km in depth, sampled on a uniform 6.25-m grid. (Represent-
ing topography and accurately modeling its eects on seismic
waves will be a top goal of the Foothills model; in addition,
members of the SEAM Phase II consortium are now enter-
taining a proposal to extend the project by an additional year
to add additional features such as topography to Unconven-
tional and Arid models.)
Within these constraints, construction of geologic features
in the digital Arid Model was based on size and shape dis-
tributions seen in natural analogs. Figure 4 shows two major
components of the model: a series of buried stream channels
making up a shallow wadi and a deep karst eld. Tese digital
realizations of geologic structures were made with modeling
software that can populate 3D volumes with a variety of re-
alistic shapes, based on the statistics of natural occurrences,
while retaining each objects identity as an individual geo-
body. Figure 5 shows two shallow depth slices through the
compressional-wave velocity model, illustrating the juxtaposi-
tion of wadi and karst structures. Te subsurface of the Arid
Model below 500 m incorporates the stratigraphic section of
the Unconventional Model, including full orthorhombic elas-
tic anisotropy in the overburden and fractured shale reservoirs
(see the update in the March 2013 TLE).
Of the three geologic models in Phase II, the Arid Model
will likely encompass the largest range of velocity contrasts.
Karsts in arid regions are often air-lled voids, which means
juxtaposing the seismic velocity of air (about 343 m/s) with
that of hard limestone bedrock (35004000 m/s)a 10:1
contrast. Te next big advance in exploration seismic model-
ing will come from routine use of conforming nite-element
or other types of structured, multiresolution grids that can
handle large contrasts with high accuracy without imposing a
ne grid on the entire model.
Acknowledgments and further information: Design of the
near-surface geologic model for the SEAM Phase II Arid Model
608 The Leading Edge June 2013
was carried out in the near-surface modeling technical committee
chaired by Tim Keho of Saudi Aramco. Carl Regone of BP and
Joe Stefani of Chevron made helpful early contributions to un-
derstanding how karst elds could be represented realistically for
numerical seismic modeling. Peter Wang of Schlumberger West-
ernGeco built the nal digital realization of the model for the
SEAM Phase I
Produces Results!!
24 industry participants
$2.6 million budget
40 X 35 X 15-Km model
62,478 free-surface acoustic
shots
2403 absorbing surface shots
Up to 436,921 traces per shot
200 TB uncompressed data
11 "classic datasets
Project expansion with a $2.6
million RPSEA subcontract
under prime U.S. DOE contract
Complete project documentation
SEAM DataPublic Availability Coming Soon...
Watch the SEAM Website for more announcements: www.seg.org/SEAM.
Thanks to these Phase I Participants
SEAM consortium using the Petrel E&P Software Platform. Te
images in Figures 13 are provided courtesy of Saudi Aramco;
those in Figures 4 and 5 are courtesy of Schlumberger.
MICHAEL ORISTAGLIO
SEAM Phase II Project Manager
ENRICHED
DEEPER INSIGHT INTO THE SUBSURFACE TO MEET
YOUR E&P CHALLENGES
With Paradigm, interpretation isnt just a step in the process; it is the process.
Modeling is an integral part of interpretation, with geology, geophysics and
petrophysics combined and energized by a modern, user-friendly interface.
This leads your team to trade linear thinking for dynamic collaboration,
increasing productivity with no compromise in quality. The result? Greater
condence in your E&P business decisions.
Get enriched results. Find out how at pdgm.com
Less time, higher accuracy:
A unied interpretation and
shared earth model.
er accuracy: Less time highe er accuracy: Less time highe
Visit Us at Booth 1230 at EAGE in London
Interpreting and Modeling
Geology, Geophysics, Petrophysics
Nonreflection seismic and inversion of surface and guided waves
610 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Introduction to this special section: Nonrefection seismic and
inversion of surface and guided waves
S
eismic imaging need not be synonymous with or rely
on the presence of refectors in the Earth. Much can be
gleaned from the nonrefected wavefeld. For example, direct
waves map out smooth velocity variations in crosswell seismic
tomography, wide-angle refracted waves play a crucial role
in full waveform inversion (Hole et al., 2005), and surface
waves provide unmatched sensitivity to near-surface shear-
wave velocity structure. Guided waves exist in both cracks
(Korneev, 2008) and boreholes, the latter referred to as tube
waves. Te famous Biot slow wave is itself a guided-wave
phenomenon akin to a tube wave (Norris, 1987). Surface-
wave dispersion has a long history in seismology and was the
frst seismic characteristic to be subjected to an automated
inversion procedure (Dorman and Ewing, 1962).
Surveys based on surface waves provide a low-cost, non-
invasive means of probing the shallow subsurface using either
active sources (Xia et al., 1999) or in passive mode using mi-
crotremors (Aki, 1957, 1965; Louie, 2001; Okada, 2003).
In fact, the past decade has witnessed a revival in the micro-
tremor method because of the realization that ambient noise
correlations are closely related to the surface-wave Greens
function (Campillo and Paul, 2003). Recent work has shown
that other surface-wave observables besides dispersion pos-
sess sensitivity to density in addition to shear-wave velocity
(Lin, et al., 2012). When velocity decreases with depth, the
existence of leaky waves (Ryden and Park, 2004) attests to the
richness of nonrefected-wave phenomena. In a way, it is too
bad that these topics must be collectively identifed by what
they are not (i.e., nonrefection); however, the negative termi-
nology emphasizes the gaping hole in our understanding of
the subsurface if only refections are taken into account.
It is with this background that the special section has
come together. Surface waves feature prominently in many
of the outstanding articles that follow. Guided, leaky, and re-
fracted waves round out the cast of seismic wave types that
appear in this special section. Te articles are grouped into
four subsections:
1) Surface wavesactive and passive
2) Surface wavesindustry data applications
3) Surface wavescase histories
4) Refracted waves and instrumentation
Leading of the subsection on active and passive surface
waves, Xia et al. give an overview of the multichannel analysis
of surface waves (MASW) and its extension to Love waves.
Because of their complete insensitivity to P-wave velocity,
Love waves, in comparison to Rayleigh waves, have simpler
dispersion curves with a more stable inversion. Xia et al. fur-
ther discuss the amplitude of Love waves and the inversion
Matthew M. haney, Anchorage, USA
Rick MilleR, Lawrence, USA
for Q profles from active-source data. Next, Hayashi et al.
determine S-wave velocity structure to depths of 23 km at
sites within the Los Angeles Basin using two-station micro-
tremor array measurements (2ST-MAM). Hayashi et al. com-
pare their method to MASW, borehole velocity logs, and a
community velocity model for Southern California. In the
third article of the section, Behm and Snieder show that Love
waves ofer the possibility of further reducing the inherent
ambiguity in surface-wave imaging compared to the more
commonly used Rayleigh waves. From a continuous broad-
band data set of more than 50 stations that recorded traf-
fc noise, Behm and Snieder were able to measure Love-wave
dispersion and relate its spatial variation to the local geology.
Te authors comment on the unexpectedly high signal-to-
noise ratio of the trafc-induced Love waves over the frequen-
cy band of 1.55 Hz. Tis observation is in harmony with
that reported in the earlier article by Xia et al. for Love waves.
Te subsection ends with an article by Xu et al. that discusses
the application of seismic interferometry to surface-wave im-
aging. Close attention is paid by Xu et al. to the efects of sta-
tion density and spread length on dispersion-curve resolution
with passive techniques. Te authors are able to detect two
low-velocity zones near a tunnel on the Yangtze River that
they interpret as potential landslide planes.
Te subsection on industry applications includes articles
by Boiero et al. and Douma and Haney. Boiero et al. demon-
strate remarkable success fusing leaky waves into the surface-
wave inversion problem. Te authors show defnitively that
subsurface velocity information accurately obtained from
guided P- and S-waves can be jointly inverted with S-wave
velocity information estimated from Rayleigh or Scholte
waves. Boiero et al. compare their results to acoustic full
waveform inversion and point out that the enhanced near-
surface models computed from joint inversion of surface and
guided waves can provide better shallow starting models for
full waveform inversion. In a study of the nonlinearity inher-
ent to dispersion curve inversion, Douma and Haney present
a method designed to quantify the nonuniqueness of accept-
able subsurface models. Te authors compare linearized sur-
face-wave dispersion-curve inversion initialized with diferent
starting models to the results of a nonlinear search. Douma
and Haney make the point that, for bandlimited data, the
highest frequencies defne a shallow unresolved region in the
subsurface in a similar way to how the lowest-frequencies de-
termine the maximum depth of resolution.
Case histories form the theme of the third subsection. Te
utility of MASW to a broad array of problems in civil engi-
neering is demonstrated in Parks article by novel approaches
to site characterization. Park highlights applications target-
ing quantifcation of a ground-shaking hazard, foundation
June 2013 The Leading Edge 611
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
with the potential to provide nearly continuous sampling in
space, in contrast to traditional point sensors. Daley et al.
conduct tests at sites in Alabama, Australia, and Germany in
both surface and borehole environments where high-ampli-
tude ground roll was observed; as result, the authors speculate
that fber optic cables may be well-suited for MASW surveys.
References
Aki, K., 1957, Space and time spectra of stationary stochastic waves,
with special reference to microtremors: Bulletin of the Earthquake
Research Institute: Tokyo University, 25, 415457.
Aki, K., 1965, A note on the use of microseisms in determining the
shallow structures of the Earths crust: Geophysics, 30, no. 4,
665666, http://dx.doi.org/10.1190/1.1439640.
Campillo, M. and A. Paul, 2003, Long-range correlations in the dif-
fuse seismic coda: Science, 299, no. 5606, 547549, http://dx.doi.
org/10.1126/science.1078551.
Dorman, J. and M. Ewing, 1962, Numerical inversion of seismic sur-
face wave dispersion data and crust-mantle structure in the New
York-Pennsylvania area: Journal of Geophysical Research, 67, no.
13, 52275241, http://dx.doi.org/10.1029/JZ067i013p05227.
Hole, J. A., C. A. Zelt, and R. G. Pratt, 2005. Advances in controlled-
source seismic imaging: Eos, Transactions, American Geophysical
Union, 86, 177 and 181.
Korneev, V. A., 2008, Slow waves in fractures flled with viscous fuid:
Geophysics, 73, no. 1, 17, http://dx.doi.org/10.1190/1.2802174.
Lin, F.-C., B. Schmandt, and V. C. Tsai, 2012, Joint inversion of
Rayleigh wave phase velocity and ellipticity using USArray: con-
straining velocity and density structure in the upper crust: Geo-
physical Research Letters, 39, no. 12, L12303, http://dx.doi.
org/10.1029/2012GL052196.
Louie, J. N., 2001, Faster, better: shear-wave velocity to 100 meters
depth from refraction microtremor arrays: Bulletin of the Seis-
mological Society of America, 91, no. 2, 347364, http://dx.doi.
org/10.1785/0120000098.
Norris, A., 1987, Te tube wave as a Biot slow wave: Geophysics, 52,
no. 5, 694696, http://dx.doi.org/10.1190/1.1442336.
Okada, H., 2003, Te microtremor survey method: SEG, http://
dx.doi.org/10.1190/1.9781560801740.
Ryden, N. and Park, C. B., 2004, Surface waves in inversely dispersive
media: Near Surface Geophysics, 2, 187197.
Xia, J., R. D. Miller, and C. B. Park, 1999, Estimation of near-surface
shear-wave velocity by inversion of Rayleigh waves: Geophysics,
64, no. 3, 691700, http://dx.doi.org/10.1190/1.1444578.
properties of a proposed nuclear plant site, and improving the
understanding of highway stability. Park describes the advan-
tages of combining active and passive surveys to sense shal-
low and deep structure in South Africa. A second article in
this special section with Boiero as lead author describes two
feld cases from Scandinavia concerning water resource map-
ping and site characterization for tunnel construction. Boiero
et al. analyze surface-wave dispersion curves with a Monte
Carlo inversion technique and in the process address the
nonuniqueness problem, thereby avoiding falling into local
minima of the misft function. Comparisons of the surface-
wave profles are made to borehole data and the results of seis-
mic refection surveying. Several case studies by Suto describe
MASW surveys at landfll sites in Australia where objectives
include precise delineation of the base of a landfll, monitor-
ing of compaction, characterization of an incipient sinkhole,
and gauging ground strength in lieu of a large commercial
development. Suto discusses a study in which disagreement
exists between results from MASW data and dynamic cone
penetration tests. In closing, the author speculates as to why
shear-wave velocity analysis of the near surface is yet to be
embraced by geotechnical engineers.
Articles in the fnal grouping push the state-of-the-art in
near-surface refraction imaging and instrumentation. Palmer
discusses the generalized reciprocal method of refraction im-
aging and adaptations of the method to work with common-
ofset gathers (COG). Te adaptations ofer benefts in the
form of convenient determination of the crossover distance
and improved resolution at the base of the weathering zone,
in particular for low-angle thrust faults. An advantage of the
COG adaptations lies in their ability to rapidly assess large
volumes of refraction data. Ivanov et al. present a study on
the joint analysis of refractions with surface waves, or JARS
method. In contrast to conventional refraction tomography,
the JARS method addresses the inherent nonuniqueness of
the inverse refraction problem by defning an initial V
s
model
based on MASW. Ivanov et al. describe several feld applica-
tions of the method, including one from a levee in southern
New Mexico where a velocity inversion is resolved at 10-m
depth. Daley et al. give an account of feld testing of a fber
optic cable for seismic applications. Te fber optic cable is
an exciting new development in near-surface instrumentation
Nonreflection seismic and inversion of surface and guided waves
612 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Near-surface shear-wave velocities and quality factors derived
from high-frequency surface waves
S
hear (S)-wave velocities of near-surface materials can
be derived from inverting the dispersive phase velocity
of high-frequency ( 2 Hz) surface (Rayleigh and/or Love)
waves (e.g., Song et al., 1989). Multichannel analysis of
surface waves (MASW) uses phase information of high-
frequency Rayleigh waves recorded on vertical component
geophones to determine S-wave velocities (Miller et al.,
1999). Multichannel analysis of Love waves (MALW) uses
phase information of high-frequency Love waves recorded
on horizontal (perpendicular to the direction of wave
propagation) component geophones to determine S-wave
velocities (Xia, 2012b). Both MASW and MALW possess
stable and e cient inversion algorithms to invert phase
velocities of surface waves but MALW has some attractive
advantages: (1) Love-wave dispersion curves are simpler
than those of Rayleigh waves; (2) dispersion images of Love-
wave energy have a higher signal-to-noise ratio and are more
focused than those generated from Rayleigh waves; and (3)
inversion of Love-wave dispersion curves is less dependent on
initial models and more stable than from Rayleigh waves.
S-wave velocities of near-surface materials that are de-
rived from high-frequency surface waves utilize only the sur-
face waves phase information. Te feasibility of also using the
amplitude information to estimate near-surface quality factors
(Q
s
and/or Q
p
) has been studied (Xia et al., 2002a and 2012b).
Quality factors (Q
p
, Q
s
) can be obtained by inverting attenua-
tion coe cients calculated from the amplitude of high-frequen-
cy surface (Rayleigh and/or Love) waves. Inversion of attenua-
tion coe cients of Love waves to estimate Q
s
is simpler than for
Rayleigh waves because they are independent of Q
p
.
Methods that use high-frequency surface waves to estimate
near-surface S-wave velocities and quality factors are noninva-
sive, nondestructive, environmental-friendly, low-cost, fast,
and in-situ methods. Tis article describes key aspects of the
multichannel analysis of high-frequency surface-wave methods
through discussion of inversion systems and real-world exam-
ples. Challenges in applying high-frequency surface-wave meth-
ods in practice are also presented the end of the article.
Inversion of Rayleigh-wave phase velocities
Phase information of high-frequency surface waves can be
used to estimate near-surface S-wave velocities. Te Ray-
leigh-wave phase velocity of a layered Earth model is a func-
tion of frequency and four properties: P-wave velocity (V
P
),
S-wave velocity (V
S
), density (), and thickness (h) of layers.
Rayleigh-wave phase velocity, c
Rj
, is determined by a charac-
teristic equation F (Schwab and Knopo, 1972), in its non-
linear, implicit form F( f
j
, c
Rj
, V
S
, V
P
, , h) = 0, (j = 1, 2, ...,
m, f is frequency).
Analysis of the Jacobian matrix provides a measure of the
dispersion-curve sensitivity to Earth properties (Xia et al.,
JIANGHAI XIA, CHAO SHEN, and YIXIAN XU, China University of Geosciences (Wuhan)
1999). S-wave velocity is the dominant inuence on a dis-
persion curve so, for our purposes, only S-wave velocities are
considered unknowns in their inversion. A key consideration
Figure 1. (a) 48-channel Rayleigh-wave data acquired in Wyoming
along a WE line. Te data were acquired with the source at the west
end of the line. (b) Dispersion property of Rayleigh-wave was clearly
shown in the f-v domain. Te solid dots represent phase velocities that
were used in inversion to estimate near-surface S-wave velocities.
June 2013 The Leading Edge 613
Nonreflection seismic and inversion of surface and guided waves
is the degree that solution nonuniqueness will increase if the
thickness values are treated as unknowns. Tis is a reason
thickness should be dened as an input value in the inversion
of Rayleigh-wave phase velocities.
A Rayleigh-wave survey was conducted in Wyoming to
determine S-wave velocities in near-surface materials (upper
7 m). Rayleigh-wave energy was strong and easily recognized
(Figure 1a and Figure 1b). By following the peaks of the energy
trend in Figure 1b, we can pick up phase velocities at dierent
frequencies (dots in Figure 1b). Inversion of the phase velocities
with an initial model determined with the suggested formula by
Xia et al. (1999) provided an S-wave velocity model that was
supported by borehole measurements (Figure 2).
S-wave velocity proles derived by MASW compared fa-
vorably to direct borehole measurements at numerous sites
(e.g., Xia et al., 2002b). On average, the dierence between
MASW-calculated V
S
and borehole-measured V
S
is less than
15%. MASW not only provides accurate near-surface S-wave
velocities but in some geologic settings it may be the only
method to obtain S-wave velocity informatione.g., a set-
ting with a dipping layer where converted P-wave energy
could be dominant in a shear-wave refraction survey (Xia et
al., 2002b) or a velocity inversion (a higher-velocity layer on
the top of a lower-velocity layer), resulting in no refraction
event from the interface.
Accuracy of S-wave velocity models derived from MASW
can be improved when inversion simultaneously includes the
fundamental- and high-mode phase velocities of Rayleigh
waves (Xia et al., 2003). Moreover, the inversion is more
Figure 2. S-wave velocities from inverted S-wave velocities labeled
as MASW W-E (E) and MASW W-E (W) that represent sources at
the east and west ends of the line, respectively, and the suspension log.
Te S-wave velocity model labeled MASW W-E (W) was inverted
from phase velocities shown as dots in Figure 1b. Results of SH-wave
reection data were also shown in the gure, which were interpreted as
velocities of converted waves except for the rst layer (adapted from Xia
et al., 2002b).
Figure 3. (a) SH-wave data acquired in the grassy area in front of
the KGS building. (b) An image of Love-wave energy generated with
high-resolution linear Radon transform (Luo et al., 2008). Te sharp
Love-wave energy trend makes picking phase velocities easier and more
accurate. Te solid dots represented phase velocities that were used in
inversion to estimate near-surface S-wave velocities.
614 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
stable when higher-mode data are included in the process.
Tis is because high-mode Rayleigh-wave data behave like
Love waves, which as we know are independent of P-waves.
Numerical modeling results (Xia et al., 2003) indicate that
the eects of P-wave energy are negligible in high modes of
Rayleigh waves.
Inversion of Love-wave phase velocities
Te Love-wave phase velocity of a layered Earth model is in-
dependent of P-wave velocity and is a function of frequency
and three properties: SH-wave velocity (V
sh
), density (), and
thickness (h) of layers. S-wave velocity associated with Love
waves is only the SH. Love-wave phase velocity, c
Lj
, is deter-
mined by a characteristic equation F (Schwab and Knopo,
1972), in its nonlinear, implicit form F( f
j
, c
Lj
, v
hs
, , h) = 0, (j
= 1, 2, ..., m, f is frequency).
Some (Renalier et al., 2011; Xia et al., 2012b) suggest
estimating SH-wave velocity using Love-wave inversion for
near-surface applications may become more appealing than
Rayleigh-wave inversion because of the simplicity of Love-
wave dispersion curves. Numerical modeling results suggest
the independence of Love-wave energy from P-waves reduces
the chances that Love-wave dispersion curves will be com-
plicated by mode kissing (an undesired and frequently oc-
curring phenomenon in Rayleigh-wave analysis which often
results in mode misidentication).
Real-world examples (Xia et al., 2012b) demonstrate two
other advantages of inverting Love-wave dispersion curves.
Love-wave dispersion images in the frequency-velocity (f-v)
domain have a higher signal-to-noise ratio and are more fo-
cused than equivalent images generated from Rayleigh waves.
Tis advantage is generally related to the long geophone
spreads commonly used for SH-wave refraction surveys. Pick-
ing Love-wave phase velocities is easier and more accurate
from the images of Love-wave energy from longer-oset data
because they possess higher resolution than those from near-
oset data. In addition, inversion of Love-wave dispersion
curves is more stable and less dependent on initial models
than Rayleigh waves.
Te same algorithm used for the inversion of Rayleigh-
wave dispersion curves displayed previously was used to in-
vert Love-wave phase velocities. Love-wave data (Figure 3a)
were acquired on the grass-covered lawn at the Kansas Geo-
logical Survey with 40 14-Hz horizontal-component geo-
phones oriented perpendicularly (SH) along the same survey
line where Rayleigh-wave data were collected previously (Xia
et al., 1999). Geophones were spaced at 1-m intervals. A po-
larized seismic energy pulse was provided by delivering an im-
pact to a block, oriented in a direction parallel to that of the
geophones maximum sensitivity, with a 6.3-kg hammer. A
resolution overtone image (Figure 3b) was generated from the
shot gather. Phase velocities from the fundamental mode can
be picked from 12 to 45 Hz on the image (dots in Figure 3b).
An initial model used for inversion of the dispersion curve
(dots in Figure 3b) was general because inversion of Love-
wave phase velocities is not sensitive to initial models. Obvi-
ously, in this example, an initial value for the rst layer could
be more accurately determined (Figure 4) because phase ve-
locities approach the S-wave velocity of around 110 m/s for
the rst layer at the high-frequency range.
After ve iterations, the root-mean-squares error was re-
duced from 280 m/s to 5 m/s. Te nal inverted S-wave ve-
locity model (Figure 4) compares favorably to the borehole
measurements. We note that relatively large dierences exist
between inverted results and borehole measurements in the
top layers (03 m) and layers from 6 to 12 m (Figure 4). Te
asymptote of the Love-wave energy (Figure 3b) clearly indicates
the SH-wave velocity of the top layer is consistent with the in-
verted velocity of 110 m/s. Te dierence observed between 6
and 12 m may be indicative of S-wave velocity anisotropy.
Figure 4. S-wave velocity models: the initial model, the inverted
model from the picked phase velocities shown by solid dots in Figure
3b, and borehole measurements. See the text for explanation of the
discrepancy between the inverted model and borehole measurements.
Figure 5. An example of an L-curve generated with dierent
damping factors. A wise choice of solutions should be within a trade-o
zone.
Building Partnerships.
Every company is different, every project unique. At Weatherford, we
believe in getting every job right, listening to your concerns, and working
with you to meet your needs and your expectations.
Delivering Results.
We dont succeed until you dothats the bedrock principle of our
CVTJOFTTBOEUIFNFBTVSFPGPVSDPNNJUNFOU'SPNTUBSUUPOJTI
our resources are focused on your objectives.
Contact and collaborate with us at
weatherfordcommunications@weatherford.com.
Stop by the Weatherford stand, #220, at the 2013 EAGE Conference in London.
Formation Evaluation
|
Well Construction
|
Completions
|
Production
2013 Weatherford. All rights reserved.
616 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Inversion of attenuation coe cients of Rayleigh waves
Amplitude information from high-frequency surface waves
can be used to estimate near-surface quality factors (Q).
Laboratory experiments (Johnston et al., 1979) showed that
Q may be independent of frequency over a broad bandwidth
(10
2
10
7
Hz), especially for some dry rocks. Te quality fac-
tor as a function of depth, which is directly related to the
material damping ratio D (= 0.5Q
1
) (Rix et al., 2000), is of
essential interest in earthquake engineering (Kramer, 1996),
geotechnical engineering, groundwater, and environmental
Figure 6. (a) Raw Rayleigh-wave data used for determining near-surface quality factors. (b) Attenuation coe cients from 23 to 75 Hz shown
in diamonds were determined from data shown in (a) and those in solid squares were determined with inverted quality factors shown in (d). (c)
Te L-curve used to determine the trade-o value of a damping factor (a solid square). Te numbers next to the symbols are values of damping
factors associated with the solutions. (d) A trade-o Q
s
model (in the logarithmic scale) under the constraint 0 < Q
s
< 100.
studies, as well as in oil exploration and earthquake seis-
mology. To fully understand seismic-wave propagation in
the Earth, the quality factors are parameters that must be
known.
Modeling results (Xia et al., 2002a) suggest it is feasible
to estimate Q
P
and Q
S
in a layered Earth model by inverting
Rayleigh-wave attenuation coe cients when V
S
/V
P
reaches
0.45. Only Q
S
can be estimated from Rayleigh-wave attenu-
ation coe cients when V
S
/V
P
is less than 0.45, which is a
common situation for near-surface materials.
June 2013 The Leading Edge 617
Nonreflection seismic and inversion of surface and guided waves
Shallow Rayleigh-wave data (Figure 6a) were acquired in
an arid region of the southwestern United States during the
spring of 2010. Te objective of this survey was to determine
the seismic properties of near-surface sediments. Rayleigh-
wave data were acquired using a towed streamer that con-
sisted of 24 4.5-Hz vertical component geophones with a
nearest-oset of 41 m, a geophone interval of 1.2 m, and an
accelerated weight drop as a seismic source.
We calculated attenuation coe cients (triangles in Figure
6b) using the assumption Q
P
= 2Q
S
during inversion. S-wave
velocities of the model were determined by inverting phase
velocities of Rayleigh waves before inversion of attenuation
coe cients.
Inversion of attenuation coe cients of Love waves
Information about the amplitude of high-frequency Love waves
can be used to estimate near-surface quality factors (Q
s
). Un-
like Rayleigh waves, attenuation coe cients (amplitude) of Love
waves are independent of the quality factors of P-waves and are
functions of quality factors of Love waves. In theory, fewer pa-
rameters make the inversion of attenuation coe cients of Love
waves more stable and reduce the degree of nonuniqueness.
SH-wave refraction data (Figures 7a) acquired in Wyo-
ming were used to estimate near-surface quality factors. Te
data were acquired using 48 28-Hz horizontal-component
geophones oriented NS. Geophones were deployed at a 0.9-m
interval along a WE spread. A polarized seismic pulse was
generated by a 6.3-kg hammer impacting a coupled plate (S-
wave source plate) perpendicular to the spread. Attenuation
coe cients from 10 to 45 Hz (diamonds in Figure 7b) were
determined from data shown in Figure 7a. As mentioned in
the previous sections, S-wave velocities were estimated by in-
verting phase velocities of Love waves using MALW.
We used the same layer model in inverting attenuation
coe cients in terms of thickness of layers as inverting phase
velocities of Love waves using MALW. Te nal model pos-
sesses a reasonable range of quality factors for near-surface
materials. Because of the constraints applied to the inverse
system, we expect the t in the data space to decrease. Tis
is the price we have to pay for no a-priori information on
quality factors at this site. Calculated attenuation coe cients
(solid squares in Figure 7b) from the inverted quality factor
model (Figure 7c) t measured data (diamonds in Figure 7b)
reasonably well.
Figure 7. (a) 48-channel SH-wave refraction data in Wyoming. (b) Attenuation coe cients from 10 to 45 Hz (diamonds) from data shown in
(a) and attenuation coe cients (solid squares) determined with the inverted model shown in (c). (c) Te inverted Q
s
model (in the logarithmic
scale) under the variable constraints (see the text for details).
618 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Other developments and challenges
High-frequency surface-wave methods have received in-
creased attention in the near-surface geophysics and geo-
technical communities in the last 20 years. Te noninvasive,
nondestructive, e cient, environmental friendly, and low-
cost nature of these methods matched with their proven suc-
cess in environmental and engineering applications and the
upswing in interest should not be unexpected. Tese meth-
ods are viewed by the near-surface geophysics community
as one of most promising techniques for determining elastic
properties. However, they face unique problems related to
the extremely irregular velocity variations common in near-
surface geology or man-made structures (e.g., highway, foun-
dation, dam, levee, jetty, etc.). It is these irregularities that
inhibit solving this characterization problem with techniques
or algorithms that are in wide use in earthquake seismology
or oil/gas seismic exploration.
Calculating dispersion curves by existing algorithms may
fail for velocity models with velocity inversions (a high-ve-
locity layer on the top of a low-velocity layer). Tere are two
velocity models that are most common in near-surface appli-
cations: a low-velocity half-space model and a high-velocity
surface-layer model. Te low-velocity half-space model re-
sults in a complex matrix that has no roots in the real number
domain. Terefore, based on current algorithms, no phase ve-
locities can be calculated in certain frequency ranges. A work-
around for this dilemma is to use only the real part of the root
of the complex number (Pan et al., 2013). It is well-known
that surface-wave phase velocities approach about 92% of
the surface layer S-wave velocity when wavelengths of surface
waves are much shorter than the thickness of the rst layer.
For the high-velocity surface-layer model, however, phase ve-
locities calculated using the current algorithms approach, in
a high-frequency range, the value associated with the lowest
S-wave velocity of the model rather than with the S-wave ve-
locity of the surface layer. We are working on this problem.
Horizontal resolution of an S-wave velocity prole de-
rived from inversion of surface-wave phase velocities is critical
in imaging subsurface in near-surface applications, although
it is not a serious challenge for some applications, such as
seismic zonation (Yilmaz et al., 2009). Te horizontal resolu-
tion is mostly inuenced by the length of the receiver spread.
A short receiver spread may be the ultimate solution for in-
creasing horizontal resolution. For this to become common
practice, algorithms need to be developed that could generate
an image of surface-wave energy with high resolution in the
f-v domain or are stable in terms of handling noise.
Accurately determining attenuation coe cients still re-
mains a challenge. It requires highly consistent geophones
and careful placement of those geophones. To the reduce ef-
fects of geophone consistency and placement, averaging mul-
tiple traces in the frequency domain may be useful.
References
Johnston, D. H., M. N. Toksz, and A. Timur, 1979, Attenuation of
seismic waves in dry and saturated rocks: II. Mechanisms: Geo-
physics, 44, no. 4, 691711, http://dx.doi.org/10.1190/1.1440970.
Kramer, S. L., 1996, Geotechnical earthquake engineering: Prentice Hall
Luo, Y., J. Xia, R. D. Miller, Y. Xu, J. Liu, and Q. Liu, 2008, Rayleigh-
wave dispersive energy imaging by high-resolution linear Radon
transform: Pure and Applied Geophysics, 165, no. 5, 903922,
http://dx.doi.org/10.1007/s00024-008-0338-4.
Miller, R. D., J. Xia, C. B. Park, and J. Ivanov, 1999, Multichannel
analysis of surface waves to map bedrock: Te Leading Edge, 18,
no. 12, 13921396, http://dx.doi.org/10.1190/1.1438226.
Pan, Y., J. Xia, and C. Zeng, 2013, Verication of correctness of us-
ing real part of complex root as Rayleigh-wave phase velocity by
synthetic data: Journal of Applied Geophysics, 88, 94100, http://
dx.doi.org/10.1016/j.jappgeo.2012.09.012.
Renalier, F., G. Bivre, D. Jongmans, M. Campillo, and P.-Y. Bard,
2011, Clayey landslide investigations using active and passive V
S
measurements, in R. Miller, J. Bradford, and K. Holliger, eds., Ad-
vances in near-surface seismology and ground-penetrating radar:
SEG, 397414, http://dx.doi.org/10.1190/1.9781560802259.
Rix, G. J., C. D. Lai, and A. W. Spang Jr., 2000, In situ measurement
of damping ratio using surface waves: Journal of Geotechnical
and Geoenvironmental Engineering, 126, no. 5, 472480, http://
dx.doi.org/10.1061/(ASCE)1090-0241(2000)126:5(472).
Schwab, F. A. and L. Knopo, 1972, Fast surface wave and free mode
computations, in B.A. Bolt, ed., Methods in computational phys-
ics: Academic Press, 87180.
Song, Y. Y., J. P. Castagna, R. A. Black, and R. W. Knapp, 1989, Sensi-
tivity of near-surface shear-wave velocity determination from Ray-
leigh and Love waves: 59th Annual International Meeting, SEG,
Expanded Abstracts, 509512.
Xia, J., R. D. Miller, and C. B. Park, 1999, Estimation of near-surface
shear-wave velocity by inversion of Rayleigh waves: Geophysics,
64, no. 3, 691700, http://dx.doi.org/10.1190/1.1444578.
Xia, J., R. D. Miller, C. B. Park, and G. Tian, 2002a, Determining
Q of near-surface materials from Rayleigh waves: Journal of Ap-
plied Geophysics, 51, no. 2-4, 121129, http://dx.doi.org/10.1016/
S0926-9851(02)00228-8.
Xia, J., R. D. Miller, C. B. Park, E. Wightman, and R. Nigbor, 2002b,
A pitfall in shallow shear-wave refraction surveying: Journal of Ap-
plied Geophysics, 51, no. 1, 19, http://dx.doi.org/10.1016/S0926-
9851(02)00197-0.
Xia, J., R. D. Miller, C. B. Park, and G. Tian, 2003, Inversion of
high frequency surface waves with fundamental and higher modes:
Journal of Applied Geophysics, 52, no. 1, 4557, http://dx.doi.
org/10.1016/S0926-9851(02)00239-2.
Xia, J., Y. Xu, R. D. Miller, and J. Ivanov, 2012a, Estimation of near-
surface quality factors by constrained inversion of Rayleigh-wave
attenuation coe cients: Journal of Applied Geophysics, 82, 137
144, http://dx.doi.org/10.1016/j.jappgeo.2012.03.003.
Xia, J., Y. Xu, Y. Luo, R. D. Miller, R. Cakir, and C. Zeng, 2012b, Ad-
vantages of using multichannel analysis of Love waves (MALW) to
estimate near-surface shear-wave velocity: Surveys in Geophysics,
33, no. 5, 841860, http://dx.doi.org/10.1007/s10712-012-9174-2.
Yilmaz, O., M. Eser, and M. Berilgen, 2009, Applications of engineer-
ing seismology for site characterization: Journal of Earth Science,
20, no. 3, 546554, http://dx.doi.org/10.1007/s12583-009-0045-9.
Acknowledgements: Tis work is partly supported by the National
Natural Science Foundation of China (NSFC), under Grant No.
41274142.
Corresponding author: jianghai_xia@yahoo.com
A TGS COMPANY
See the Value
Discover how Arcis
can deliver your solution with
industry-leading service and quality.
Custom Integrated Geotechnical
Solutions Adds Value
Data Library
Project Management
Reservoir Analysis
Data Marketing
Geo-Consulting Services
Land / Marine / Time / Depth Processing
Alberta
Manitoba
British Columbia
Saskatchewan
Nunavut
Northwest Territories
Yukon Territory
2D Seismic
3D Seismic
Nonreflection seismic and inversion of surface and guided waves
620 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Estimating deep S-wave velocity structure in the Los Angeles Basin
using a passive surface-wave method
T
his article summarizes a passive surface-wave method
that uses only two sensors and its application to the
estimation of deep S-wave velocity structure.
Tree-dimensional S-wave velocity structure to a depth
of several kilometers has a large eect on long-period ground
motion in tectonic basins, such as the Los Angeles (LA) Ba-
sin. Recent studies of long-period ground motion in the LA
Basin (e.g., Hatayama and Kalkan, 2012) show that observed
ground motion in some areas cannot be explained by the S-
wave velocity models in current use. Most studies of basin
velocity structure rely on geologic information, surface and
borehole geophysical data, and observed earthquake records
to deduce or measure seismic velocities. Geophysical data
and seismic stations commonly used for velocity analysis are
sparsely distributed and most well data are too shallow to
characterize deep S-wave velocity structure. To establish more
accurate basin velocity structure, there is a need for more
densely distributed deep S-wave velocity data.
Active and passive surface-wave methods have been in-
creasing in popularity over the last 15 years. Te passive
surface-wave method or microtremor array measurements
(Okada, 2003), in which surface waves from ambient noise
are used, is particularly attractive to estimate deep S-wave ve-
locity structure. Tis is because the method does not require
an articial source and the depth of investigation can easily
be extended by increasing the size of the array, providing the
necessary low-frequency microtremor energy is present.
Large-scale microtremor array measurements have been
widely used in the last 15 years in Japan for estimating S-wave
velocity structure to a depth of several kilometers. In these
investigations, triangle arrays with dimensions of several kilo-
meters are used for calculating Rayleigh-wave phase velocity
in the frequency range of 0.21 Hz.
Many practitioners use the spatial autocorrelation (SPAC)
method (Aki, 1957) for calculating phase velocities from am-
bient noise data and the method requires at least four sensors
placed on the center and vertices of an equilateral triangle.
Margaryan et al. (2009) show that SPAC using only two sen-
sors yields almost identical phase velocities as triangle arrays
with four sensors. Recently, Hayashi and Underwood (2012a,
2012b) and Hayashi et al. (2013) show that S-wave velocity
proles down to a depth of 23 km can be determined by
using two sensors and SPAC in the San Francisco South Bay
area in California and Seattle and Olympia in Washington.
SPAC using a small number of sensors (less than four) enables
acquisition of microtremor array data much more e ciently
and is, therefore, more cost-eective.
To evaluate the applicability of SPAC using two sensors,
KOICHI HAYASHI, Geometrics
ANTONY MARTIN, GEOVision
KEN HATAYAMA, National Research Institute of Fire and Disaster
TAKAYUKI KOBAYASHI, OYO Corporation USA
herein referred to as two-station microtremor array measure-
ments (2ST-MAM), we performed measurements at four
sites in the southwestern portion of the LA Basin and com-
pared resulting S-wave velocity proles with existing borehole
S-wave velocity logs and a 3D seismic velocity model based
on geologic structure models and sparse geophysical data.
Data acquisition and processing
Two-station microtremor array measurements (2ST-MAM)
were performed at four sites (Carson, South Gate, Willow-
brook, and Manhattan Beach) in the southwestern portion of
LA Basin (Figure 1). At each site, one sensor was established
at a xed location with microtremor data acquired at this lo-
cation for the duration of the survey. Microtremor data were
acquired with a second sensor using variable station separa-
tions ranging from 10 to 3430 m, 3802 m, 820 m, and 2645
m at the Carson, South Gate, Willowbrook, and Manhattan
Beach sites, respectively. At each measurement location, we
recorded microtremor data for several hours per site using
intervals of 10 minutes to 1 hour and a 10-ms sample rate.
As the separation of sensors increased, the record length of
ambient noise was increased (Figure 2). Acquisition of data
from smaller arrays (less than 300 m) and larger arrays was
performed during the day and night, respectively. Receivers
were placed in relatively quiet locations such as in parks or
residential areas.
Seismographs used for this study were McSEIS-MT Neo
by OYO Corporation. Te units are self-contained with
a single set of three-component accelerometers and a GPS
clock. Te GPS clock allows synchronizing of multiple seis-
mographs.
A vertical component of ambient noise is used in pro-
cessing. Recorded data were divided into several time blocks
with overwraps. Each block consists of 8192 samples with a
data length of 81.92 s. Several blocks containing nonstation-
ary noise were rejected before processing. Coherence was cal-
culated for each block with the real part of coherence for all
blocks averaged to obtain the spatial auto correlation (SPAC).
A velocity that minimizes the error in between SPAC and
a Bessel function (rst kind, zero order) can be considered
as the phase velocity. An outline of the processing based on
SPAC using two sensors is summarized in Figure 3.
Example of spatial autocorrelation
Figure 4 shows an example of spatial autocorrelations at the
Manhattan Beach site. It is evident that there are clear dis-
tinctions between the coherence for various sensor spacings
(Figure 4a). Frequency-based coherences (Figure 4b) have
June 2013 The Leading Edge 621
Nonreflection seismic and inversion of surface and guided waves
At the Carson, South Gate, and Manhattan Beach sites,
the longest wavelengths associated with observed phase veloc-
ities are greater than 5 km and may include information on
the S-wave velocity structure to a depth of 23 km. Te maxi-
mum Rayleigh-wave phase velocities at these sites are about
2000 m/s, which implies that S-wave velocities at the depth
of 23 km are greater than 2000 m/s. In general, the longest
wavelength obtained applying SPAC is 24 times the receiver
separation. Te maximum receiver separations at these sites
are 3430 m, 3802 m, and 2645 m, respectively and, therefore,
it is reasonable that the maximum wavelengths at these sites
are 510 km. At the Willowbrook site, the longest wavelength
was only about 3 km because of a lack of large sensor spacing
good agreement between the ob-
served coherences and the theoretical
Bessel functions. Dispersion curves
are clearly evident in the frequency
range of 0.250.5 Hz in the low-fre-
quency image and from 0.5 to 20 Hz
in the high-frequency image (Figure
4c).
Clear variation of coherence and
phase velocity are observed in the
frequency range of 0.20.4 Hz in
the low-frequency chart and from 1
to 5 Hz in the high-frequency chart.
Tese variations correspond to deep
(several kilometers) and shallow (sev-
eral tens of meters) velocity structure,
respectively. Te results demonstrate
the applicability of the 2ST-MAM to
both deep and shallow investigations.
Dispersion curves
Small-scale microtremor array mea-
surements and active surface-wave
measurements (MASW) were also performed at the Carson
site (Dolphin Park). Triangle and linear arrays were used for
small-scale passive data acquisition. Te maximum size of
the triangle array was 60 m and the length of the linear array
was 115 m. In the triangular array, seven geophones with a
natural frequency of 1 Hz were used for data acquisition. In
the linear array, 24 geophones with a natural frequency of 4.5
Hz were deployed on 5-m intervals.
Passive surface-wave data were recorded for about 10
minutes with a 2-ms sample rate. Te conguration of the
linear array was similar to that of a popular passive surface-
wave method often referred to as the refraction microtremor
method (ReMi
TM
).
In the active surface-wave method (MASW), 24 geo-
phones (4.5 Hz) were spaced along a single line at 1-m inter-
vals. A 10-kg sledge hammer was used as the energy source
for the MASW testing. Figure 5 shows the comparison of
dispersion curves together with corresponding wavelength.
It is clear that a dispersion curve obtained by the 2ST-
MAM agrees with that of the small passive arrays (triangle
and linear) at the overlapping frequency range of 23 Hz.
Maximum wavelengths obtained using the 2ST-MAM, the
small arrays, and the MASW were about 10 km, 200 m, and
50 m, respectively (Figure 5). As a rule of thumb, 1/2 to 1/3
of the maximum Rayleigh-wave wavelength is indicative of
the penetration depth of the surface-wave method. Te ex-
tremely deep penetration capability of the 2ST-MAM is obvi-
ous when compared to conventional surface-wave methods,
such as MASW or ReMi (Figure 5).
At all sites, phase velocities were obtained at the frequency
range from 0.413 Hz. We estimated Rayleigh-wave phase
velocities down to a frequency of 0.25 Hz, except for the Wil-
lowbrook site which had maximum frequencies varying from
13 to 30 Hz (Figure 6).
Figure 2. Example of 2ST-MAM array conguration (Carson). Only
receivers more than 900 m from the xed receiver are shown.
Figure 1. Sites of investigation. Note that the velocity log associated with the Manhattan Beach
site is several kilometers north of the 2ST-MAM array.
622 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
and, therefore, the surface-wave data likely provide information
only on S-wave velocity structure to a depth of about 1 km.
Te phase velocities at the Manhattan Beach site are lower
than those of the other sites over the frequency range from
0.4 to 1.5 Hz. Te phase velocities at the Willowbrook and
the Manhattan Beach sites are much faster than those of other
sites across the frequency range of 310 Hz.
Inversion
An inversion scheme (Suzuki and Yamanaka, 2010) was ap-
plied to the observed dispersion curves to develop S-wave
velocity proles for the four sites. During the inversion, the
phase velocities of the dispersion curves were used as obser-
vation data with the unknown parameters being layer thick-
ness and S-wave velocity. A genetic algorithm (Yamanaka
and Ishida, 1995) was used for optimization. Te search
area for the inversion was determined based on initial veloc-
ity models created by a simple wavelength transformation in
which wavelengths calculated from phase velocity and fre-
quency pairs are divided by three and mapped as depth.
Te theoretical phase velocity was dened as an eective
mode that was generated by calculating the weighted aver-
age of the fundamental mode and higher modes (up to the
20th mode) based on the medium response. Te inversion
was based on minimization of dierences between the ob-
served and the eective-mode phase velocities. To objectively
evaluate the capability of the 2ST-MAM, the inversion was
performed without a priori information so that the investiga-
tion can be considered a blind test, to a certain degree. In a
later section, existing borehole velocity logs and a crustal scale
3D S-wave velocity model will be shown for comparison.
Figure 7 shows an example of a comparison of observed
and theoretical dispersion curves. Yellow circles indicate the
eective mode of theoretical phase velocities as mentioned
above. Te theoretical dispersion curve (eective mode)
agrees reasonably well with the observed data.
S-wave velocity proles
At the Carson, South Gate and Manhattan Beach sites, S-
wave velocity proles were determined to a depth greater
than 2.5 km (Figure 8). At the Willowbrook site, an S-wave
velocity prole was determined only to a depth of about 1
km. Te relatively shallow penetration depth at the Willow-
brook site was because of a lack of microtremor measure-
ments with a large sensor separation.
At all four sites, there is a near-surface layer with S-wave
velocity less than 300 m/s. A shallow sti sediment layer with
S-wave velocity of more than 300 m/s was determined at a
depth ranging from 10 to 50 m, depending on the site. Tis
layer was relatively deep at the Carson and South Gate sites
compared with the other two sites.
An intermediate velocity layer with S-wave velocity great-
er than 1000 m/s was calculated for a depth range from 500
to 750 m. Above this layer, S-wave velocity at the Manhattan
Beach site in the 250750 m depth range is clearly lower than
the other three sites. Tis lower S-wave velocity at the Man-
hattan Beach site is because of relative lower phase velocity in
the frequency range of 0.41.5 Hz. Tis intermediate velocity
bedrock unit corresponds to a thick layer in which S-wave
velocity ranges from 1000 to 1500 m/s in the depth range of
7502500 m.
At the three sites (Carson, South Gate, and Manhattan
Beach) where S-wave velocity proles were estimated to a
depth of around 3000 m, the bedrock with S-wave velocity
greater than 2000 m/s was estimated to be at a depth of more
than 2500 m. Te deepest bedrock modeled at South Gate
posses an S-wave velocity lower than that at the Carson and
Manhattan Beach sites.
Figure 3. Outline of processing based on the spatial autocorrelation
using two sensors. If f(i,t) and g(i,t) are two traces of the i
th
block
obtained at two sensors (A and B) with separation x, then the fast
fourier transform (FFT) of these two functions (waveforms) for each
block can be expressed in the frequency domain as F(i,) and G(i,).
Terefore, complex coherence (COH) for i
th
block is calculated as
(1)
where CC
fg
(i,) is the crosscorrelation of two traces F(i, ) and G(i,
) and A
f
(i,) and A
g
(i, ) are the autocorrelations of F(i, ) and
G(i, ), respectively. Te spatial autocorrelation SPAC is dened as
the real part of the averaged complex coherences:
(2)
where n denotes the number of blocks. Coherence (COH) was
calculated for each block and then the real parts of all blocks were
averaged to obtain the SPAC. Ten to 100 blocks were averaged for
calculating nal SPAC. If we assume that microtremor propagates
in all directions equally, the SPAC forms a Bessel function of the rst
kind, zero order (Aki, 1957).
(3)
where, c() is phase velocity at angular frequency and J
0
is the rst
kind and zero order of the Bessel function. Te velocity that minimizes
the error in Equation 3 can be considered as the phase velocity at the
angular frequency .
1957).
1957).
Leading the way with customer-driven data, integrated software
and services for your upstream decision-making needs.
geoSCOUT | gDC | petroCUBE at www.geoLOGIC.com
Now, get geoLOGICs value-added data almost any place, any time,
any way you want it. Available through gDCweb on your tablet,
smartphone or computer.
With 30 years of data experience behind it, gDC is the source for high
quality, value-added well and land data from across Western Canada and
the Northern United States. Another plus our data is accessible through
an expanding range of industry software utilizing our own easy-to-use
gDC GIS and our geoSCOUT software.
View, search, import and export well, land and production data,
documents, logs and more from almost anywhere. For more information
visit our website at www.geoLOGIC.com
Warning: Our data has gone mobile
(You may never return to the ofce)
624 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Te S-wave velocity proles ob-
tained using 2ST-MAM were com-
pared with existing borehole velocity
logs. Te seismic velocity logs at the
Carson, South Gate, and Willow-
brook sites were acquired as part of
the ROSRINE (Resolution of Site
Response Issues from the Northridge
Earthquake) project and until re-
cently were available at http://geoinfo.
usc.edu/rosrine. Te velocity log as-
sociated with the Manhattan Beach
site was acquired by the USGS (Dan
Ponti, written communication) at a
location several kilometers north of
the 2ST-MAM array.
P- and S-wave velocity borehole
measurements were made by OYO
Corporations suspension PS logging
system. Tis system directly deter-
mines the average velocity of a 1-m
segment of the lithologic column im-
mediately surrounding the boring.
Tis measurement is the elapsed time
between the arrival of a controlled-
source wave propagating upward
through the rock/soil column and
between a pair of receivers separated
by rubber isolation tubes. Depths of
the velocity logs were 250 m at the
South Gate (Downey), Willowbrook,
and Manhattan Beach sites and 350
m at the Carson (Dolphin Park) site.
Te approximate locations of the
borehole velocity logs are close to the
measurement areas, except for the
Manhattan Beach site, which is sev-
eral kilometers from the area of mea-
surements (Figure 1).
Te S-wave proles were also
compared with a 3D seismic veloc-
ity model for Southern California:
the Southern California Earthquake
Center Community Velocity Model
(SCEC 4.0; http://www.data.scec.org/
research-tools/3d-velocity.html). Tis
velocity model was primarily based
on geologic models, empirical rela-
tionships between seismic velocity,
sediment age and depth of burial,
and existing surface and borehole
geophysical data.
Comparisons of S-wave velocity proles in the deep (left)
and shallow (right) regions from the 2ST-MAM method with
the velocity logs and the SCEC 4.0 model suggest good agree-
ment between the various data sets (Figure 9). S-wave velocities
obtained using the 2ST-MAM are reasonably matched with the
Figure 4. Example of spatial autocorrelation and phase-velocity images at the Manhattan Beach
site. Te left side of the gure shows the low-frequency component which reects S-wave velocity
structure down to several kilometers, whereas the right side shows the high-frequency component
reecting velocity structure less than about 100 m. (a) Examples of frequency-dependent coherences
comparing larger sensor spacing (left) and smaller spacing (right). (b) Typical coherences as a
function of sensor distance with theoretical Bessel functions calculated for phase velocities that
yield minimum error between the observed coherences and the theoretical Bessel function. Te
symbols indicate observed coherences and solid lines indicate the theoretical Bessel functions. (c)
Error between observed coherences and theoretical Bessel functions with magenta indicating large
error and blue indicating small error. Red dots indicate minimum-error phase velocities at each
frequency and they can be considered as the observed dispersion curves.
velocity logs when considering the resolution dierence between
surface-wave methods and borehole velocity data. At the Man-
hattan Beach site, S-wave velocities recorded on the borehole ve-
locity log are slightly faster than that of the 2ST-MAM, which is
likely associated with the physical separation between the veloc-
ity log and surface-wave measurement area.
June 2013 The Leading Edge 625
Nonreflection seismic and inversion of surface and guided waves
It is not reasonable to establish the accuracy of the 2ST-
MAM results by directly comparing them to the SCEC
model, which was constructed from interpolating sparsely
distributed geophysical and drilling information with geolog-
ic information. Here, we compare the 2ST-MAM with the
SCEC model from an approximate basin structure perspec-
tive. In general, the S-wave velocity proles developed using
2ST-MAM agree with the SCEC model.
If we compare S-wave velocity at a depth of 3000 m from
the SCEC model at the Carson and South Gate sites, the
Carson site clearly has higher velocity than the South Gate
site. Tis is consistent with the 2ST-MAM velocity models
and the fact that the South Gate site is in the deepest part of
the LA Basin. At the Manhattan Beach site, S-wave velocity
obtained using the 2ST-MAM is clearly lower than that in
the SCEC model in the depth range of 3002500 m. Te site
is along the coast and this dierence may be because of the
inaccuracy of the SCEC model. Tis example implies that the
2ST-MAM can complement large 3D velocity models such as
the SCEC model.
Conclusions
Two-station microtremor array
measurements (2ST-MAM) were
performed at four sites in the south-
western portion of the Los Angeles
Basin, California to estimate deep
S-wave velocity structure and evalu-
ate the applicability of the method
to such investigations. Resultant
S-wave velocity proles were favor-
ably compared to existing velocity
logs and a 3D seismic velocity mod-
el. Our investigation results imply
that the 2ST-MAM can accurately
detect Rayleigh-wave phase veloci-
ties down to a frequency of 0.2 Hz
and penetrate to a depth of 23 km.
Tese results have shown that using
Figure 5. Comparison of dispersion curves obtained by MASW, small
arrays, and 2ST-MAM (Carson) together with corresponding wavelength.
Figure 6. Comparison of observed dispersion curves together with
corresponding wavelength.
Figure 8. Comparison of the S-wave velocity proles in both the deep (left) and shallow (right)
regions obtained by the inversion.
Figure 7. Comparison of observed and theoretical dispersion curves.
Red solid line with white circles indicates observed dispersion curve.
Solid and broken lines indicate theoretical dispersion curves and
their relative amplitude (medium response) respectively. Yellow circles
indicate the eective mode of theoretical phase velocities.
626 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
the 2ST-MAM method is applicable
to both deep and shallow investiga-
tions.
As mentioned earlier, the 2ST-
MAM is based on the assumption
that microtremors propagate in all
directions equally. Investigation re-
sults described in this article together
with examples in other basins imply
that assumption is valid in large tec-
tonic basins along the west coast of
the United States.
Microtremor measurements using
a large number of sensors are much bet-
ter than measurements using only two
sensors. Te 2ST-MAM, however, is
much more e cient and cost-eective
compared to conventional microtrem-
or array measurements using many sen-
sors. Considering that the demand for
precise S-wave velocity models in urban
areas is increasing, the method present-
ed here can play an important role for
such investigations.
References
Aki, K., 1957, Space and time spectra of
stationary stochastic waves, with spe-
cial reference to microtremors: Bulletin
of the Earthquake Research Institute,
35, 415456.
Hatayama, K. and E. Kalkan, 2012, Long-
period (3 to 16 s) ground motions
in and around the Los Angeles Basin
during the Mw 7.2 El Mayor-Cucapah
earthquake of April 4, 2010: Presented
at 15th World Conference on Earth-
quake Engineering.
Hayashi, K. and D. Underwood, 2012a,
Estimating deep S-wave velocity struc-
ture using microtremor array mea-
surements and three-component mi-
crotremor array measurements in San
Francisco Bay Area: Proceedings of
the Symposium on the Application of
Geophysics to Engineering and Envi-
ronmental Problems.
Hayashi, K. and D. Underwood, 2012b,
Microtremor array measurements and
three-component microtremor measure-
ments in San Francisco Bay Area: Presented at the 15th World Confer-
ence on Earthquake Engineering.
Hayashi, K., R. Cakir, and T. Walsh, 2013, Using two-station mi-
crotremor array method to estimate shear-wave velocity proles in
Seattle and Olympia, Washington: Proceedings of the Symposium
on the Application of Geophysics to Engineering and Environmen-
tal Problems.
Margaryan, S., T. Yokoi, and K. Hayashi, 2009, Experiments on the
stability of the spatial autocorrelation method (SPAC) and linear
array methods and on the imaginary part of the SPAC coe cients
as an indicator of data quality: Exploration Geophysics, 40, no. 1,
Figure 9. Comparison of S-wave velocity proles in deep (left) and shallow (right) regions with the
seismic logs and the SCEC 4.0 model.
121131, http://dx.doi.org/10.1071/EG08101.
Okada, H., 2003, Te microtremor survey method: SEG, http://
dx.doi.org/10.1190/1.9781560801740.
Suzuki, H. and H. Yamanaka, 2010, Joint inversion using earthquake
ground motion records and microtremor survey data to S-wave prole
of deep sedimentary layers: Butsuri Tansa, 65, 215227 (in Japanese).
Yamanaka, H. and J. Ishida, 1995, Phase-velocity inversion using
genetic algorithms: Journal of Structural and Construction Engi-
neering, 468, 917 (in Japanese).
Corresponding author: KHayashi@geometrics.com
AREAS OF EXPERTISE
Unconventional Reservoirs
Challenging Environments
Complex Geologies
Basin Exploration
Reservoir Exploitation
>
Nonreflection seismic and inversion of surface and guided waves
628 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Love waves from local trafc noise interferometry
S
urface-wave interferometry on local scale usually aims
at recovering Rayleigh waves. Tis is because of the
predominant use of vertical component geophones in
exploration seismology and the fact that Rayleigh waves
occur for any given subsurface structure. On the other hand,
Love waves are present only in layered media and require
horizontal component geophones for their observation. As
they depend on shear-wave velocity structure and density
only, the analysis of Love waves provides a potentially
powerful supplement to Rayleigh wave inversion. Perhaps
surprisingly, recent studies show that
low-frequency Love waves (0.050.1
Hz) excited by the interaction of
ocean waves with the ocean oor
(the Earths microseism) can be
recovered by interferometry, and
that their S/N is high compared to
Rayleigh waves (Lin et al., 2008).
On a regional scale, Jay et al. (2012)
analyzed the ambient noise eld in
a volcanic region and found that
Love waves with frequencies of about
0.3 Hz are observed more clearly
than corresponding Rayleigh waves.
In this article, we show that Love
waves in the frequency band of 1.5
to 5 Hz can be obtained from local
noise interferometry, and that they
are of comparable S/N as Rayleigh
waves. Tus they may also be used to
constrain the near-surface structure.
Love waves in a nutshell
Love waves are horizontally polar-
ized because they result from interac-
tion of shear (SH) waves. As opposed
to Rayleigh waves, Love waves exist
in layered media only. For the one-
layer case, the Love wave represents
the superposition of multiply, criti-
cally reected downgoing SH waves
from the bottom of the layer (e.g.,
Stein and Wysession, 2003). Te lay-
er of a thickness H is then considered
as a wave guide and the Love-wave
velocity c
L
is inbetween the shear-
wave velocities of the layer and the
half-space (Figure 1). In contrast,
Rayleigh-wave velocities are always
less than the layers shear-wave ve-
locity. Te dispersion relation shows
that Love-wave velocities at low fre-
quencies tend toward the half-space
M. BEHM and R. SNIEDER, Colorado School of Mines
velocity
2
, while observations at high frequencies give the
layer velocity
1
. Equation 1 relates the Love-wave velocity
c
L
to its frequency f, layer thickness H, layer and half-space
shear-wave velocities
1
,
2
, and densities
1
,
2
:
(1)
Figure 1. Love-wave phase velocity as a function of frequency and layer thickness H for a layer-
over-half-space model (dispersion relation). and refer to shear-wave velocity and density,
and i
c
is the critical angle. Te shown raypath is not the actual raypath of the Love wave, but
schematically describes critically reected SH waves. Te dashed gray line is the wavefront of the
downgoing SH waves which interfere to constitute the Love wave at the surface point P.
Figure 2. Te deployment of the La Barge Passive Seismic Experiment in southwestern Wyoming.
White dots indicate locations of three-component instruments. Te black line represents the state
road contributing dominantly to the ambient noise. Te Hogsback thrust is the main structural
feature and separates carbonate outcrops in the west from siltstones and sandstones in the east.
June 2013 The Leading Edge 629
Nonreflection seismic and inversion of surface and guided waves
densities, also the layer thickness. Eslick et al. (2008)
examine the constraints on the subsurface settings and re-
cording parameters for successful retrieval of Love-wave dis-
persion. In case of several layers, the shear-wave velocities of
the layers can be determined analogously to Rayleigh waves
(Xia et al., 2009). From a practical point of view, it is inter-
esting to note that even for the one-layer case the dispersion
relation is highly nonlinear and a so-
lution for Love-wave velocity must
be determined numerically. Solu-
tions for multilayer-models and later-
ally varying layer thicknesses require
more computational eort (e.g., Ben-
Hador and Buchen, 1999). On the
other hand, the nonlinearity of the
dispersion relation can be employed
to impose constraints on shear waves,
densities, and layer thickness, pro-
vided the Love-wave velocity can be
reliably observed over an appreciable
frequency range.
Data and interferometric processing
Te La Barge Seismic Experiment
is a industry-academia cooperation
aiming at evaluating the feasibility of
passive seismology for local subsur-
face characterization (Saltzer et al.,
2011). From November 2008 to June
2009, 55 3C broadband stations were
deployed at a spacing of 250 m in an
active hydrocarbon production site
in southwestern Wyoming (Figure
2). Te continuous recordings and
the small aperture of the array make
the data set well suited for local inter-
ferometry analyses. Previous investi-
gations (Behm et al., accepted) show
that both Rayleigh- and Love-wave
velocity information are obtained
from tra c noise originating from a
state road. We rst summarize their
approach and their most important
ndings, and then discuss the Love
waves in more detail.
Each of the stations is turned into
a virtual source by correlating its am-
bient noise recording with the am-
bient noise recording of every other
station. It turns out that ve days of
continuous noise data are su cient
to recover surface waves travelling be-
tween the stations up to distances of 5
km. Analysis further shows that traf-
c activity from the Wyoming state
road 235 in the eastern part of the
deployment provides the main source
Equation 1 illustrates that, opposed to Rayleigh waves, Love-
wave velocities do not depend on compressional-wave veloci-
ties. Te use of Love waves thus reduces the ambiguity inher-
ent in inversion for shear-wave structure. If the assumption
of the one-layer case is well justied, the dispersion relation
potentially enables us to estimate the shear-wave velocities of
both the layer and the basement, and by further assuming
Figure 3. Interferograms for virtual source L17 (red star) and stations L42 to L55 (red dots).
Te lack of causal energy results from the receiver stations being closer to the noise source (WY
state road 235) than the virtual source. Dashed red lines indicate linear moveouts for velocities of
1500, 2000, and 2500 m/s. Note the higher apparent phase velocity at the transverse component
(T) compared to vertical (Z) and radial (R) components. Te black rectangle depicts the waveform
from which the dispersion curve for raypath A (Figure 6) is calculated.
Figure 4. Common-oset stacks of Hilbert-transformed interferograms from 530 virtual-source
receiver combinations in the central part of the investigated area. Note the overall similarity of the
vertical (Z) and radial (R) components, and the higher apparent group velocity of the transverse
(T) component data.
630 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
of coherent noise. Te location of the road with the respect
to the deployment ensures a large number of stationary phase
points such that most interferograms feature clear surface-
wave arrivals with high S/N. Rayleigh waves are polarized in
the plane defned by the vertical and the propagation direc-
tion, and Love waves are polarized in the horizontal plane
and perpendicular to the propagation direction. By rotating
the horizontal components into the radial (virtual source
Figure 5. Near-surface phase-velocity maps and their ratio as obtained from traveltime
tomography of vertical and transverse component interferograms (from Behm et al. accepted,
slightly modifed). Te arrows (A, B) denote the virtual source-receiver pairs for which dispersion
curves are shown in Figure 6.
receiver azimuth) and transverse (90 clockwise to the azi-
muth) directions, it is possible to separate Love and Rayleigh
waves. Te slower Rayleigh wave is present on the vertical
and radial components, and the faster Love wave appears on
the transverse component (Figure 3). Although of overall
high S/N, the interferograms are characterized by a limited
bandwidth peaking at 2.5 Hz, and subsequently a sometimes
ringy wavelet. Te calculation of the envelope (modulus of
the Hilbert-transformed interfero-
gram) compresses oscillating wavelets
and improves the delectability of the
onset of the waves. It is important to
note that envelope interferograms no
longer represent phase velocities, but
group velocities instead. All envelope
interferograms from the central part
of the investigated area are further
stacked in ofset bins (Figure 4). Al-
though lateral velocity variations may
degrade the stacks, these results also
support the existence of both Ray-
leigh and Love waves. Vertical and
radial component data appear simi-
lar with respect to apparent velocity
and maximum ofsets (25003500
m), while transverse component data
feature higher apparent velocity and
also more consistent arrivals at large
ofsets (35005000 m).
Dispersion of surface waves en-
ables us to invert for shear-wave veloc-
ity structure. In exploration seismol-
ogy and near-surface investigations,
the multichannel analysis of surface
waves (MASW) is commonly applied
to obtain Rayleigh-wave phase-veloc-
ity dispersion (Park et al., 1999). Te
relatively low central frequency of the
obtained trafc noise interferograms
in conjunction with the station spac-
ing hampers the observation of phase
velocity dispersion, while the large
number of clear surface-wave arrivals
allows inverting picked traveltimes for
laterally varying group and phase ve-
locities. Vertical component traveltimes
provide Rayleigh-wave velocities, and
transverse component traveltimes are
inverted for Love-wave velocities (Fig-
ure 5). With respect to the wavelength,
those velocities represent average sur-
face-wave velocities from the upper
100300 m. Te results correlate well
with the surface geology as the carbon-
ates west of the Hogsback thrust are
represented by relatively high velocities.
Lateral resolution of the velocity maps
Figure 6. Group velocity dispersion curves obtained from frequency-time analysis for two virtual
sourcereceiver pairs (A: L46 > L17; B: L55 > L31). Te white curve depicts the maximum
amplitude at each frequency. Note the overall similarity of the vertical (Z) and radial (R)
components, and the higher velocity and diferent appearance of the transverse (T) component. Te
gray bar shows the average phase velocity (Figure 5) along the raypath.
June 2013 The Leading Edge 631
Nonreflection seismic and inversion of surface and guided waves
depends on ray coverage and picking accuracy, and we estimate
it to range between 5001000 m.
Group velocity dispersion of Rayleigh and Love waves
In contrast to the MASW method, the relatively sparse dis-
tribution of broadband instruments on continental scale
led to methods to derive group velocity dispersion between
station pairs based on the frequency-time analysis (FTAN;
Dziewonski et al., 1969; Levshin et al., 1989). With this ap-
proach, the surface wave travelling between two stations is
reconstructed by interferometry. Te interferogram is ltered
in dierent frequency bands, where each band is dened by
a Gaussian function of a central frequency and given half-
width. After ltering, the envelope of the trace is calculated.
By knowing the oset between the two stations, the time
axis of the trace is converted to velocity and the maximum
of the envelope is picked for each central frequency. Te ob-
tained group velocity dispersion curve can then be inverted
for a shear-wave velocity-depth function representing the
region between the stations. If station coverage is dense, a
tomographic approach for a 3D shear-wave velocity model
is also feasible. Tis method has been applied successfully
to ambient noise from globally distributed earthquakes to
delineate crustal and mantle structures (e.g., see the overview
given by Bensen et al., 2007). As with MASW, the inversion
for shear-wave velocities assumes a layered 1D model. Te
evident lateral variation in the investigated area limits the
general applicability of the FTAN algorithm, but nonethe-
less we are able to calculate group velocity dispersion curves
for selected receiver pairs (Figure 6). To minimize the con-
tribution of spurious energy, we mute the interferograms for
apparent velocities larger than 4000 m/s and less than 1000
m/s prior to the dispersion analysis. Te raypath A com-
prises the stations L46 (virtual source) and L17 (receiver) in
the high-velocity carbonates. As virtual source and receiver
are interchangeable, the actual surface wave used for the cal-
culation of the dispersion curve is seen in the acausal part of
Figure 3. In contrast, the raypath B connects the stations
L55 (virtual source) and L31 (receiver) in the low-velocity
eastern part.
Vertical and radial component dispersion curves appear
similar with a gentle tendency of lower velocities toward
higher frequencies, while the dispersion of the transverse
component with its steep slope toward the low-frequency end
resembles typical Love-wave dispersion characteristics (com-
pare with Figure 1).
Group velocities U
G
and phase velocities U
P
are related by
. (2)
For a given frequency f and realistic velocities, group ve-
locities are less than phase velocities if the phase velocities
632 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
decrease with frequency (dU
p
/df < 0) and vice versa. Te mag-
nitude of the diference between group and phase velocities
is inversely proportional to the phase velocity and to the rate
of change of phase velocity with frequency. Te observations
for ray path A qualitatively agree with this general relation
of phase and group velocities in that sense that group veloci-
ties are less than phase velocities. In case of raypath B, the
group velocities surprisingly appear higher than the phase
velocities. However, it is noted that raypath B crosses the
region with the strongest lateral variation, representing the
westward dipping low-angle Hogsback thrust where a gradu-
ally thickening sheet of high-velocity carbonates overthrusts
low-velocity sediments. Tis defnitely represents a challenge
to the simplifcations inherent to both traveltime inversion
and dispersion interpretation, and also illustrates limits to
surface-wave inversion.
Outlook
Our study shows that locally excited Love waves in a typical
exploration environment and in the frequency range of 1.55
Hz are of comparable, if not higher S/N than Rayleigh waves.
Tis frequency range does not necessitate costly broadband
stations as in our test study, but can be well targeted by low-
frequency geophones more commonly used in exploration
seismology. With more and more industrial applications rely-
ing on three-component instruments (e.g., seismic monitor-
ing, shear-wave retrieval), the recording and potential use of
Love waves for subsurface characterization becomes feasible.
In particular passive seismic deployments are well suited, as
surface waves can be efciently recovered from interferom-
etry applied to local ambient noise. Love waves are enticing
because, compared to Rayleigh waves, they do not depend on
P-wave velocity and thus reduce the ambiguity in extracting
shear-wave velocity structure. Near-surface shear-wave veloc-
ity inversion zones might by quickly mapped by the absence
of Love waves. Te complementary information of Rayleigh
and Love waves provides improved assessment of seismic ve-
locities and densities. Combined dispersion measurements of
Love and Rayleigh might also be used to constrain lateral
variations in Earth structure (Levshin and Ratnikova, 1984)
and seismic anisotropy (Montagner and Nafaf, 1986). In
case when the near surface is sufciently described by a one-
layer model, the distinct shape of the Love-wave dispersion
curve could facilitate to estimate layer and half-space shear-
wave velocities and densities simultaneously.
References
Ben-Hador, R. and P. Buchen, 1999, Love and Rayleigh waves in
non-uniform media: Geophysical Journal International, 137, no.
2, 521534, http://dx.doi.org/10.1046/j.1365-246X.1999.00790.x.
Bensen, G. D., M. H. Ritzwoller, M. P. Barmin, A. L. Levshin, F. Lin,
M. P. Moschetti, N. M. Shapiro, and Y. Yang, 2007, Processing
seismic ambient noise data to obtain reliable broad-band surface
wave dispersion measurements: Geophysical Journal Interna-
tional, 169, no. 3, 12391260, http://dx.doi.org/10.1111/j.1365-
246X.2007.03374.x.
Behm, M., R. Snieder, and G. M. Leahy, Retrieval of local surface
wave velocities from trafc noisean example from the LaBarge
basin (Wyoming): accepted for publication in Geophysical Pros-
pecting.
Dziewonski A. M., S. Bloch, and M. Landisman, 1969, A technique
for the analysis of transient seismic signals: Bulletin of the Seismo-
logical Society of America, 59, 427444.
Eslick, R., G. Tsofias, and D. Steeples, 2008, Field investigation of
Love waves in near-surface seismology: Geophysics, 73, no. 3, G1
G6, http://dx.doi.org/10.1190/1.2901215.
Jay, J. A., M. E. Pritchard, M. E. West, D. Christensen, M. Haney, E.
Minaya, M. Sunagua, S. R. McNutt, and M. Zabala, 2012, Shal-
low seismicity, triggered seismicity, and ambient noise tomography
at the long-dormant Uturuncu volcano, Bolivia: Bulletin of Vol-
canology, 74, no. 4, 817837, http://dx.doi.org/10.1007/s00445-
011-0568-7.
Levshin, A. L., and L. I. Ratnikova, 1984, Apparent anisotropy in in-
homogeneous media: Geophysical Journal of the Royal Astronom-
ical Society, 76, no. 1, 6569, http://dx.doi.org/10.1111/j.1365-
246X.1984.tb05022.x.
Levshin, A. L., T. B. Yanovskaya, A. V. Lander, B. G. Bukchin, M. P.
Barmin, L. I. Ratnikova, and E. N. Its, 1989, Seismic surface waves
in a laterally inhomogeneous Earth: Norwell.
Lin, F., M. P. Moschetti, and M. H. Ritzwoller, 2008, Surface wave
tomography of the western United States from ambient seismic
noise: Rayleigh and Love wave phase velocity maps: Geophysical
Journal International, 169, 12391260.
Montagner, J. P. and H. C. Nataf, 1986, On the inversion of the
azimuthal anisotropy of surface waves: Journal of Geophysi-
cal Research, 91, B1, 511520, http://dx.doi.org/10.1029/
JB091iB01p00511.
Park, C. B., R. D. Miller, and J. Xia, 1999, Multichannel analysis
of surface waves: Geophysics, 64, no. 3, 800808, http://dx.doi.
org/10.1190/1.1444590.
Saltzer, R., G. M. Leahy, J. Schmedes, J. Roth, and E. Rumpfhuber,
2011, EarthquakesA naturally occurring source of low frequen-
cy data: 81st Annual International Meeting, SEG, Expanded Ab-
stracts, 36893693, http://dx.doi.org/10.1190/1.3627967.
Stein S. and M. Wyssesion, 2003: An introduction to seismology,
earthquakes, and earth structure: Blackwell publishing.
Xia, J., R. Cakir, R. D. Miller, C. Zeng, and Y. Luo, 2009, Estima-
tion of near-surface shear-wave velocity by inversion of Love waves:
79th Annual International Meeting, SEG, Expanded Abstracts,
13901394, http://dx.doi.org/10.1190/1.3255109.
Acknowledgments: Tis work was funded by ExxonMobil. We
thank Matt Haney for reviewing the manuscript. IRIS DMC was
used to access the waveform data.
Corresponding author: mbehm@mines.edu
Nonreflection seismic and inversion of surface and guided waves
634 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Surface-wave observations after integrating active and passive
source data
E
mpirical Greens function (EGF) retrieval and turning
ambient noise into useful signal by crosscorrelation or
seismic interferometry (Curtis et al., 2006) has been a popular
topic in recent years in the seismological community. Many
have discussed how the interstation distance or its equivalent
afects the accuracy of the Greens function that can be
retrieved by crosscorrelation of long-range noise between two
stations (e.g., Snieder, 2004; Bensen et al., 2007; Halliday
and Curtis, 2008; Tsai, 2009; Kimman and Trampert, 2010).
It is generally accepted that the necessary or optimum in-
terstation distance strongly depends on source distribution,
length of records (and, hence, is naturally related to source
distribution), and the duration of the Greens function to be
retrieved. Noise generated by a surface source can be efciently
used to reconstruct the Greens function of surface waves if we
focus on the accuracy of the phase, and not be too concerned
with the amplitude accuracy of the retrieved Greens function
(Halliday and Curtis, 2008; Kimman and Trampert, 2010).
Free-space theory for surface source distribution
Noise sources that contribute to the retrieved Greens func-
tion resulting from the crosscorrelation of two stations, A
and B, in the xoy plane that are extracted from an acoustic
free space with a velocity c, must distribute on a hyperbola
(Figure 1) (Roux et al., 2005).
Te acute angle formed by the asymptote with the x-
axis in the frst quadrant is
(1)
It is obvious that d D if 0 . Te angle formed
by connecting any point S on the hyperbola to the origin will
be less than .
Constructive interference will occur between records of
receivers A and B when
, n N
as defned by the frst Fresnel zone, where denotes wave-
length of a plane wave with angle frequency for a random
source S. On the other hand, from Equation 1, the following
inequality always holds in the frst quadrant,
(2)
Yixian xu, Baolong Zhang, Yinhe luo, and Jianghai xia, China University of Geosciences
Figure 1. Hyperbola defned by Equation 1 with foci A and B. Te
acute angle formed by the asymptote of a hyperbola with the x-axis
in the frst quadrant.
Figure 2. Coverage angle versus minimum interstation distance.
Mask (75.6, 1) means the interstation distance is at least one
wavelength when the noise coverage angle approaches 75.6.
Equation 2 states that the minimum interstation distance D
is determined by the coverage angle (and, hence, its upper
limit ) and the wavelength . Te smaller the angle , the
smaller the interstation distance D could be (Figure 2). Based
on a rigorous analysis for realistic medium by Tsai (2009), the
far-feld plane wave or frst term approximation (Equation 12 in
that paper) of the measured delay time between two stations
tan cos
cos cos cos
cos
June 2013 The Leading Edge 635
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
will be related to coverage angle and interstation distance as
. (3)
If the delay-time between two stations is measurable
only if is greater than a quarter of the period T, (i.e., D cos
/ c T / 4), Equation 2 has value.
As suggested by Bensen et al. (2007), the empirical
minimum interstation distance must be larger than three
Figure 3. Dispersion images derived from the same shot gather:
(a) spread length of 24 m ( one wavelength at 10 Hz); (b) spread
length of 12 m ( half wavelength at 10 Hz); (c) spread length of 6
m ( one-fourth wavelength at 10 Hz). Te model is a top layer with
thickness 10 m underlying a half-space. Te S-wave velocities of the
top layer and the half-space are 200 m/s and 800 m/s, respectively. Te
wavelet peak frequency is 10 Hz. Te dispersion image is calculated
from a shot gather with 0.5-m receiver interval by the phase-shift
method (Park et al., 1999) and the energy is normalized for every
frequency. Te dot line is the analytical result calculated by Schwab
and Knopofs scheme (Schwab and Knopof, 1972).
times the wavelength. Tis is related to the coverage angle of
random noise and its general distribution everywhere in real
world (Figure 1 and Figure 2).
Extension to multichannel analysis of surface waves
(MASW)
An interesting case results when all the sources are located
at x axis (i.e., D = d) and = = 0. Tis source orientation
is applicable to the multichannel analysis of surface waves
(MASW) method when using an active source (Xia et al.,
1999; Park et al., 1999; Xu et al., 2006). Te minimum inter-
station distance for an MASW spread will be roughly equiv-
alent to / 4 (Equation 2). To demonstrate this relationship,
a minimum spread necessary for extracting 10-Hz dispersive
energy with 200 m/s phase velocity is 5 m. Please note that
this result is for an inline, long-ofset, random source that is
independent of the source energy, attenuation, and degree
of heterogeneity of the media that the surface wave passes
through.
Te resolution of the MASW dispersion image depends
on the geophone spread not the geophone interval (Forbriger
2003; Xia et al., 2006), and in the frequency-velocity do-
main, resolution increases with increasing geophone spread
length (Figure 3). In spite of a drop in resolution in frequency
versus phase velocity image as the spread length decreases,
phase velocities can be accurately estimated at frequencies
lower than the source peak frequency as long as the spread
length is greater than one fourth wavelength. Tis is analo-
gous to improvements in surface-wave EGF extracted from
crosscorrelation of noise wavefelds (Tsai, 2009) (Figure 6)
obtained by increasing interstation distance. It is noted again
Figure 4. Shot gather (band-fltered within 130 Hz) generated
seismic interferometry for a virtual source at the seventh trace.
cos
636 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
dispersive energy image is actually sensitive to the spatially
averaged properties of subsurface. Tis confrms the previ-
ous statement that spatial resolution can be improved by us-
ing narrow, denser arrays. Using a rough estimate of a 5-Hz
Rayleigh-wave phase velocity of 1600 m/s at the base rock of
the tunnel produces a wavelength is 320 m. It is clear that we
can also get a good estimate of the dispersion-curve image
even when the spread length approaches half a wavelength of
the passive seismic energy.
Seventeen shot gathers were constructed from the entire
data set. We used the total dispersion curve generated for each
of the 17 shot gathers when inverting for each 1D S-wave
velocity profle used to generate the 2D velocity cross section
(Figure 6b). Two low-velocity layers (LVL) were revealed.
Te top LVL is continuously distributed from the midpoint
of receivers 15 and 16, 60 m deep to the right up to about
Figure 5. Dispersion-energy images calculated by the phase-shift
method from traces (a) 140 traces, (b) 123, and (c) 1840.
that the spread length is consistent with interstation distance
as defned in this case.
Increasing the spread length or interstation distance will
decrease the spatial resolution of the MASW or EGF output.
From the previous analysis, the upper limit of the horizontal
spatial resolution (determined to be half the interstation dis-
tance for surfaces waves) was one-eighth wavelength for the
EGF extraction. Tis is true when all possible random noise
sources are distributed inline with at least two stations (a con-
straint also true for MASW). For MASW, incrementally mov-
ing the efective receiver spread less than a spread length after
every source recording (analogous to the multifold technique
used in refection seismic) allows spatial sampling points on
sub-spread-length intervals. Tis process efectively improves
spatial resolution by inverting dispersion curves derived from
the intraspread records. Some good examples from the real
world can be found in Miller et al. (1999).
An example of seismic interferometry
From the passive seismic perspective, we might expect denser
one- or two-dimensional seismic arrays comparatively im-
prove spatial resolution by inverting dispersion curves from
the extracted EGF for a profle or an area. Tis scheme should
beneft from the partial duplication of information content
within the subsurface. It is possible to recover the same in-
formation content by inverting spatially denser dispersion
curves. We hence report an example of ambient noise-based
seismic interferometry.
We deployed a dense array consisting of 47 Texans 125A
seismic recorders with 4.5-Hz vertical receivers across the base
of a tunnel structure near Badong, a town on the southern
side of the Yangtze River in the frst section of Tree Gorges.
Te tunnel was excavated through the core of Huangtuling
landslide by China University of Geosciences to monitor
any movement along the slide plane (Figure 6a). Te array
was 368 m in length with 8-m receiver intervals. Te record
length was consistently 26 hours. We used the standard seis-
mic interferometry approach (e.g., Curtis et al., 2006) to ob-
tain the shot gather generated by the virtual source.
We know from independent measurements that the S-
wave velocity of the rock exposed at the base of the tunnel is
between 1000 and 2000 m/s. Te dominant Rayleigh waves
evident in the shot gather have a frequency range from 3 to
8 Hz (Figure 4). Using the tau-p transform (McMechan and
Yedlin, 1981), we generated an image of the dispersive en-
ergy. For the shot gather depicted in Figure 4, the energy im-
ages are calculated for traces 1-40 (Figure 5a), 123 (Figure
5b), and 1840 (Figure 5c), corresponding spread lengths of
312, 176, and 176 m, respectively.
Consistent with previous observations, the resolution
of the energy images decreases with decreasing the spread
lengths. Energy images displayed in Figure 5b and Figure
5c are obviously diferent because the number of traces (ex-
cept traces 1823 traces) used to generate each is diferent.
However, it is interesting to note that the energy image in
Figure 5a (40 traces) can be constructed by simply averaging
Figure 5b and Figure 5c. Tis result demonstrates that the
June 2013 The Leading Edge 637
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
20 m and potentially linked to the slide plane exposed at the
tunnel base of receiver position 25. Te second LVL can be
found at the middle of the cross section approaching 100 m
deep to the right endpoint up to 50 m deep. Tese LVLs of
the S-wave are interpreted as the potential slide planes of the
Huangtuling landslide.
References
Bensen, G., M. Ritzwoller, M. Barmin, A. Levshin, F. Lin, M. Mos-
chetti, N. Shapiro, and Y. Yang, 2007, Processing seismic am-
bient noise data to obtain reliable broad-band surface wave
dispersion measurements: Geophysical Journal International,
169, no. 3, 12391260, http://dx.doi.org/10.1111/j.1365-
246X.2007.03374.x.
Curtis, A., P. Gerstoft, H. Sato, R. Snieder, and K. Wape-
naar, 2006, Seismic interferometry-turn noise into signal:
Te Leading Edge, 25, no. 9, 10821092, http://dx.doi.
org/10.1190/1.2349814.
Forbriger, T., 2003, Inversion of shallow-seismic wavefelds: I. Wave-
feld transformation: Geophysical Journal International, 153,
720734.
Halliday, D. and A. Curtis, 2008, Seismic interferometry, surface
waves, and source distribution: Geophysical Journal Interna-
tional, 175, no. 3, 10671087, http://dx.doi.org/10.1111/j.1365-
246X.2008.03918.x.
Kimman, W. P. and J. Trampert, 2010, Approximations in seismic in-
terferometry and their efects on surface waves: Geophysical Jour-
nal International, 182, 461476.
McMechan, G. A. and M. J. Yedlin, 1981, Analysis of dispersive waves
by wave feld transformation: Geophysics, 46, 869874, http://
dx.doi.org/10.1190/1.1441225.
Miller, R. D., J. Xia, C. B. Park, and J. Ivanov, 1999, Multichannel
analysis of surface waves to map bedrock: Te Leading Edge, 18,
no. 12, 13921396, http://dx.doi.org/10.1190/1.1438226.
Park, C. B., R. D. Miller, and J. Xia, 1999, Multichannel analysis
of surface waves: Geophysics, 64, no. 3, 800808, http://dx.doi.
org/10.1190/1.1444590.
Roux, P., K. G. Sabra, W. A. Kuperman, and A. Roux, 2005, Ambi-
ent noise cross correlation in free space: theoretical approach: Te
Journal of the Acoustical Society of America, 117, no. 1, 7984,
http://dx.doi.org/10.1121/1.1830673.
Schwab, F. A. and L. Knopof, 1972, Fast surface wave and free mode
computations, in B. A. Bolt ed., Methods in Computational Phys-
ics, Academic Press, 87180.
Snieder, R., 2004, Extracting the Greens function from the cor-
relation of coda waves: a derivation based on stationary phase:
Physical review. E, Statistical, nonlinear, and soft matter
physics, 69, no. 4, 046610, http://dx.doi.org/10.1103/Phys-
RevE.69.046610.
Tsai, V., 2009, On establishing the accuracy of noise tomography trav-
el-time measurements in a realistic medium: Geophysical Journal
International, 178, no. 3, 15551564, http://dx.doi.org/10.1111/
j.1365-246X.2009.04239.x.
Figure 6. Te tunnel location marked on the Google image (a), where
the points mark receivers and the shaded section corresponds to the
cross section and the inverted 1D S-wave velocity cross section along
the tunnel base of Huangtuling landslide (b). Te origin of the cross
section locates at the midpoint of the receivers 15 and 16.
Acknowledgments: Y. Xu is supported by the National Natural Sci-
ence Foundation of China (NSFC) under grant no. 40974079. Y.
Luo is supported by the Special Fund for Basic Scientifc Research
of Central Colleges, China University of Geosciences (Wuhan)
(#CUGL100402). And J. Xia is partly supported by the National
Natural Science Foundation of China (NSFC), under grant no.
41274142.
Corresponding author: xyxian@cug.edu.cn
Nonreflection seismic and inversion of surface and guided waves
638 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Surface- and guided-wave inversion for near-surface modeling
in land and shallow marine seismic data
I
n several domains of applied geophysics, surface, and
guided waves are considered as a source of information
for characterizing the near surface, which in a marine
environment includes the seabed. By contrast, in exploration
seismic surveys, these waves have traditionally been regarded
as coherent noise that should be fltered out as soon as
possible. Te authors consider that surface and guided
waves are not noise but a signal that can be lifted from the
seismic record and exploited for a variety of well-established
geophysical solutions. Surface and guided waves constitute a
large part of the recorded energy and with proper acquisition,
analysis, and inversion they can be used to characterize
the near surface with surprisingly high resolution. In this
role, they can provide valuable information for tasks such
as perturbation correctionadjustment for near-surface
traveltime distortions. Tey can also be used for velocity and
geological modeling. In this article, the authors discuss a
workfow for the analysis and joint inversion of surface and
guided waves in both land and ofshore seismic data.
Introduction
Tere is no satisfactory defnition of
near-surface in the context of ex-
ploration seismic. It is often consid-
ered as the shallow part of the sub-
surface whose properties, although
not directly of interest, can distort
or otherwise degrade the observed
response of deeper targets. It is fre-
quently the part of the subsurface in
which seismic coherent noise propa-
gates. In this view, the near surface
can be generally described as a lay-
ered waveguide in which the upper
boundary is the free surface and the
lower boundary is the bottom of the
weathering layer. A large part of the
wavefeld recorded in surface seismic
consists of energy trapped in this
waveguide, which manifests itself in
the form of surface and guided waves.
Surface and guided waves con-
sist of several modes of Rayleigh
waves (Scholte waves in shallow
water environments), Lamb waves
(when strong velocity inversions are
present), Love waves (on horizontal
components when properly excited),
Stoneley waves (that typically propa-
gate along a solid-fuid interface, and,
more rarely, a solid-solid interface),
Daniele Boiero, eDwarD wiarDa, and Peter Vermeer, WesternGeco
and guided P- and S-waves. In many cases, some of these
modes may be present simultaneously and are superimposed
on each other.
Analysis of surface waves is widely adopted for build-
ing near-surface S-wave velocity models (Socco et al., 2010)
and the method is under continuous and rapid evolution for
exploration seismology applications (Strobbia et al., 2011);
however, the use of guided waves is not yet considered a com-
mon practice.
Although, from a theoretical point of view, guided waves
is a general term that includes surface waves, the terminology
adopted in the seismic exploration community refers to guid-
ed waves as those events generated by multiple refections in
the near surface that propagate horizontally from the source
and are recognized usually by a characteristic high-amplitude
interference pattern that is often called shingling.
In exploration seismology, guided P- and S-waves are of-
ten observed in surface seismic data (Muyzert, 2007). Figure
1a shows a characteristic land shot record. Te wavefeld is
Figure 1. (a) Example land shot gather with Rayleigh waves (A) and guided P- and S-waves (B).
(b) Example shallow water OBC receiver gather with Scholte waves (C) and guided P-waves (D).
(c) Example shallow water towed-streamer shot gather with Scholte waves (E) and guided P-waves
(F). (d) f-k transform along the same line as (a). Event 1 is a Rayleigh-wave mode. Event 2 is a
guided-wave mode. (e) f-k transform along the same line as (b). Events in the inset are Scholte-
wave modes. Other events are guided wave modes. (f ) f-k transform along the same line as (c).
Events in the inset are Scholte-wave mode. Other events are guided-wave modes.
June 2013 The Leading Edge 639
Nonreflection seismic and inversion of surface and guided waves
near-surface model can be used for short- to long-wavelength
static and perturbation corrections, for velocity model build-
ing, and for geotechnical applications.
Surface-wave analysis
Te common physical principle of dierent surface-wave
characterization methods is related to the fact that their
propagation depends on their wavelengths, which in turn
is responsible for the geometric dispersiondierent fre-
quencies have dierent phase velocities. Te dispersion is
strictly related to the site properties and can be inverted to
a near-surface velocity model (Socco et al., 2010; Strobbia
et al., 2011). For this reason, we analyze guided waves by
looking at their main property: geometric dispersion. Sev-
eral approaches have been developed for the analysis of sur-
face waves in dierent elds of application depending on the
available data, ranging from sparse 3D networks for earth-
quake seismology to small-scale linear arrays of geophones
in geotechnical characterization (Socco et al., 2010). For the
applications of interest in the reection seismic industry, a
robust and exible implementation of surface-wave analysis
techniques is needed (Strobbia et al., 2011).
Te objective of the analysis is the extraction of the local
wavenumber as a function of frequency, k(f ), for the dier-
ent modes. It can be observed that in a laterally gently vary-
ing medium the gradient of the modal phase is essentially a
surface-consistent parameter. If the waveform is obviously af-
fected by the full propagation path, the kinematic properties
of the surface wave, when excluding the near eld, can be ex-
pressed in terms of local properties (Vignoli et al., 2011). At
this stage, each location is considered one-dimensional and
the local phase velocity can be inverted to obtain the vertical
distribution of the near-surface velocities.
Estimation of phase velocities is done here following the
approach proposed by Strobbia et al. (2011), which is based
on the use of high-resolution, unevenly spaced f-k transforms
(Figures 1d, e, and f ) to estimate the local properties of sur-
face and guided waves within a patch of receivers. Te analysis
complex because of the interference
of reected and refracted multiples
of P- and S-waves and of converted
waves. Guided P-waves are disper-
sive events that appear with relatively
high phase velocities, which can ap-
proach the moveout of reections at
large osets. Guided S-waves have
slower phase velocities than guided
P-waves and overlap the ground-roll
cone; they are usually called higher-
order Rayleigh modes on vertical
component data. From a theoreti-
cal point of view, Roth et al. (1998)
showed that the Rayleigh waves are
the superposition of normal modes
and that shingled guided P-waves are
the superposition of leaking modes
(Haddon, 1984).
A shallow marine environment supports guided P-waves
in the water layer. Tey usually display a number of char-
acteristic features: their dispersion patterns have a resonant
frequency-tuned appearance; they have relatively high cut-o
frequencies; and their phase velocities exceed the velocity of
the water (Shtivelman, 2004). Figures 1b and 1c show ex-
amples of an ocean-bottom cable (OBC) receiver gather and
a towed-streamer shot gather that display two distinct groups
of waves: the rst consists of low-frequency, low-velocity nor-
mal modes (Scholte waves), and the second consists of leak-
ing modes (guided P-waves) that have higher velocities and
frequencies. When the subwater layers are composed of rela-
tively soft saturated rocks with high Poissons ratio, the leak-
ing modes can be approximated by guided acoustic waves.
Guided waves, jointly with the other types of surface
waves mentioned above, can be used to obtain a near-surface
velocity model for P- and S-waves as their propagation prop-
erties depend directly upon the elastic properties of the near-
surface (Roth and Holliger, 1999; Ritzwoller and Levshin,
2002; Klein et al., 2005; Muyzert 2007). In particular, guid-
ed P-waves can provide constraints on the P-velocities where-
as guided S-waves appear as higher surface-wave modes and
are used for the estimation of S-velocities (Shtivelman, 2004;
Boiero et al., 2009).
Dierent dispersive modes can be analyzed together to
build a reliable near-surface velocity model according to the
following workow:
1) Obtain a high-resolution spatial distribution of the modes
properties, in particular the velocity
2) Invert the modes properties to a near-surface model
A surface-wave analysis and inversion method is proposed.
It exploits the dierent modes, including guided waves. Tis
changes the perspective of coherent noise in the context of
reection seismic; whereby, surface-wave analysis becomes
part of the data processing workow and surface and guided
waves are considered as signal and not noise. Te obtained
Figure 2. (a) Rayleigh-wave phase-velocity distribution along the line (event 1 in Figure 1d). (b)
Guided-wave phase-velocity distribution along the line (event 2 in Figure 1d). Te black contour
line indicates the boundary between guided P-waves (below) and guided S-waves (above).
640 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
workow aims at the extraction of the local properties of the
linear event of interest (surface- or guided-wave modes) and
makes use of redundancy in the data to remove the eect of
the propagation path from the source to the analysis point via
extraction of the local average-phase gradient. Te analysis
can be run on source and receiver lines for typical 3D ac-
quisition geometries and results are merged into a volume
representing the surface-wave properties within a survey. An
example of continuous proling along a 12-km receiver line is
shown in Figure 2. Te main two modes identied in Figure
1c (labeled 1 and 2) are extracted along the seismic line and
shown in Figures 2a and 2b respectively. In this representa-
tion, each vertical line represents the dispersion curve for that
location.
Surface-wave inversion
Inversion of phase velocities at each location provides access
to the medium velocities. Tree challenges can be identied
with dispersion curve inversion, especially when looking at
guided waves:
1) Curve inversion traditionally needs identication of the
order (nature) of the modes. Mode identication is mainly
limited by the bandwidth of recorded data, in particular
at low frequencies. Furthermore, guided P-waves do not
have a characteristic reference mode like the fundamental
Rayleigh mode for S-waves.
2) Guided P-waves can interfere with converted-wave modes
and guided S-waves. In this case, mode identication is
di cult because the dispersion curves split into short
pieces.
3) Surface waves propagate as normal modes dened by the
real-valued roots of the dispersion equation whereas guid-
ed wavesespecially P-guided wavesin many condi-
tions propagate as leaking modes dened by the complex-
valued roots of the dispersion equation (Aki and Richards,
1980), which can be di cult to deal with.
To address these challenges, we follow the approach pro-
posed by Ernst (2007), which involves minimizing the de-
terminant of the stiness matrix: an implicit function whose
zeros are the solution of the secular function and correspond
to modal curves. In particular, we consider the mist function
proposed by Maraschini et al. (2010) based on the Haskell-
Tomson matrix method adapted to take into account leak-
ing modes (Boiero et al., 2009). Tis mist function allows
surface and guided modes to be inverted without the need to
associate experimental data points to a specic mode, thus
avoiding mode misidentication errors in the retrieved veloc-
ity proles.
Figure 3 schematically shows the inversion scheme. On its
left hand side, black asterisks represent the dispersive events
estimated for a synthetic example at a certain location. Sev-
eral modes of Rayleigh waves and guided P-waves can be
recognized: the two events with higher-phase velocities. Te
background orange shadows represent the real and complex
solutions of the secular function for the velocity model on the
right hand side.
Te inversion algorithm modies S- and P-velocities in
order to match the estimated dispersive events with the secu-
lar function solutions. Tis approach does not need to de-
scribe leaking modes as acoustic or pseudo-acoustic modes
(Roth and Holliger, 1999; Shtivelman, 2004), which is a
Figure 3. Schematic representation of the inversion scheme. On the
left side, black asterisks represent the dispersive events estimated for
a synthetic example at a certain location. Orange shadows in the
background are the real and complex solutions of the secular function
for the velocity model on the right side. From top to bottom, the
inversion algorithm modies the S- and P-velocities to match the
estimated dispersive events with the secular function solutions.
ACHIEVE SUPERIOR ACCURACY AND PRECISION WITH TRANSFORM 3.3,
FEATURING ANALYTIC INTERPRETATION & MODELING
Addressing the complexity and subtlety of todays conventional and unconventional reservoirs, Transform 3.3
unleashes the power of Transforms new Analytic Interpretation & Modeling (AI&M) platform. Tackling the
expanding challenges of data overload, AI&M focuses precious E&P skills on what uniquely defines sweetspot
trends and optimal engineering. Spanning all disciplines and all E&P phases, AI&M enables operators to connect
the dots and gain key insights into what drives better well performance. Interested in learning more?
Visit transformsw.com/aim.html.
ANALYTIC INTERPRETATION & MODELING
21st Centur y Tools for 21st Centur y Reser voirs
A New Look for the Leader in E&P Software Innovation
642 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
reasonable approximation when deal-
ing with soft saturated rocks with
high Poissons ratio but may not be
appropriate in other cases (Roth and
Holliger, 1999). Forbriger (2003)
also describes a robust procedure
that accounts for normal and leaky
modes with their amplitudes without
the need of approximations, but this
approach could not be used here be-
cause it does not allow for using the
redundancy of the data to remove the
eect of the propagation path from
the source to the analysis point via
extraction of the local average phase
gradient.
It is worth noticing that dier-
ent modes (surface or guided) may
be sensitive to dierent parameters
at dierent depths of the near-surface
velocity model (Liner 2012). Tis
also reminds us that the parameter-
ization, along with the regularization
in the vertical and horizontal direc-
tions and the use of a-priori informa-
tion, is an important aspect of the
inverse problem and should be evalu-
ated for each particular case (Taran-
tola, 2005). However, we shall not
discuss this here.
In the following, we describe the
possible surface and guided modes
that can be found in land environ-
ments and OBC and towed-streamer data from shallow
marine environments; how to analyze them measuring their
phase velocities; and how to invert them to build near-surface
P- and S-velocity models.
Near surface in a land environment
Te example data set was acquired using point receivers. Te
receiver line spans a distance of approximately 12 km. Figure
1a shows the complexity of the waveeld, including interfer-
ence of reected and refracted multiples of P- and S-waves
and of converted waves. Te guided P-waves are the disper-
sive events that appear with relatively high phase velocities
(event B in Figure 1a) while the guided S-waves have slower
phase velocities and overlap the ground-roll cone (mainly
composed by Rayleigh waves indicated as event A). By ana-
lyzing the local properties of surface and guided waves along
the receiver line (Figure 2), we can immediately observe
the high lateral variability of the phase velocities with an
abrupt transition at a distance of 9000 m. By observing the
Rayleigh-wave phase velocity behavior in Figure 2a, we can
predict a velocity inversion in the right side of the line. Tis
means that on the left we are dealing with a waveguide that
is mainly supporting Rayleigh modes and guided P-waves
propagating as leaking modes (Roth et al., 1998) while on
the right we are mainly observing the exural (Figure 2a)
and extensional (Figure 2b) waves that characterize the
propagation of pseudo-Lamb waves in sti layers overlaying
a weaker one (Ryden and Park, 2006).
Figure 4 shows the S-wave (4a) and P-wave (4b) velocity
model along the receiver line. Te two sections are inferred by
inverting the phase velocities (Rayleigh and guided waves) es-
timated at each location in Figure 2. Figure 4c shows the V
P
/
V
S
ratio with values oscillating around 2slightly higher than
2.5 for soft materials and lower than 2 for sti layerswhich
is in agreement with published data (Ivanov et al., 2006).
Figure 4d shows the solution of the secular function for two
dierent locations compared with the estimated phase veloci-
ties. A good match exists, indicating that the depth velocity
models are able to explain the phase velocity of the measured
modes.
Te near-surface model obtained by means of the inver-
sion process gives geometric information about the near-sur-
face layers and the geology and velocity distribution down
to the investigation depth. Te results of the inversion show
rapid changes of weathering, compaction, and lithology in
the near surface even within a single receiver line. Tese varia-
tions can be retrieved thanks to the high lateral resolution of
surface and guided waves.
Figure 4. (a) Near-surface model obtained from phase velocities in Figure 2: P-wave. (b) Near-
surface model obtained from phase velocities in Figure 2: S-wave. (c) V
P
/V
S
ratio. (d) Solutions
of the secular function for two dierent locations (719 m and 10,656 m distance, respectively)
compared with the estimated phase velocities.
June 2013 The Leading Edge 643
Nonreflection seismic and inversion of surface and guided waves
Seabed in a shallow marine environment
Seismic data acquired in shallow waters often display well-de-
ned dispersion patterns related to surface and guided waves.
Figures 1b and 1c show minimally preprocessed hydrophone-
component data from an OBC and towed-streamer sail line,
respectively. Te waves propagating as normal modes are rep-
resented by a low-velocity, low-frequency wavetrain identi-
ed with Scholte waves (events C and E). Te high-frequency
part of Scholte waves consists mainly of Stoneley waves local-
ized in the vicinity of the liquid/solid interface, whereas at
lower frequencies they consist of Rayleigh waves propagat-
ing in the layers below the seabed (Shtivelman 2004). As the
interface waves attenuate rapidly with increasing distance to
the liquid-solid interface, the receivers have to be as close as
possible to the seabed: OBC cables can easily record Scholte
waves as well as towed streamers in shallow water. Numerous
examples obtained in areas with various geological condi-
tions using dierent acquisition geometries show that most
of the energy of the waves is localized within a narrow range
of low frequencies limited to 220 Hz (Shtivelman, 2004).
Te phase velocities of Scholte waves are related to the shear-
wave velocities below the water-bottom and can be inverted
to estimate V
S
in the subwater layers.
Te guided waves propagating as leaking modes are com-
posed mostly of multiply reected P-waves, whereas their
resonant character is because of the leaking of S-waves out-
wards from the upper layers (events D and F in Figures 1b
and c). By inverting the guided-wave dispersion curves, the
vertical distribution of the P-wave velocity (V
P
) in the shal-
low sub-water layers can be estimated.
Figure 5 shows an example of continuous proling along
an OBC cable. Te phase velocities for the two Scholte-wave
modes and ve guided P-wave modes labeled in Figure 1e are
extracted along the line and shown in Figures 5a and 5b re-
spectively. In this representation, each vertical line represents
the dispersion curve for that location.
Figure 6 shows the P-wave (Figure 6a) and S-wave (Fig-
ure 6b) velocity model along the OBC cable. Te two sec-
tions are inferred by inverting the phase velocities estimated
at each location in Figure 5. Figure 6c shows the V
P
/V
S
ratio
with values ranging from 2.5 up to 9 in the shallower lay-
ers, which are in agreement with published data (Ritzwoller
and Levshin, 2002). Figure 6d shows the stack of the shal-
low section superimposed with results from acoustic full-
waveform inversion (FWI). Te velocity models are in good
agreement, highlighting the capability of guided P-waves for
retrieving and characterizing lateral variations in the shallow
sediments. From the P- and S- wave velocity models inferred
from guided and surface waves, we can observe a penetra-
tion depth for surface-wave inversion of about 200 m below
the seabed. Tis could be used for velocity inversion imag-
ing and the characterization of laterally variant complexities
within this layer. Figure 7 shows the solution of the secular
function for three dierent locations compared with the esti-
mated phase velocities. A good match exists, indicating that
the depth model is able to explain the phase velocity of the
measured modes.
Figure 5. (a) Scholte-wave phase velocity distribution for two
dierent modes along the line (events in the inset in Figure 1e). (b)
Guided-wave phase-velocity distribution for ve dierent modes along
the line (events in Figure 1e).
Figure 8 shows an example of continuous proling
along a towed-streamer line following the same approach
as described for OBC data. Te phase velocities for the one
Scholte-wave mode and three guided P-wave modes labeled
in Figure 1f are extracted along the line and shown in Figures
8a and 8b, respectively. Figure 9 shows the P-wave (Figure
9a) and S-wave (Figure 9b) velocity models. Te two sec-
tions are inferred by inverting the phase velocities estimated
at each location in Figure 8. Figure 9c shows the V
P
/V
S
ratio
with values ranging from 2 up to 8 in the shallower layers,
which are in agreement with the previous example. Also in
this case the solution of the secular function for three dier-
ent locations is compared with the estimated phase velocities
(Figure 10). Te depth model is able to identify the lateral
variations that reect the variations of phase velocities of the
measured modes.
Conclusions
In exploration seismic surveys, surface and guided waves,
which constitute a large part of the radiation of seismic sourc-
es at the surface, can be analyzed and inverted to character-
ize the near surface. Te size and quality of modern seismic
644 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Figure 6. (a) Sea-bottom model obtained from phase velocities in Figure 5: P-wave. (b) Sea-
bottom model obtained from phase velocities in Figure 5: S-wave. (c) V
P
/ V
S
ratio. (d) Shallow
stack and P-wave velocity model from acoustic FWI. Te white line is a ducial marker and is not
related to any interpretation.
Figure 7. Solutions of the secular function for the sea-bottom velocity model in Figure 6 at three
dierent locations (1347 m, 2847 m, and 4843 m distance, respectively) compared with the
estimated phase velocities.
data sets enable the implementation
of a robust workow for the pro-
cessing of surface waves, providing
accurate estimates of their propaga-
tion properties with high lateral and
vertical resolution. By combining
dierent processing techniques, the
lateral resolution can be adapted to
the characterization objectives, from
small-scale perturbation correction
to long wavelength statics and ve-
locity modeling. A proper inversion
scheme can then infer S- and P-wave
velocities from the propagation prop-
erties of dierent type of surface and
guided waves that can be found in
land data and shallow marine en-
vironments, whether recorded by
OBC, ocean-bottom nodes (OBN)
or towed streamer.
Beside the use of the inverted
velocity models to support static
estimation and correction, mainly
for land environments, surface- and
guided-wave inversion has the capa-
bility to enhance conventional shal-
low velocity model building, depth
imaging and FWI e ciency and
results. Te assumptions typically
underpinning current acoustic FWI
methods are generally not justied in
the elastic near-surface environment,
and surface-wave inversion may pro-
vide P-wave velocity models that
can be incorporated into the FWI
initial model. In practice, the shal-
lowest tens of meters are notoriously
di cult to update using FWI. FWI
often uses Gardners rule to obtain
a density model, which is invalid in
the near surface for unconsolidated
sediments. Obtaining both P- and
S-wave velocity models in the near
surface through guided- and Scholte-
wave inversion, respectively, may
provide essential constraints on the
wet bulk density of unconsolidated
sediments, particularly in shallow
marine environments. It is also note-
worthy that surface-wave inversion is more cost-eective and
less computationally intensive than FWI.
For multicomponent data, a further benet of accurate
near-surface P- and S-wave velocity models is their poten-
tial contribution to improved PP-PS matching, imaging and
(joint) inversion. Generation of an accurate near-surface
model enables imaging processes to more accurately account
for dierences in traveltimes and paths rather that reliance on
1D time statics. Tis will result in more accurate depth im-
ages for both PS and PP data, better amplitude handling in
oset/angle gathers, and hence greater condence in match-
ing PS-PP events in migrated stacks and joint inversion using
the results of amplitude versus oset (AVO) analysis of the
gathers.
Te inferred near-surface properties can also be used to
design and optimize the coherent noise-ltering workow,
June 2013 The Leading Edge 645
Nonreflection seismic and inversion of surface and guided waves
and can be used for local-adaptive
lters. Model-based noise genera-
tion can be used to predict the co-
herent noise even beyond aliasing.
Finally, interpretation of the inver-
sion results can provide a robust
geological, structural, and lithologi-
cal model of the near surface from
which geotechnical parameters and
drilling hazards may be identied.
References
Aki, K. and P. G. Richards, 1980, Quan-
titative seismology: theory and meth-
ods: Freeman and Co.
Boiero, D., M. Maraschini, and L. V. Soc-
co, 2009, P and S wave velocity model
retrieved by multi modal surface wave
analysis: 71st Conference and Exhibi-
tion, EAGE, Extended Abstracts.
Ernst, F., 2007, Long-wavelength statics
estimation from guided waves: 69th
Conference and Exhibition, EAGE
Extended Abstracts.
Forbriger, T., 2003, Inversion of shallow-
seismic waveelds. Part 2: Inferring
subsurface properties from waveeld
transforms: Geophysical Journal In-
ternational, 153, no. 3, 735752,
http://dx.doi.org/10.1046/j.1365-
246X.2003.01985.x.
Haddon, R. A. W., 1984, Computation of
synthetic seismograms in layered earth
models using leaking modes: Bulletin
of the Seismological Society of Ameri-
ca, 74, 12251248.
Ivanov, J., R. D. Miller, J. Xia, D. Stee-
ples, and C. B. Park, 2006, Joint anal-
ysis of refractions with surface waves:
An inverse solution to the refraction-
traveltime problem: Geophysics, 71,
no. 6, R131R138, http://dx.doi.
org/10.1190/1.2360226.
Klein, G., T. Bohlen, F. Teilen, S. Ku-
gler, and T. Forbriger, 2005, Acquisi-
tion and inversion of dispersive seismic
waves in shallow marine environ-
ments: Marine Geophysical Research-
es, 26, no. 2-4, 287315, http://dx.doi.
org/10.1007/s11001-005-3725-6.
Liner, C. L., 2012, Elements of seis-
mic dispersion: A somewhat practi-
cal guide to frequency-dependent
phenomena: SEG, http://dx.doi.
org/10.1190/1.9781560802952.
Maraschini, M., F. Ernst, S. Foti, and
L. V. Socco, 2010, A new mist func-
tion for multimodal inversion of sur-
face waves: Geophysics, 75, no. 4,
Figure 8. (a) Scholte-wave phase-velocity distribution along the line (events in the inset in Figure
1f ). (b) Guided-wave phase-velocity distribution for three dierent modes along the line (events in
Figure 1f ).
Figure 9. (a) Sea-bottom model obtained from phase velocities in Figure 8: P-wave. (b) Sea-
bottom model obtained from phase velocities in Figure 8: S-wave. (c) V
P
/V
S
ratio.
646 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
G31G43, http://dx.doi.org/10.1190/1.3436539.
Muyzert, E., 2007, Near surface models derived from ground roll,
guided waves and Scholte waves: 69th Conference and Exhibition,
EAGE, Extended Abstracts.
Roth, M., K. Holliger, and A. G. Green, 1998, Guided waves in near-
surface seismic surveys: Geophysical Research Letters, 25, no. 7,
10711074, http://dx.doi.org/10.1029/98GL00549.
Roth, M. and K. Holliger, 1999, Inversion of source-generated noise
in high-resolution seismic data: Te Leading Edge, 18, no. 12,
14021406, http://dx.doi.org/10.1190/1.1438230.
Ritzwoller, M. H. and A. L. Levshin, 2002, Estimating shallow shear
velocities with marine multicomponent seismic data: Geophysics,
67, no. 6, 19912004, http://dx.doi.org/10.1190/1.1527099.
Ryden, N. and C. B. Park, 2006, Fast simulated annealing inversion of
surface waves on pavement using phase-velocity spectra: Geophys-
ics, 71, no. 4, R49R58, http://dx.doi.org/10.1190/1.2204964.
Shtivelman, V., 2004, Estimating shear wave velocities below the sea
bed using surface waves: Near Surface Geophysics, 2, 241247.
Socco, L. V., S. Foti, and D. Boiero, 2010, Surface-wave analysis for
building near surface velocity modelsEstablished approaches
and new perspectives: Geophysics, 75, no. 5, A83A102, http://
dx.doi.org/10.1190/1.3479491.
Figure 10. Solutions of the secular function for the sea-bottom velocity model in Figure 9 at three diferent locations (700 m, 2700 m, and
4850 m distance, respectively) compared with the estimated phase velocities.
Strobbia, C., A. Laake, P. Vermeer, and A. Glushchenko, 2011, Surface
waves: use them then lose them. Surface-wave analysis, inversion
and attenuation in land refection seismic surveying: Near Surface
Geophysics, 9, 503514.
Tarantola, A., 2005, Inverse problem theory and methods for model
parameter estimation: Society for industrial and applied math-
ematics.
Vignoli, G., C. Strobbia, G. Cassiani, and P. Vermeer, 2011, Statisti-
cal multiofset phase analysis for surface-wave processing in later-
ally varying media: Geophysics, 76, no. 2, U1U11, http://dx.doi.
org/10.1190/1.3542076.
Xia, J., R. D. Miller, and C. B. Park, 1999, Estimation of near-surface
shear-wave velocity by inversion of Rayleigh wave: Geophysics, 64,
no. 3, 691700, http://dx.doi.org/10.1190/1.1444578.
Acknowledgments: Te authors thank Statoil and WesternGeco for
permission to publish the data and WesternGeco for permission to
publish this work. Tanks also to our former colleague Claudio
Strobbia for valuable discussions and to Ayman Zaghloul for his
help in processing part of the data.
Corresponding author: dboiero@slb.com
2003
2013
CELEBRATING
10 YEARS
PROCESSING
PSET
TM
FRACSTAR
TM
BURIEDARRAY
TM
PRODUCTIVE-SRV
TM
SURFACE ARRAY
NEAR-SURFACE ARRAY
PROPPED FRACTURE ESTIMATION
Microseismic events and associated focal mechanisms
provide a wealth of diagnostic information, not least fracture
size and geometry. Stimulated reservoir productivity, however,
relies on fractures remaining open through the placement of
microseismically undetectable proppant.
Propped fractures can now be estimated by the integrated interpretation of microseismic events with treatment parameters
and target zone geomechanical characteristics. The multiple patent-pending Productive-SRV workflow models proppant
distribution more precisely, producing a well constrained volume of effective stimulation. The technique accounts for
proppant pumped, fracture size, orientation, aperture, and tortuosity, by stage within the target zone of interest.
This breakthrough technique correlates microseismic activity with 90 day production, helping operators optimize their
completions and stay ahead of the drill bit.
866.593.0032 / MICROSEISMIC.COM
H OU S T ON D E NV E R P I T T S B U R GH CA L GA R Y
C
M
Y
CM
MY
CY
CMY
K
MSI_FatFrac_Full-Page_Ad_SEG-TLE_June2013_rev01.pdf 1 5/7/2013 1:03:49 PM
2013 Europe Honorary Lecturer
Joint Inversion
The Way Forward to a
Comprehensive Earth Model
Marion D. Jegen, Geomar/Helmholtz Centre for Ocean Research, Kiel, Germany
For more information or to view
previous HL presentations, visit: www.seg.org/hl
Jegens presentation will overview how joint-inversion problems are formulated and solved. The application and
benefts of joint inversion will be illustrated through case studies in subsalt/sub-basalt exploration and quantifcation
of gas hydrate and gas in marine sediments.
DATE LOCATION SECTION
4 Jun ....... Istanbul, Turkey ................. Istanbul Chamber of Geophysical Engineers
6 Jun ....... Baku, Azerbaijan ................ Azerbaijan State Oil Academy
DATE LOCATION SECTION
18 Jun ..... Torino, Italy ........................ Politecnico di Torino Geophysical Society
24 Jun ..... Dublin, Ireland ................... Dublin Inst. of Advanced Studies (DIAS)
Sponsored by Shell
2003
2013
CELEBRATING CELEBRATING CCE EB B AT AT NGG CCELLEB BRRAT ATINGG CEL LEBRRATINNG LL RR NN
10 YEARS
PROCESSING
PSET
TM
FRACSTAR
TM
BURIEDARRAY
TM
PRODUCTIVE-SRV
TM
SURFACE ARRAY
NEAR-SURFACE ARRAY
PROPPED FRACTURE ESTIMATION
Microseismic events and associated focal mechanisms
provide a wealth of diagnostic information, not least fracture
size and geometry. Stimulated reservoir productivity, however,
relies on fractures remaining open through the placement of
microseismically undetectable proppant.
Propped fractures can now be estimated by the integrated interpretation of microseismic events with treatment parameters
and target zone geomechanical characteristics. The multiple patent-pending Productive-SRV workflow models proppant
distribution more precisely, producing a well constrained volume of effective stimulation. The technique accounts for
proppant pumped, fracture size, orientation, aperture, and tortuosity, by stage within the target zone of interest.
This breakthrough technique correlates microseismic activity with 90 day production, helping operators optimize their
completions and stay ahead of the drill bit.
866.593.0032 / MICROSEISMIC.COM
H OU S T ON D E NV E R P I T T S B U R GH CA L GA R Y
Nonreflection seismic and inversion of surface and guided waves
648 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Exploring nonlinearity and nonuniqueness in surface-wave
inversion for near-surface velocity estimation
W
ith the rapid increase in prospecting for unconventional
oil and gas, a large part of which remains on land,
the demand for land seismic data processing has increased
substantially and is expected to further increase in the
future. It is known that land seismic data are often of much
poorer quality than marine seismic data. Tis is to a large
extent caused by the presence of unconsolidated rock in the
near surface with often complex velocity structure, which is
absent in many marine settings. Such near-surface variations
cause the wavefeld to scatter or even lose its coherence as
it propagates. Tis makes it difcult to accurately image the
deeper-lying targets in land seismic data.
Moreover, the data contain surface waves that have sub-
stantially larger amplitudes than the refected body waves and
as such are considered noise that hides part of the refections
needed for accurate imaging. Besides attempts to increase the
data quality from a data-acquisition point of view, one can try
to improve the imaging with a data-processing point of view
and use the surface waves to invert for the near-surface (shear-)
wave velocity structure (e.g., Xia et al., 1999). Once this velocity
structure is known, it can be used to model the surface waves and
subtract them from the data to increase the signal-to-noise ratio
or, e.g., to calculate shear-wave receiver statics for converted-
wave seismic imaging. In that way, the hampered imaging of the
deeper lying targets can be improved.
Surface waves are sensitive to the Earths properties up to a
depth of roughly one wavelength. Terefore, with observed fre-
quencies typically somewhere between 3 and 30 Hz, assuming
typically observed velocities, they can be used to invert for the
Huub Douma, ION Geophysical/GXT Imaging Solutions
mattHew m. Haney, U.S. Geological Survey
(shear-wave) velocity up to depths of 100150 m (e.g., Ivanov et
al., 2006; Muyzert, 2007). Recently Haney and Douma (2012)
inverted group- and phase-velocity Rayleigh-wave dispersion
curves for the near-surface shear-wave velocity using a pertur-
bational approach applied to the forward method known as the
thin-layer method (Lysmer, 1970; Kausel, 1999).
Te forward problem of modeling dispersion curves for
surface waves, however, remains nonlinear. In linearized in-
versions, this inherent nonlinearity is evident because the
sensitivity kernels are dependent on the model parameters,
making the inversion dependent on the starting model.
Trough the mere acceptance of uncertainties in the data by
ftting the data up to a certain tolerance, the problem also be-
comes inherently nonunique. Even though linearized inver-
sions are in the daily practice of exploration geophysics often
the only practical option, they cannot deal with nonlinearity
and nonuniqueness. To get a feeling for the nonlinearity and
nonuniqueness of surface-wave inversion in an exploration
geophysical context, we compare the results obtained from a
linearized inversion with that of a nonlinear search technique.
For this article, we illustrate our methods and results by in-
verting a single dispersion curve obtained from feld data. We
focus on the inversion of the phase velocity only.
Data
Figure 1a shows a receiver gather that was band-pass fltered
between 2 and 15 Hz, with ofsets between 50 and 300 m.
By limiting the ofsets to a narrow region around the receiver
gather, spatial averaging or path efects are minimized, such
Figure 1. Band-pass fltered (215 Hz) common-receiver gather from a land data set with ofsets between 50 and 300 m showing a
fundamental-mode Rayleigh wave (a), and its associated phase-velocity spectrum (b). Te white dotted line in (b) is the picked dispersion curve.
Good signal in the phase-velocity spectrum was obtained only between 4 and 8.25 Hz.
June 2013 The Leading Edge 649
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
that the fnal inverted shear-wave velocity profle is indeed
mostly associated with the local area around the receiver
gather. Te surface (Rayleigh) wave is clearly visible and can
be used in a slant-stacking (or beamforming) procedure as a
function of frequency to obtain the phase-velocity spectrum
(Van der Kruk, 2007). Because of the irregular distribution
of ofsets and azimuths, the surface wave has a somewhat
ragged appearance.
Figure 1b shows the associated velocity spectrum and the
picked dispersion curve (white line). Te dispersion curve was
picked through fnding the global maximum in the velocity
spectrum and following the maximum in a search window in
the direction of both increasing and decreasing frequency. In
this way, mode jumping can be minimized, although this receiver
gather does not contain higher modes beyond the fundamental
mode. A good quality dispersion curve could be obtained only
over a narrow bandwidth (here 48.25
Hz). Te dispersion curve was picked at
a frequency interval of 0.25 Hz.
Linearized inversion results
For the inversion, we chose a layer
thickness for the forward modeling of
3 m. With a maximum frequency of
8.25 Hz in the dispersion curve and
minimum observed phase velocity of
about 350 m/s, this layer thickness is
less than a tenth of the estimated short-
est wavelength of 350/8.25 = 42 m.
Tis ensures the accuracy of the thin-
layer method used in the forward mod-
eling (Kausel, 1999). We imposed an
exponential smoothness constraint on
the inversion using a model-covariance
matrix with a 1/e length of 9 m (about a
quarter of the smallest wavelength) and
a model standard deviation of 34 m/s.
Te data standard deviation was set to
10 m/s for all frequencies.
Te Rayleigh-wave phase-velocity
is primarily sensitive to the shear-wave
velocity. Terefore, in practice, only
the shear-wave velocity can be reliably
estimated in phase-velocity disper-
sion-curve inversion. Te sensitivity
to the P-wave velocity and the density
are taken into account by assuming a
constant V
P
/V
S
ratio and using Gard-
ners relation to relate density to the
P-wave velocity.
To set up a linearized inversion based
on the thin-layer method, a linearized
perturbation analysis is used. Because the
matrix in the obtained linear inverse rela-
tion is sparse, we use the LSQR algorithm
of Paige and Saunders (1982). Conver-
gence is established when
2
< 1.
We used two diferent starting models for our linearized
inversion: a linearly increasing shear-wave velocity with depth
(Figures 2a and 2b), and a smoothed version of a low-velocity
layer on top of a half-space (Figures 2c and 2d). As can be
seen in Figure 2e, both fnal models ft the data well (within
one standard deviation because convergence was found for
both models with
2
< 1). When comparing both models
(Figure 2f ), we note that the models difer mostly in the up-
per 20 m and below 50 m. When looking at the update of
the models as a function of iteration (Figures 2b and 2d), we
notice that below 7080 m both fnal models are not much
diferent from their respective initial models.
Figure 3 shows the mode shapes for the fundamental-
mode Rayleigh wave for both the horizontal and vertical par-
ticle velocities, using the linear initial velocity model. It is clear
from Figure 3 that below 7080 m there is little sensitivity
Figure 2. Phase-velocity spectrum with superimposed modeled dispersion curves as a function of
the iteration number for a linearized inversion using a linear starting model (a) and the related
shear-wave velocity models (b). Te starting model is indicated by the black dashed line in (b)
and the associated dispersion curve is indicated by the black dashed line in (a). Te diferent
model updates are shown from the frst iteration (dark green) to the fnal model in bright green.
Te associated dispersion curves in (a) have the same colors as the models in (b). (c) and (d) are
similar to (a) and (b) but for a starting model that is a smoothed version of a low-velocity layer
on top of a half-space. (f ) and (e) compare the fnal models obtained using both starting models
and their associated dispersion curves, respectively. Te white dashed line in (a), (c), and (e) is
the measured dispersion curve.
650 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
to the model parameters. Tis lack of
sensitivity is caused by the absence of
low frequencies (< 4 Hz).
Te results from this linearized
inversion emphasize the nonlinear
character of the surface-wave disper-
sion-curve inversion; the inversion
can get stuck in a local minimum of
the objective function, making the
fnal model dependent on the start-
ing model. Without knowledge of
the uncertainties in the model pa-
rameters, there is no way of knowing
which model to choose, as both mod-
els ft the data equally well (within
one standard deviation,
2
< 1). To try
and get an idea of the uncertainty in
the model parameters, we employed
a nonlinear search method to search
the model space for an ensemble of
models that ft the data equally well.
Nonlinear search and the neigh-
borhood algorithm
Tere are many diferent methods to
search a model space, amongst which
are genetic algorithms, neural net-
works, and simulated annealing, to
name but a few. Here we chose the
neighborhood algorithm (NA) by
Sambridge (1999). Te choice of this
algorithm was motivated by its sim-
plicity in both parameterization and
implementation as well as its previous
use in surface-wave dispersion-curve
inversion (e.g., Huang et al., 2010).
Figure 4 explains how the NA
works for the simple case of a two-pa-
rameter model space. Te algorithm
is initialized by a random distribu-
tion of initial models in the model
space (10 in this case). For each ini-
tial model, the algorithm performs
the forward modeling and calculates
the associated misft. Subsequently, it
chooses the (user-defned) N
r
models
with the lowest misfts (two in Figure
4b, indicated by the green regions)
and repopulates each nearest neigh-
borhood region of these models with
a (user-defned) number N
s
of new
models (one in Figure 4c, indicated
by the red dots). For each new model,
the associated misft is calculated and
again the N
r
models with the low-
est misft are chosen (the orange re-
gions in Figure 4d) and their nearest
Figure 4. Cartoon illustration of the neighborhood algorithm for a two-parameter model space.
Te algorithm is initialized with 10 random models (a), and N
r
= 2 regions of lowest misft are
selected as shown by the green regions in (b) and each repopulated with N
s
= 1 new model as shown
by the red dots in (c). Te algorithm proceeds by calculating the misfts of the newly added models
and continues to again select the N
r
= 2 regions of lowest misft as shown by the orange regions in
(d) and repopulate each of these regions with N
s
= 1 new model as shown by the green dots in (d).
Te algorithm proceeds in the same fashion until a (user-defned) maximum number of iterations
is reached.
Figure 3. Mode shapes of the fundamental-mode Rayleigh wave for the horizontal particle velocity
(a) and the vertical particle velocity (b) for one of the starting models. It is clear that there is little
sensitivity to the model parameters below about 70 m.
June 2013 The Leading Edge 651
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
neighborhood regions repopulated with N
s
new models each.
Te algorithm proceeds onward in the same fashion until a
maximum number of iterations is reached.
Te NA thus zooms in on the regions of model space that
most reduce the misft. It is able to avoid getting trapped in a
local minimum because it resamples multiple nearest neigh-
borhood regions with the lowest misfts. As such, it has the po-
tential to fnd a global minimum, although convergence to this
minimum, as with most search algorithms, is not guaranteed.
To reduce the number of parameters, we enforced a regular-
ized inversion grid where three consecutive layers have the same
shear-wave velocity (and thus the same density and compres-
sional-wave velocity). Tis gives an efective layer thickness of
3 3 = 9 m which is about 1/4 of the smallest (approximate)
wavelength of 42 m. In this way, we reduce the size of the model
space considerably while maintaining the necessary resolution.
Te forward modelling grid still has a smallest layer thickness of
3 m, ensuring the accuracy of the thin-layer method.
Te NA was initialized with 2500 uniformly distributed
random models. At each iteration, we decided to repopulate
the N
r
=100 best nearest neighborhood regions with two ad-
ditional models, leading to a total of 200 added models per
iteration. Te total number of iterations was set to 50. Tis
resulted in 12,500 models, shown by the gray lines in Figure
5c. Te bounds of the search space are clearly visible. Figure
5a shows the velocity spectrum and the measured dispersion
curve (white dashed line), as well as all the dispersion curves
(gray lines) associated with each model in the total ensemble.
It is clear that most models have associated dispersion curves
that do not ft the data. For comparison, the models and their
associated dispersion curves resulting from both linearized in-
versions are shown.
We chose a subset of the generated ensemble that best ft
the data. Tis resulted in an ensemble of 75 models shown by
the black lines in Figure 5c. Teir associated dispersion curves
are shown in Figure 5b, and it is clear that these models ft
Figure 5. Ensemble of shear-wave velocity models (gray lines) generated by the NA (c) and their associated dispersion curves as shown by the gray
lines in (a), and the ensemble of best 75 models as shown by the black lines in (c) and their associated dispersion curves as shown by the black
lines in (b). Note that the ensemble of best shear-wave velocity models ft the data well (b). Te white dashed line in (a) and (b) indicates the
measured dispersion curve. For comparison, both models from the linearized inversion and their associated dispersion curves are shown by the
blue and magenta curves in (b) and (c).
Figure 6. Cartoon illustrating that the variation of models in an ensemble (green ellipse) is largest in the direction of the long axis of the ellipse
(a), and smallest in the direction of the short axis (b). To obtain the most robust features in the ensemble (i.e., the ones that vary the least), the
ensemble needs to be projected onto the short axis of the ellipse (c).
652 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
robust pattern that the models have
in common is defned by the short
axis of the ellipse. Hence to extract
these patterns from the ensemble we
must project the ensemble onto the
short axis of the ellipse.
Te major axes of the ellipse are
found by the eigenvectors of the covari-
ance matrix of the models, where the
shear-wave velocities as a function of
depth are treated as the independent
parameters and the diferent model re-
alizations as diferent measurements of
these parameters. Te eigenvector with
the largest eigenvalue is then related to
the long axis of the ellipse, while the ei-
genvector with the smallest eigenvalue
is related to the short axis. Te eigenval-
ue itself is a direct measure of the vari-
ability of the models in the direction of
the associated eigenvector.
In the simple case of the cartoon
in Figure 6, it is easy to determine
which eigenvector to keep because
there are only two such vectors; one
with a bigger eigenvalue and one with
a smaller value. We would keep the
small eigenvalue and project the en-
semble onto its associated eigenvector
only. In case we have more than two
parameters, such as in our inversion
case, we need to look at the eigenval-
ue spectrum and see if we can identify
a knee-point or a break in the eigen-
value spectrum, above which the ei-
genvalues are substantially larger or
become larger more rapidly as a function of the eigenvector
index. In that case we project the ensemble onto the eigenvec-
tors with eigenvalues below the knee-point.
We can perform this ensemble inference on both the
whole ensemble generated using the NA (gray lines in Figure
5c) and on the ensemble of best models only (black lines in
Figure 5c). At frst sight, doing this for the whole ensemble
might not seem meaningful, because this ensemble contains
mostly models that do not ft the data (Figure 5a). However,
knowing that the NA focuses its model space sampling on
those regions that reduce the misft the most, it follows that
the most robust features in this ensemble are related to the
parameters that are best resolved by the data.
Plotting the eigenvalues of the covariance matrix of the
whole ensemble identifes a knee-point in the eigenvalue spec-
trum (Figure 7a) around eigenvector index 17 or 19. Project-
ing the whole ensemble onto these frst 17 or 19 eigenvectors
gives the projected ensemble shown in blue in Figure 7b and
Figure 7d, respectively. Te projected ensembles indicate that
the shear-wave velocity from 15 to 75 m is best resolved by
the data. From the linearized inversion, we already knew that
Figure 7. Eigenvalue spectra, (a) and (c), of the eigenvectors of the covariance matrix for the whole
ensemble of all models generated using the NA as shown by the gray lines in (b) and (d). Projecting
the ensemble on the frst 17 (a) or 19 (c) eigenvectors gives the fltered ensembles shown by the blue
lines in (b) and (d). Clearly the most robust information in the ensemble is that the shear-wave
velocity between about 1575 m is best resolved. Te deeper (> 75 m) and shallow (< 15 m) shear-
wave velocity appears to be poorly resolved because of the lack of low (< 4 Hz) and high (> 8.25
Hz) frequencies in the dispersion measurement.
the data well (cf. the dispersion curves related to both models
obtained using the linearized inversion shown by the magenta
and blue curves in Figure 5b). Te individual models exhibit
a trade-of relation between velocity and depth because no
a-priori smoothness constraint was imposed on the models,
as opposed to the linearized-inversion case. Note that both
models obtained using the linearized inversion appear to be con-
tained within the bounds of the ensemble of 75 best models.
Ensemble inference
Once we have an ensemble of models that ft the data to a pre-
scribed tolerance, we can ask what all these models have in com-
mon. Hence, instead of looking for an optimum solution as in
the linearized inversion case, we now ask what robust information
we can infer from the ensemble. To do so, we proceed in the
same way as Douma et al. (1996).
Suppose that, for the sake of argument, we can depict our
ensemble of models as an ellipse in a two-dimensional model
space (Figure 6). In that case, it is clear that the models vary
the most in the direction of the long axis of the ellipse, and
the least in the direction of the short axis. Tat is, the most
June 2013 The Leading Edge 653
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
there was little resolution below 7080 m, but the ensemble
inference applied to the whole ensemble generated using the
NA search method seems to further reveal that the shallow
shear-wave velocity is poorly constrained by the data.
To confrm this somewhat surprising observation in the
context of the linearized inversion, we calculated the resolu-
tion matrices for the linearized inversions for both fnal mod-
els (Figure 8). If resolution would be perfect, the resolution
matrix would be the identity matrix. For both models, how-
ever, it is clear that at shallow depths (< 20 m) the resolution
matrices contain substantial side lobes away from the diago-
nal. Tis is consistent with the above conclusion drawn from
the ensemble inference on the whole ensemble of models
(Figure 7) and explains why for the linearized inversion the
models obtained with diferent starting models difer consid-
erably at shallow depths (Figure 8c).
Figure 8. Resolution matrices for the linearized inversion with both the smoothed layer over a half-space initial model (a) and the linear initial
model (b). Above 20 m, the resolution matrices contain strong side lobes away from the diagonal and below 70 m the matrices are near zero. At
the depths between 20 and 70 m, the resolution matrices are reasonably diagonal, indicating good resolution at this depth range only. Te fnal
models obtained using the linearized inversion for both starting models are shown in (c) for comparison.
MERLIN
+
provides a unique combination of paleogeography, paleo-Earth systems and predictive models.
Subscribers to MERLIN
+
gain access to a unique, innovative, data constrained solution that delivers genuine
commercial advantage in seeking investment opportunities in under-explored and frontier basins.
Visit our website to learn more at robertson.cgg.com/merlinplus
Paleo-Earth Wizardry
MERLIN
+
: A wizard at reducing exploration risk
cgg.com/robertson
654 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
We attribute the lack of resolution at the shallow depths
(< 20 m) to the maximum frequency in our dispersion curve
data being limited to 8.25 Hz. It appears that without higher-
frequency surface waves the shallow shear-wave velocity can-
not be accurately determined. Tis is important, since the
majority of the problems with the near surface often stem
from the complexity of the Earth in the frst 1020 m. Tere-
fore, when using dispersion-curve surface-wave inversion to
try and fnd this shallow complexity, it is important to focus
on methods of dispersion-curve estimation that maximize the
largest usable frequency in the data. Possible measurement
and inversion of other physical quantities (such as the H/V
ratio) could be considered when inverting for the shallow
shear-wave velocity.
By applying the ensemble inference to the ensemble of
75 best models, we obtain the eigenvalue spectrum shown in
Figure 9a. Te only clear break in the eigenvalue spectrum
appears to be at eigenvector index 23 (Figure 9c). By pro-
jecting the ensemble on the frst 23 eigenvectors, we get the
projected ensemble shown by the green lines in Figure 9d.
Tis projection reduces the model parameter uncertainty for
depths shallower than about 70 m when comparing the range
of values of the shear-wave velocity in
the projected ensemble (green lines)
with the original ensemble of best 75
models (black lines) but not much for
depths below 70 m. When, using in-
stead the frst 14 eigenvectors for the
projection (Figure 9b), we see that
the uncertainty is further reduced,
while the fnal models obtained from
the linearized inversion remain con-
tained within the bounds of the pro-
jected ensemble for the well-resolved
depth range of 1575 m. Hence,
both models obtained from the lin-
earized inversion seem to be consis-
tent with the robust information in
the ensemble. Tis confrms that the
linearized inversion seems to perform
well in the well-resolved depth range
of 1575 m. As such, it seems both
models are both good estimates of the
shear-wave velocity at those depths.
We observe that in the well-
resolved part of the model space,
for the projections in Figure 9b and
Figure 9d, the models resulting from
the linearized inversion tend to be on
the low end of the shear-wave veloc-
ity range indicated by the projected
ensemble. Tis is likely caused by the
smoothness constraint that was im-
posed on the linearized inversion, be-
cause this constraint is known to in-
clude a minimization of the norm of
the model (Yanovskaya and Ditmar,
1990). Tis causes the linearized inversion to tend toward the
model with the minimum shear-wave velocity. No smooth-
ness constraint was imposed on the nonlinear search.
Te fltered ensemble shows a range of possible shear-
wave velocities at each depth. Terefore, the fltered ensemble
can be used to estimate the uncertainty of the shear-wave
velocity. However, the range of estimate shear-wave veloci-
ties depends on the number of eigenvectors used to flter the
ensemble. When there is a clear knee-point or break in the ei-
genvalue spectrum, it is easy to fnd the number of eigenvec-
tors on which to project. However, when no clear knee-point
or break is present, it becomes harder to determine how many
eigenvectors to use for the projection. To avoid ambiguity in
choosing the number of eigenvectors to use and thus deter-
mining the uncertainty in the model parameters, a more ob-
jective criterion would be desirable.
Conclusion
We have explored the nonlinearity and nonuniqueness of the
inversion of fundamental-mode Rayleigh-wave dispersion
curves by comparing linearized inversion results based on the
fnite-element thin-layer method to the results obtained from
Figure 9. Eigenvalue spectra (a) and (c) of the eigenvectors of the covariance matrix for the
ensemble of best 75 models shown by the black lines in (b) and (d). Projecting the ensemble on the
frst 14 (a) or 23 (c) eigenvectors gives the fltered ensembles shown by the green lines in (b) and
(d). Te fnal models from both linearized inversion are shown for comparison as shown on the
blue and magenta lines in (b) and (d) as well as their associated initial models (the dashed blue
and magenta lines).
June 2013 The Leading Edge 655
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
ensemble inference of an ensemble of models generated
through a nonlinear search method (the neighborhood al-
gorithm). Te ensemble inference applied to the whole en-
semble highlighted the depths that were well resolved, and
emphasized the need for high frequency dispersion mea-
surements to be able to resolve the near-surface (<1020
m) shear-wave velocity. In the well-resolved depth-range,
the results from two linearized inversions with diferent
starting models compared well with the results obtained
from the nonlinear search.
References
Douma, H., R. Snieder, and A. Lomax, 1996, Ensemble inference
in terms of empirical orthogonal functions: Geophysical Jour-
nal International, 127, no. 2, 363378. Yanovskaya, T. and P.
Ditmar, 1990, Smoothness criteria in surface wave tomography:
Geophysical Journal International, 102, no. 1, 6372.
Haney, M. and H. Douma, 2012, Rayleigh wave tomography at
Coronation Field, Canada: Te topography efect: Te Leading
Edge, 31, no. 1, 5461, http://dx.doi.org/10.1190/1.3679328.
Huang, H., H. Yao, and R. van der Hilst, 2010, Radial anisotropy
in the crust of SE Tibet and SW China from ambient-noise
interferometry: Geophysical Research Letters, 37, no. 2, n/a,
http://dx.doi.org/10.1029/2010GL044981.
Ivanov, J., R. Miller, P. Lacombe, C. Johnson, and J. Lane Jr.,
2006, Delineating a shallow fault zone and dipping bedrock
strata using multichannel analysis of surface waves with a
land streamer: Geophysics, 71, no. 5, A39A42, http://dx.doi.
org/10.1190/1.2227521.
Kausel, E., 1999, Dynamic point sources in laminated media via the
thin-layer method: International Journal of Solids and Struc-
tures, 36, no. 3132, 47254742, http://dx.doi.org/10.1016/
S0020-7683(98)00262-5.
Lysmer, J., 1970, Lumped mass method for Rayleigh waves: Bul-
letin of the Seismological Society of America, 60, no. 1, 89104.
Muyzert, E., 2007, Seabed property estimation from ambient-
noise recordings: Part 2Scholte-wave spectral-ratio in-
version: Geophysics, 72, no. 4, U47U53, http://dx.doi.
org/10.1190/1.2719062.
Paige, C., and M. Saunders, 1982, Algorithm 583, LSQR: Sparse
linear equations and least-squares problems: ACM Transactions
on Mathematical Software, 8, no. 2, 195209, http://dx.doi.
org/10.1145/355993.356000.
Paige, C., and M. Saunders, 1982, LSQR: an algorithm for sparse
linear equations and sparse least squares: ACM Transactions on
Mathematical Software, 8, no. 1, 4371.
Sambridge, M., 1999, Geophysical inversion with a neighborhood
algorithmI. Searching a parameter space: Geophysical Jour-
nal International, 138, 479494.
Van der Kruk, J., S. Arcone, and L. Liu, 2007, Fundamental and
higher mode inversion of dispersed GPR waves propagating
in an ice layer: IEEE Transactions on Geoscience and Remote
Sensing, 45, no. 8, 24832491, http://dx.doi.org/10.1109/
TGRS.2007.900685.
Xia, J., R. Miller, and C. Park, 1999, Estimation of near-surface
shear-wave velocity by inversion of Rayleigh waves: Geophysics,
64, no. 3, 691700, http://dx.doi.org/10.1190/1.1444578.
Acknowledgments: We thank ION Geophysical/GXT Imaging
solutions for permission to publish these results.
Corresponding author: hdouma@iongeo.com
With OMNI 3D 2013 seismic survey design
& modeling software you can plan, execute,
monitor, and analyze multiple land, marine,
transition zone, vertical seismic profle, and
multicomponent surveys in the same project
simultaneously on a Windows
operating system.
Visit us at EAGE stand no. 960
www.GEDCO.com/OMNI3D
All the Survey Design Tools
You Need in One Package.
OMNI 3D 2013
*
M
a
r
k
o
f
S
c
h
l
u
m
b
e
r
g
e
r
.
2
0
1
3
S
c
h
l
u
m
b
e
r
g
e
r
.
1
2
-
X
X
-
X
X
X
X
A Schlumberger Company
O
M
N
I
3
D
i
s
a
r
e
g
i
s
t
e
r
e
d
t
r
a
d
e
m
a
r
k
o
f
S
c
h
l
u
m
e
r
g
e
r
.
O
t
h
e
r
c
o
m
p
a
n
y
,
p
r
o
d
u
c
t
,
a
n
d
s
e
r
v
i
c
e
n
a
m
e
s
a
r
e
t
h
e
p
r
o
p
e
r
t
i
e
s
o
f
t
h
e
i
r
r
e
s
p
e
c
t
i
v
e
o
w
n
e
r
s
.
C
o
p
y
r
i
g
h
t
2
0
1
3
S
c
h
l
u
m
b
e
r
g
e
r
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
1
3
-
s
e
-
0
0
6
3
Nonreflection seismic and inversion of surface and guided waves
656 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
MASW for geotechnical site investigation
M
ultichannel anaylsis of surface waves (MASW) is a
seismic surface-wave technique developed specically
for near-surface applications at depths usually shallower than
a few tens of meters (Park et al., 1999). Since its introduction
in the late 1990s, use of the technique has rapidly increased
for two reasons: (1) it provides the shear-wave velocity (V
S
)
of ground materials, which is one of the most important
geotechnical parameters in civil engineering, and (2) it is
easier to use than other common seismic approaches (e.g.,
refraction, reection, and surface-wave surveys).
Elastic moduli are commonly used in geotechnical en-
gineering to describe the behavior of Earth materials under
stress, which is ultimately related to such tasks as properly
designing earthworks and structural foundations, risk assess-
ment under specic site conditions, and monitoring various
types of existing infrastructures for public safety. Among three
primary types of modulusYoungs (E), shear (), and bulk ()
modulithe rst two are most commonly used because of what
they represent. Youngs modulus simply describes the deforma-
tion tendency along the axis of stress, whereas the shear modulus
describes the tendency of shape deformation (shearing) that, in
turn, is related to the viscosity of material.
Youngs and shear moduli are determined from the pa-
rameters of density (), shear-wave velocity (V
S
), and Pois-
sons ratio () (Figure 1). From the two dening equations
shown in the gure, it is obvious that V
S
plays the most im-
portant role as it is included as squared terms. In addition, V
S
in reality changes through a broader range than density and
Poissons ratio. Terefore, accurate evaluation of V
S
can be ex-
tremely valuable in geotechnical engineering. As shown in the
equations, the shear modulus can be determined fairly accu-
rately once V
S
is known. On the other hand, Youngs modulus
requires Poissons ratio to obtain a comparable accuracy.
MASW provides V
S
information of ground materials by
processing Rayleigh-type surface waves that are dispersive
when travelling through a layered media (dierent frequen-
cies travel at dierent speeds). Tis dispersion property is de-
termined from a materials shear-wave velocity (V
S
) (by more
than 95%), P-wave velocity (V
P
) ( 3%), and density () (
2%). By analyzing dispersion properties, we can therefore de-
termine V
S
fairly accurately by assuming some realistic values
for V
P
and . Te accurate evaluation of the dispersion prop-
erty is most important with any surface-wave method in this
sense.
By using a 2D waveeld transformation (for example, f-k
transformation), the MASW method converts raw eld data
in a time-oset (t-x) domain directly into a frequency-phase
velocity (f-v) domain in which dispersion patterns are evident
through the waveeld maxima. Te remaining procedure ex-
tracts a dispersion curve of, usually, fundamental mode that
will be used in a subsequent process in search for the one-
dimensional (1D) V
S
prole. An accurate dispersion analysis
is obviously an important part of data processing, and this is
CHOON PARK, Park Seismic LLC
Figure 1. Dening equations for Youngs and shear moduli showing
relationship with shear-wave velocity ( V
S
) and other parameters.
Figure 2. A diagram showing the relative amplitude change with
oset among surface waves, body waves, and ambient noise indicating
that the most commonly used oset range for MASW data acquisition
is usually shorter than 100 m. Tis almost always falls into the
optimum oset due to the strong energy of surface waves.
Site Class S-velocity (V
S
)
(ft/s)
S-velocity (V
S
)
(m/s)
A
(Hard rock)
> 5000 > 1500
B
(Rock)
25005000 7601500
C
(Very dense soil and
soft rock)
12002500 360760
D
(Sti soil)
6001200 180360
E
(Soft clay soil)
< 600 < 180
F
(Soils requiring
additional response)
< 600, and meeting
some additional
conditions
< 180, and meeting
some additional
conditions
Table 1. NEHRP seismic site classication based on shear-wave
velocity ( V
S
) ranges.
June 2013 The Leading Edge 657
Nonreflection seismic and inversion of surface and guided waves Nonreflection seismic and inversion of surface and guided waves
as liquefaction evaluation, is related to the elastic property
of V
S
that is closely linked to the viscosity of material; the
lower the V
S
, the more viscous is the material. On the other
hand, ground amplifcation for a given earthquake mag-
nitude, which causes most earthquake-related damages,
changes with ground stifness at relatively shallow depths.
Based on the premise established from empirical studies that
the top 30 m infuences the most, and also from the fact that
the shear-wave velocity (V
S
) is the best indicator of stifness,
the average V
S
in the top 30 m (usually denoted as Vs30m)
is used as an important criterion in the design of building
structures. In general, a site with a lower Vs30m would be
subject to a greater ground amplifcation (and sufer more
damage from an earthquake).
Te National Earthquake Hazard Reduction Program
(NEHRP) established by the U.S. Congress in 1977 adopts
this criterion and classifes a site into one of several categories
(Table 1). Te International Building Code (IBC) published
the same classifcation designations in 2000 as one of the pa-
rameters that should be accounted for in structural design.
Calculation of the average V
S
for a certain depth range
(for example, the top 30 m) can be accomplished in two
ways: (1) based on relative thickness-contribution of each
layer (method 1 in Figure 3), and (2) based on the defni-
tion of velocitytotal distance (di) divided by total travel
time (ti) that is calculated by summation of thickness (di)
divided by velocity (Vsi) of each layer (method 2 in Figure
3). Both methods can yield signifcantly diferent results for
the same V
S
profle as illustrated by using a simple two-layer
V
S
profle. Vs30m as defned in International Building Code
(IBC 2000 and later editions) uses the second method, which
tends to put a heavier weight on the lower V
S
:
Vs30m = di / ti = 30 / (di/Vsi) (m/s) (1)
One of the most demanding applications for Vs30m
evaluation occurs in wind-turbine site characterization (Park
and Miller, 2005). In this case, the V
S
value provided by MASW is important
to account in the foundation design
not only for the potential earthquake
hazard, but also for the continuous and
prolonged vibration of the ground pro-
duced by rotating blades. Vs30m values
and corresponding site classes presented
in Figure 4 are selected from sites at sev-
eral diferent wind farms in the midwest
and the northeast. Tey are presented in
the typical format to deliver the results
to the engineers.
Site characterization of a potential
nuclear power plant
Another example of the application of
MASW for 1D (depth) site character-
ization comes from the seismic hazard
assessment of potential nuclear power
why it is often the signal-to-noise ratio (SNR) of dispersion
image that directly infuences reliability of MASW results.
A high SNR is required in all types of wave-based tech-
niques to achieve highly accurate results. Te surface-wave
method utilizes Rayleigh waves as signalthe most trouble-
some source-generated noise in the history of exploration
seismology, commonly known as ground roll. Surface waves
provide the highest SNR possible in any type of seismic ap-
proach. As a consequence, the feld operation for data acqui-
sition and subsequent data analysis become extremely simple
and efective, almost always ensuring the most reliable results.
MASW is the most advanced surface-wave method be-
cause of its full adaptation of the multichannel principles
long used in seismic exploration for natural resources. Figure
2 illustrates the tolerance in data acquisition with MASW by
showing that the common range of source-receiver ofsets re-
quired for most geotechnical projectsusually shorter than
100 mis optimal within which a high SNR is almost always
guaranteed. Te area too close to the source (for example, 5
m) is usually avoided because of the near-feld efects that pre-
vent full development of surface waves. On the other hand,
an excessively far ofset (for example, 100 m) is also avoided
because of far-feld efects that can make the energy level of
surface waves drop below that of ambient noise.
Because shear-wave velocity (V
S
) information is a good
indicator of the material stifness, MASW is often applied in
civil engineering to deal with mechanical aspects of ground
materials (for example, assessment of load-bearing capacity,
ground behavior under continuous and prolonged vibration,
and ground amplifcation and liquefaction potential under
earthquake). MASW also fnds application in mapping the
soil/bedrock interface, which is often more usefully and re-
alistically defned from the stifness concept than any other
characteristics (Miller et al., 1999).
Seismic site classifcation-Vs30m
One application of MASW in earthquake engineering, such
Figure 3. Two possible ways to calculate an average shear-wave velocity ( V
S
). Te second
method used for Vs30m tends to put a heavier weight on the lower V
S
.
658 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
plants that are routinely subjected to
machinery vibration and potential
ground motion from earthquake.
An instance recently implemented at
Tyspunt, South Africa, is presented.
To meet the increasing demands
for electricity generation, the gov-
ernment of South Africa is commit-
ted to the construction of several
new nuclear power plants, with the
coastal site at Tyspunt, west of Port
Elizabeth, being considered as one of
the sites for characterization (Figure
5) (Bommer et al., 2013). Although
South Africa is not a region of elevat-
ed seismicity, destructive earthquakes
have occurred. Te most recent had a
magnitude 6.2 and occurred in 1969.
Commissioned by the state-owned
energy utility (Eskom), the Council
for Geoscience (CGS), one of the
National Science Councils of South
Africa, conducted a seismic hazard
analysis following the most stringent
international standards. Te MASW
survey, adopted as one of the several
approaches for this comprehensive
analysis, was conducted at six dier-
ent locations in the area (Figure 5).
Te purpose was to evaluate V
S
struc-
ture to depths as deep as possible,
preferably down to 100 m.
Because of the unusually deep in-
vestigation depth being sought, both
active and passive surveys were con-
ducted using a 48-channel seismic
acquisition system and 4.5-Hz geo-
phones as receivers. Since all the sites
were in remote coastal areas without
strong vibration sources available,
such as tra c, passive surveys relied
on ocean activities for lower frequen-
cy surface waves (for example, 10 Hz
or lower). In addition, two dierent
active surveys were conducted at each
site: one with relatively short receiver
spacing (dx) of 1 m and a 5-kg sledge
hammer source, and another with
a longer dx of 4 m and a rock-drop
source facilitated by a tracked hoe (Figure 6). Te former
survey setup was designed to investigate relatively shallow
depths (for example, 30 m) and the latter was designed to
investigate deeper depths ( 50 m).
Most sites had soft sandy overburden of varying thick-
nesses, thereby attenuating surface waves quite rapidly,
especially in the short-spread surveys using a sledge hammer
source. Te long spread surveys with the rock-drop source
generated su cient energy at frequencies as low as 10 Hz and
lower at some sites (Figure 7). Te passive survey adopted a
two-dimensional cross-receiver array with a 10-m separation
between receivers (Figure 6).
Dispersion imaging results from these passive surveys also
showed remarkable energy at the lowest frequencies in the
range of 4 20 Hz (Figure 7). Te results from the long-spread
active surveys were quite similar, with dierences mainly in
Figure 5. Site map of a potential nuclear power plant in Tyspunt, South Africa, that shows six
MASW sites and deep borehole sites.
Figure 4. Shear-wave velocity (V
S
) proles selected from ve dierent wind-turbine sites that fall
into each dierent class in seismic site classication (as dened in Table 1).
June 2013 The Leading Edge 659
Nonreflection seismic and inversion of surface and guided waves
the higher-frequency content and resolution. All three types
of dispersion images were stacked on top of each other to
extend the usable bandwidth and increase the overall SNR
of images. Tis stacking also enhanced higher-mode patterns
that existed in dierent frequency bands on dierent images
(Figure 7).
Te inversion process to produce a 1D V
S
prole at each
site consisted of two phases. Te rst phase used only the
fundamental-mode (M0) curve to produce the rst approxi-
mation of the velocity prole. Ten, using this as an initial
model in the second phase, fundamental and higher-mode
dispersion curves were generated through the forward model-
ing process. Tese multimodal dispersion curves were then
examined against observed patterns in the stacked dispersion
image. Tis second phase of multimodal inversion was car-
ried out and repeated after manually changing the velocity
(V
S
) and thickness models until satisfactory matches were
found. Figure 8 shows the nal V
S
proles at all six sites
obtained through this two-phase inversion approach. Teo-
retical bounds for 50% change in dispersion curves are also
indicated in the prole. Borehole data from PS-suspension
logging are also presented in Figure 9 with their locations
marked on the map in Figure 5.
No borehole sites were close enough to any MASW site to
allow a meaningful direct comparison. Nonetheless, borehole
data can show possible V
S
ranges of overburden and bedrock
in the area. Tey show bedrock depths change signicantly
from one site to another in an unpredictable manner. Tey
indicate V
S
of overburden at about 200 m/s and that of bed-
rock at about 15003500 m/s with uctuations between
the two. MASW results also show V
S
values of bedrock in
a similar but slightly lower range and almost the same V
S
of
overburden (Figure 8). Depths of bedrock are also observed
changing without any predictable pattern.
Underground mine investigation
Another common application area of MASW is mapping
bedrock in depth and relative competence related to stress
Figure 6. A costal view (left) from an MASW site at the potential nuclear power-plant location in Tyspunt, South Africa. Ocean activities
generated surface waves for passive surveys that used a 2D receiver array (center). A rock-drop source using a tracked hoe (right) was used for the
active survey.
Figure 7. Dispersion images obtained from passive and active data
sets acquired at site 1 in Tyspunt, South Africa. Te image created
from combining (stacking) the two images is shown at the bottom.
660 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
these newly constructed highway segments as a means to
monitor the general condition of the bedrock. MASW sur-
veys were conducted as one of the approaches at the four
target locations marked on the map in Figure 10. Te main
purpose of the MASW surveying was to map the general to-
pography of bedrock and any other noteworthy subsurface
features that could be linked to potential progression of bed-
rock weakening or vertical migration of collapse structures.
To simultaneously survey two 12-ft wide lanes (both driv-
ing and passing), a specially built double land streamer was
from overburden and cultural activities. Te interface be-
tween overburden and underlying bedrock can be a sharp
boundary such as soil over competent basement rock, or a
gradational transition such as the buried bedrock inuenced
by a severe weathering process with no physically distinct
boundary. From a perspective of elastic property, the inter-
face is also a sharp boundary in the former case, whereas it is
a gradational change in the latter case because the weathered
top portion would consist of varying degree of rock stiness.
Tis suggests the stiness mapping by MASW would show
the interface from a highly realistic standpoint.
MASW is known to provide
highly eective and accurate informa-
tion about bedrock depth, especially
at depths shallower than 20 m or so.
Tis is because surface-wave disper-
sion properties are highly sensitive to
change in this depth range. Although
the shear-wave velocity (V
S
) of over-
burden can be accurately estimated,
V
S
of the bedrock tends to be slightly
underestimated as depth increases be-
yond the most sensitive range of 20
m unless special care is taken during
the initial model creation at the be-
ginning of inversion process.
Naturally, a common application
would be the bedrock mapping in as-
sociation with public safety where a
potential hazard of bedrock collapse
exists due to man-made or natu-
ral causes in the subsurface such as
mining and karst sinkhole develop-
ment. Mapping bedrock topography
can delineate the collapsed features,
whereas a zone of bedrock with sig-
nicantly lower V
S
than adjacent ar-
eas may indicate a potential for verti-
cal migration of a void.
In 2009, the Minnesota Depart-
ment of Transportation (MnDOT)
built a special type of pavement called
CRCP (continuously reinforced con-
crete pavement) along several seg-
ments of Trunk Highway (TH) 169
in Chisholm, Minnesota (Figure 10).
Tis construction followed several
surface collapse features in the area
near TH169 that were deemed to be
related to previous mining activities
for more than 100 years that left a
subsurface maze of abandoned mine
shafts and tunnels (Figure 10).
In 2011, the O ce of MnDOT
Materials launched a project that in-
cluded geophysical approaches to in-
vestigate subsurface conditions below
Figure 9. Deep borehole data from PS-suspension logging at six locations in Tyspunt, South
Africa.
Figure 8. MASW results of ve-layer shear-wave velocity ( V
S
) proles at six sites in Tyspunt,
South Africa.
June 2013 The Leading Edge 661
Nonreflection seismic and inversion of surface and guided waves
used to collect surface-wave data by using a 48-channel ac-
quisition system with each land streamer equipped with 24
4.5-Hz geophones installed at 1-m spacing (Figure 11). Te
left- and right-side land streamers (facing from source) were
connected to channels 124 and 2548 of the seismograph,
surveying on driving and passing lanes, respectively. A pow-
erful weight-drop source specially designed and built at the
University of Saskatchewan in Canada was used to generate
surface waves 6 m ahead and at the midpoint between the
two streamers (Figure 11). To minimize tra c control and to
avoid tra c-generated noise as much as possible, surveying
took place during the night.
Figure 12 shows typical eld records from each land
streamer and corresponding images of fundamental-mode
dispersion patterns that possess an almost ideal SNR (i.e.,
100% signal) in a broad frequency band of approximately
540 Hz. Figure 13 shows analyzed 2D shear-wave velocity
(V
S
) maps for the longest survey line on the eastbound lanes
(line 3) that were obtained with a maximum analysis depth
of 25 m.
Te bedrock surface is denoted by a relatively sharp tran-
sition boundary of velocities from approximately 200 m/s to
500 m/s. Te bedrock depth is shown to gradually increase
from about 7 m on the western end to the maximum depth
of about 20 m on the eastern end, and this general trend con-
formed to the boring results from several locations along or
near the surveyed line. Interoverburden layers of higher veloc-
ity materials are probably lenses of gravels and boulders. Tey
can be identied on both maps of driving and passing lanes,
appearing as localized lenses and continuous layers. Tis in-
terpretation is consistent with the general geology of the area
as conrmed from borings and other sources.
Although the two maps from each lane look identical at a
regional scale, dierences are noticeable when examined from
a local perspective. For example, bedrock is slightly deeper on
the eastern half of the passing-lane map, and interoverburden
layers have slightly dierent depths and lateral extent. Con-
sidering the identical and consistent acquisition conditions
Figure 10. Aerial map showing locations of four MASW survey lines
on Trunk Highway (TH) 169 near Chisholm, Minnesota. Locations
of mine properties and workings are also shown.
Figure 11. Te double land streamers (24-channel acquisition each)
built at the Minnesota Department of Transportation (MnDOT)
for the simultaneous MASW surveys over two (driving and passing)
lanes on TH169. Te weight-drop source shown in the inset used a
polyethylene impact plate that tends to increase surface-wave energy at
lower frequencies. To minimize tra c-generated noise and the burden
of tra c control, surveys took place during the night.
Figure 12. Typical eld records from MASW surveys on TH169, and
their corresponding dispersion images from each land streamer that
show almost ideal signal-to-noise ratio of 100% signal.
662 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Figure 13. MASW results of 2D shear-wave velocity (V
S
) maps for line 3 from the surveys on TH169. Results from left (channels 124, driving
lane) and right (2548, passing lane) land streamers are shown.
the two land streamers were subjected to, it is reasonable to
attribute these dierences to subtle subsurface realities.
References
Bommer, J. J., K. J. Coppersmith, E. Hattingh, and A. P. Nel, 2013,
An application of the SSHAC level 3 process to the probabilistic
seismic hazard assessment for the Tyspunt nuclear site in South
Africa: Proceedings, 22nd International Conference on Structural
Mechanics in Reactor Technology (SMiRT22).
Miller, R. D., J. Xia, C. B. Park, and J. M. Ivanov, 1999, Multichannel
analysis of surface waves to map bedrock: Te Leading Edge, 18,
no. 12, 13921396, http://dx.doi.org/10.1190/1.1438226.
Park, C. B., R. D. Miller, and J. Xia, 1999, Multichannel analysis
of surface waves: Geophysics, 64, no. 3, 800808, http://dx.doi.
org/10.1190/1.1444590.
Park, C. B. and R. D. Miller, 2005, Seismic characterization of wind
turbine sites near Lawton, Oklahoma, by the MASW method:
Kansas Geological Survey Open-le Report 2005-22.
Acknowledgments: I thank o cials at Eskom in South Africa for
permission to use the data sets in this article. Julian J. Bommer at
Imperial College, London, UK, and Artur Cichowicz at the Coun-
cil for Geosciences (CGS) in South Africa played critical roles in
getting permissions. I also acknowledge all those actively involved
in the eld operation during the MASW surveys at the Tyspunt
nuclear site. Cichowicz and Denver Birch from CGS made major
contributions to the MASW work. Henni de Beer of ESKOM
facilitated access to the site and provided assistance in clearing the
MASW test locations. Vincent Jele, Robert Kometsi, and Leonard
Tabane of the CGS assisted Birch and two TI team members, Ellen
Rathje and Adrian Rodriguez-Marek, with the MASW eld work.
Wits University provided some equipment for use in the active
MASW testing. Institute of Mining Seismology (IMS) performed
the passive MASW experiments. Special thanks to Jason Richter
at Minnesota Department of Transportation (MnDOT) for the
generosity in allowing TH169 data to be used for this publication
as well as sharing other related information.
Corresponding author: choon@parkseismic.com
now $0.99 purchase today!
N
e
w
SEG Technical Program Presentation Singles
Each year, SEG records a large number of speakers and
presentations from the Technical Program presentations and
now for the first time, you can purchase them individually.
Over 200 presentations from the 2012 SEG Annual Meeting
are currently available at introductory pricing with more
being added each day.
SEG Member pricing:
$
0.99
Nonmember pricing:
$
1.99
Choose from over 200
speaker presentations.
All available presentations include audio synchronized
with the speakers slides.
Each presentation is approximately 20 minutes in duration.
Browse topics and sessions on the SEG Web site.
Visit the SEG Online Store to purchase today!
www.seg.org/bookmart
MORE WAYS
TO SPOT THE
SWEET SPOT.
Identify fault and fracture networks from the basement to the surface. Map deep
or regional structural features that influence reservoir productivity. Delineate
brittleness or liquids variations on an areal basis. And spot near-surface geo-
hazards. With NEOS multi-measurement interpretation, you can do it all quickly.
And at a fraction of the cost of traditional ground-based data acquisition methods.
By integrating newly acquired airborne datasets with existing seismic and well
data, we deliver highly constrained 3D models of the subsurface and provide the
insights you need to make decisions about where to explore, lease, and target
future seismic investments. Find out more at neosgeo.com
G R A V I T Y | M A G N E T I C | E L E C T R O M A G N E T I C | R A D I O M E T R I C | H Y P E R S P E C T R A L | S E I S M I C
GAIN MULTI-MEASUREMENT INSIGHTS WITH NEOS.
Above, Bel ow and Beyond
SIGHTS WITH NEOSSSS.
Nonreflection seismic and inversion of surface and guided waves
664 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
Bedrock mapping in shallow environments using
surface-wave analysis
R
ayleigh-wave analysis is nowadays a standard tool for
retrieving near-surface S-wave velocity models. Te
method, usually based on the inversion of surface-wave
dispersion curves adopting a 1D forward operator, is most
often applied to laterally varying sites and often on long and
continuous seismic lines. Te processing is performed using
one of many available waveeld-transform techniques and
results in several local dispersion curves estimated along the
survey line. Te dispersion curves are inverted to provide
local S-wave velocity models which are merged to reconstruct
2D/3D structures.
If all stages of the survey are carefully planned, from ac-
quisition to inversion, the surface-wave analysis can provide
a signicant piece of information that, particularly if merged
with other geophysical and borehole data, can represent a key
in site characterization. In several engineering problems rang-
ing from seismic site-response studies to geotechnical charac-
terization, S-wave velocity is in fact an important parameter
that it is often di cult to estimate through surface-based
body-wave techniques.
Tis article presents two eld cases from Scandinavia
that represent successful examples of application of surface-
wave analysis to two engineering problems. Te rst case is
for water resource mapping in Lolland, Denmark, where the
near-surface S-wave velocity model retrieved from surface-
wave analysis is used together with high-resolution P-wave
reection seismic to map the uppermost portion of the bed-
rock (cretaceous chalk) in an area where the shallow depth
of bedrock (~20 m) make it a di cult target for seismic re-
ection. Te second case from Helsingborg, Sweden, is for a
geotechnical site characterization for a shallow railway tunnel
in a complex sedimentary rock environment with signicant
tectonics. Near-surface S- and P-wave velocity models, as well
as structural information, are retrieved from surface-wave,
refraction, and reection analysis of high-resolution P-wave
data. In both cases, data has been collected on roads in an
e cient manner using a small vibrator and a land streamer.
Te surface-wave surveys were performed according to
the following scheme:
1) Seismic data acquisition along several seismic lines
2) Estimation of a dense series of dispersion curves along the
lines
3) Dispersion curve QC
4) Monte Carlo inversion (MCI) of the dispersion curves to
set the reference model
5) Laterally constrained inversion (LCI) of the dispersion
curves to supply a pseudo-2D velocity model
DANIELE BOIERO, Politecnico di Torino, presently WesternGeco
LAURA VALENTINA SOCCO, Politecnico di Torino
STEFANO STOCCO, Gamut S.r.l.
ROGER WISN, Rambll Danmark A/S
6) Comparison of the obtained velocity models with other
geophysical investigation results and borehole data
Te surface-wave data acquisition was performed with
a land-streamer setup, designed by Rambll Danmark A/S,
utilizing geophones on stainless steel sledges with variable
spacing and a 3.5-ton vibrating source (Figure 1). Te ac-
quisition was optimized to reach the desired investigation
depth and data quality control was performed during the ac-
quisition stage. Te land streamer, specically designed for
surface-wave acquisition, consists of 77 4.5-Hz geophones,
with 1.25-m spacing for the rst 47 geophones and 2.5-m for
the last 30 geophones. Te streamer has a total active length
of 133.75 m. Te distance between the vibrator and the rst
geophone is 6.25 m. For the Lolland case, which was the rst
with this specic setup, the vibrator was complemented with
a 40-kg accelerated weight drop. A 15-s nonlinear sweep in
the frequency band of 680 Hz was used, with useful energy
being transmitted from around 8 Hz depending on site con-
ditions. Two separate data sets with dierent receiver spacing
are extracted for surface-wave processing from each record,
one with 48 receivers and 1.25-m interval and one with 55
receivers and 2.5-m interval.
A dense series of dispersion curves was then retrieved
through a processing tool (Gamut S.r.l.) developed to
handle the data acquired by the specic acquisition setup.
Te dispersion curves are estimated at each position of the
streamer by picking the energy maxima on the f-k spectrum
Figure 1. Te Rambll land-streamer setup consists of an IVI
minivib; a pulling vehicle equipped with vibrator controller,
positioning system and data QC facilities; and a streamer with
geophones on steel sledges with variable spacing. Te streamer comes in
two dierent versions, one for low-frequency surface-wave surveys and
one for higher-frequency high-resolution P-wave reection surveys.
June 2013 The Leading Edge 665
Nonreflection seismic and inversion of surface and guided waves
in which each model is linked to an experimental dispersion
curve. Moreover, neighboring 1D models are linked by lateral
constraints that impose similarity between neighboring mod-
el parameters of the same type. Te dierence with respect to
individual inversions of the retrieved dispersion curves is that,
in LCI, all the dispersion curves along the line are inverted
simultaneously, minimizing a common objective function.
Te number of output models is equal to the number of dis-
persion curves along the seismic line, but the models are not
independent and the pseudo-2D model that is obtained by
merging them is internally consistent.
Te data and the constraints act in a complementary man-
ner: where model parameters are poorly resolved by the data,
their values will be more strongly inuenced by constraints
and, hence, by the values of the same parameters in a position
and by processing the two dierently spaced congurations
separately. Consequently, for each shot gather, we estimated
two dierent dispersion curves that are spatially located at
the center of the relevant receiver spread. In this way, the
dispersion-curve data set to be inverted is constituted by a
series of dispersion curves estimated along the seismic line. A
thorough quality control has been applied to the dispersion
curves before inversion with the aim of selecting high-quality
fundamental-mode curves along the whole line.
Te inversion process of the surface-wave dispersion
curves suers from solution nonuniqueness and it is hence
strongly biased by the initial model. To mitigate this problem,
the inversion is based on a two-step process. It starts from a
broadband exploration of the model parameter space through
MCI and focuses on the high probability density (low-mist)
model space regions with a renement performed through
LCI. Both the inversion algorithms use Haskell-Tomson for-
ward modeling (Tomson, 1950; Haskell, 1953; Herrmann,
1996). Te unknown parameters are the S-wave velocity and
the thickness of the layers. Te densities of the layers and the
Poissons ratios (or P-wave velocity) are assumed a priori.
Te MCI algorithm (Socco and Boiero, 2008) uses scale
properties of Rayleigh-wave propagation to reduce the re-
quired number of simulations and e ciently explore the
model parameter space Te mist function accounts for
problem dimensionality (number of unknown parameters
with respect to available data points) and data uncertainty.
Te mist is evaluated for each prole of the population and
used to select acceptable models according to a statistical test
that selects all proles that are equivalent according to a giv-
en level of condence. Te selected models, hence, represent
a set of possible solutions which may be considered equally
probable, given the experimental data and their uncertainty,
the chosen parameterization and the level of condence. It
can be shown that the solutions of deterministic inversions
performed with dierent initial models fall in a wider model
region with respect to the models selected by the Monte Car-
lo inversion. Hence, the region where the selected proles fall
can be used as a reference for the denition of a consistent
initial model for LCI. If low-velocity layers or sti inclusions
are present, MCI allows them to be evidenced and accounted
for in the parameterization for LCI.
LCI, rst presented for the interpretation of resistivity
data (Auken and Christiansen, 2004), is a deterministic inver-
sion in which each 1D model is linked to its neighbors with a
mutual constraint to provide a single pseudo-2D model. Te
lateral constraints can be considered as a priori information
on the geologic variability in the investigated area: the smaller
the expected variation of a model parameter, the more rigid
the constraint. Moreover, it is possible to use any available a
priori information, (e.g., from drilling), to further constrain
the inversion. LCI has been validated through several applica-
tions to resistivity and electromagnetic data and has also been
successfully applied to surface-wave data (Socco et al., 2009).
Te local dispersion curves extracted along the line are
inverted through a least squares LCI algorithm to get a nal
pseudo-2D result. Te inversion result is a set of 1D models,
Figure 2. Lolland survey. (a) Site map and the seismic line with
CMPs for the reection line. (b) A seismic gather with sensor spacing
of 1.25 m. (c) Picked maxima and selection of the fundamental mode
on the f-k spectrum (normalized) of the seismogram shown above.
666 The Leading Edge June 2013
Nonreflection seismic and inversion of surface and guided waves
where the data contain greater information. Te strength of
the constraints can be tuned according to a priori information
on the geologic variability in the area or selected with several
tests as the highest possible value that does not produce an
increase in the mist with respect to unconstrained inversion.
In other words, constraints should not get in contrast with
the compliance with the data and allow lateral variations to be
depicted where present (Boiero and Socco, 2010).
Where a priori information is available from borehole and
other geophysical investigations, it can be either included in
the inversion as constraints on the initial model based on this
a priori information or used as a posteriori benchmark to
validate the results obtained by surface-wave analysis. In this
study, we have not included the a priori information in the
initial model.
Lolland
In 2009, Rambll performed a reection and surface-wave
seismic survey for the Danish Environmental Agency at
Lolland in the south of Denmark. Intensive groundwater
mapping is performed in Denmark as part of the Danish
environmental agencys work toward providing naturally
clean, untreated high-quality drinking water, a task that is
part of the Danish implementation of the European Unions
Water Framework Directive. Within this work, geophysical
surveys play an important role to cover a large enough part
of the Danish territory and are used together with several
other methods (e.g., drilling, geophysical borehole logging,
geochemical and hydrogeological mapping, and geological
modeling).
Te application of shallow reection seismic methods has
become an important tool and complements the electric and
electromagnetic methods, used successfully for cost-eective
mapping of shallow geology, with mapping of deeply (up to
several hundreds of meters) buried Quaternary valleys and
deep Tertiary aquifers as well as faults. Because the cost for
carrying out manual shallow onshore reection seismic sur-
veys acts as a limiting factor toward the use of seismic, the
development of land-streamer technology was started at the
shift of the century (Vangkilde-Pedersen et al., 2006). Te
land streamers developed at Rambll signicantly improved
the production rate and by now several generations of land-
streamer systems have been used successfully and more than
1000 km of seismic have been surveyed, both within the
groundwater mapping program and in other applications
(e.g., oil or gas prospecting and infrastructure). Land-stream-
er technology has recently also been used for dedicated low-
frequency measurements for surface-wave analysis. Together
with P-wave velocity models from refraction analysis of the
P-wave reection data, where applicable, the S- wave models
from surface-wave analysis provide important information for
static correction, interpretation of shallow features, and qual-
ity control of the near-surface parts of the reection results.
Te survey in Lolland consisted of 10.7 km of reection
seismic and 10.4 km of surface-wave proling. Te purpose
of the survey was to increase the overall understanding of the
geologic structures in the area; the results from the seismic
survey are used for building the geologic and hydrogeologic
models. In particular, the top of the PreQuarternary unit,
that consists of chalk, is shallow (~20 m) in the entire area
and the aim of surface-wave analysis is to map the surface of
the bedrock. Te data set consists of 507 common-shot gath-
ers. Each record contains 48 receivers (1.25-m spacing) and
Figure 3. (a) All fundamental-mode dispersion curves extracted
along the line shown in Figure 2. (b) Teir representation in the
distance-/2.5 domain.
Figure 4. Solution for the dispersion curve estimated from shot gather
20630 (2.5-m spacing). Te blue models on the left represent the lower
mist models with their relative dispersion curves (right upper panel).
Te red model will be used as the reference model. Te blue asterisks on
the right bottom panel correspond to the best-tting model.
June 2013 The Leading Edge 667
Nonreflection seismic and inversion of surface and guided waves
Broadband Digital Sensor
The Lightest System For High Density Surveys
A N Y WH E R E . A N Y T I ME . E V E RY T I ME .
Ahead of the Curve
SM
Exceeding todays highest channel requirements
and delivering the best signal, 428XL, with DSU1 is
WKHPRVWHIFLHQWV\VWHP
BROADBAND SIGNAL
Submit to INTERPRETATION:
INTERPRETATION seeks papers directly relevant to the practice of in-
terpretating the earths subsurface for exploration and extraction
of mineral resources and for environmental and engineering ap-
plications. Submit a paper at https://mc.manuscriptcentral.com/
interpretation.
Special Sections:
Each issue of INTERPRETATION will include at least one special section
focused on a particular topic. Watch for the rst issue of Interpretation
in August featuring a special section on interpreting stratigraphy from
geophysical data. Novembers issue will include a special section on
interpretation for unconventional resources.
Submissions are being accepted for the sections listed.
To submit a paper
to one of these
sections, visit https://
mc.manuscriptcentral.
com/interpretation and
select the appropriate
topic from the manuscript
type options.
To suggest a topic for
future special sections,
e-mail interpretation@seg.
org or contact one of the
editors.
www.seg.org/am
+Registration & Housing opens: 8 July 2013 +
S O C I E T Y O F E X P L O R A T I O N G E O P H Y S I C I S T S
Golf Tournament
GAC Lunches
Start assembling your teams!
This years golf tournament will take
place at the award winning Redstone
Golf Club. This club was voted Best
Golf Course of 2013 by National Golf
Course Owners Association. The
tournament will take place on Saturday,
21 September, before the annual
meeting begins Sunday.
We are excited to bring you
three technical lunches and four
GAC lunches in Houston. The
Gravity & Magnetics, Near Surface,
and Development & Production
committees will all hold technical
lunches. The Global Affairs Committee
will host four luncheons with regional
focus on Latin America, Africa/Middle
East, Asia/Pacifc and Former Soviet
Union/Europe. All lunches will take
place on Tuesday, 24 September
and Wednesday, 25 September.
George R. Brown Convention Center
2227 September 2013 Houston, Texas USA
The International Conference
for Geophysics
Photo courtesy of Redstone Golf Club
www.seg.org/am
+Registration & Housing opens: 8 July 2013 +
S O C I E T Y O F E X P L O R A T I O N G E O P H Y S I C I S T S
Golf Tournament
GAC Lunches
Start assembling your teams!
This years golf tournament will take
place at the award winning Redstone
Golf Club. This club was voted Best
Golf Course of 2013 by National Golf
Course Owners Association. The
tournament will take place on Saturday,
21 September, before the annual
meeting begins Sunday.
We are excited to bring you
three technical lunches and four
GAC lunches in Houston. The
Gravity & Magnetics, Near Surface,
and Development & Production
committees will all hold technical
lunches. The Global Affairs Committee
will host four luncheons with regional
focus on Latin America, Africa/Middle
East, Asia/Pacifc and Former Soviet
Union/Europe. All lunches will take
place on Tuesday, 24 September
and Wednesday, 25 September.
George R. Brown Convention Center
2227 September 2013 Houston, Texas USA
The International Conference
for Geophysics
Photo courtesy of Redstone Golf Club
714 The Leading Edge June 2013
The Leading Edge
Calendar
2013
Summer Research Workshop:
Unconventional Resources: Te
Role of Geophysics, Pittsburgh,
USA, www.seg.org/meetings/un-
conv13, (awatson@seg.org)
26 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Edin-
burgh, Scotland, www.seg.org/disc
5 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Stavan-
ger, Norway, www.seg.org/disc
17 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Bucha-
rest, Romania, www.seg.org/disc
20 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Milan,
Italy, www.seg.org/disc
24 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Massy,
France, www.seg.org/disc
26 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications
of Time-Lapse Seismic Data,
Rijswijk, Te Netherlands, www.seg.
org/disc
28 Jun
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Aber-
deen, Scotland, www.seg.org/disc
1 Jul
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Lon-
don, United Kingdom, www.seg.org/
disc
3 Jul
D&P ForumIntegrated Geophys-
ics for Unconventional Resources,
Krakow, Poland (ccoleman@seg.org)
711 Jul
Summer Research Workshop:
Seismic Uncertainties and Teir
Impact, Ban, Canada, www.seg.
org/meetings/Ban2013, (ppryor@
seg.org)
812 Jul
Near Surface Geophysics Asia Pa-
cic Conference, Beijing, China,
www.seg.org/meetings/nsgapc13,
(awatson@seg.org)
1719 Jul
IQ Earth Forum: Visualizing and
Predicting the Integrated Earth,
Boston, USA (ccoleman@seg.org)
48 Aug
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Perth,
Australia, www.seg.org/disc
9 Aug
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Mel-
bourne, Australia, www.seg.org/disc
11 Aug
ASEG-PESA 2013 23rd Interna-
tional Geophysical Conference,
Melbourne, Victoria, Australia,
(asbjorn_n_christensen@yahoo.com)
1114 Aug
Unconventional Resources Technol-
ogy Conference, Denver, USA,
www.URTeC.org
1214 Aug
Summer NAPE, Houston, USA, http://
www.napeexpo.com
1416 Aug
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Bris-
bane, Australia, www.seg.org/disc
16 Aug
SEG/ExxonMobil Student Education
Program, Rio de Janeiro, Brazil,
http:// www.seg.org/education/univer-
sity-student-programs/sep, (sep@seg.
org)
2426 Aug
AAPG/SEG Fall Student Expo, Hous-
ton, USA, http://www.studentexpo.
info
1617 Sep
International Conference & Exhibi-
tion on Reservoir Surveillance,
Xian, Shaanxi Providence, China,
(meeting@hxan.com)
1618 Sep
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Caracas,
Venezuela, www.seg.org/disc
17 Sep
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Hous-
ton, USA, www.seg.org/disc
20 Sep
SEG/ExxonMobil Student Education
Program, Houston, USA, http://
www.seg.org/education/universitystu-
dent-programs/sep, (sep@seg.org)
2022 Sep
SEG Annual Meeting Continuing
Education Courses, Houston,
USA, (www.seg.org/ce)
2122 Sep
SEG/Chevron Student Leadership
Symposium, Houston, USA, http://
www.seg.org/education/university-
student-programs/sls, (sls@seg.org)
2122 Sep
SEG International Exposition and
83rd Annual Meeting, Houston,
USA, www.seg.org/am
2227 Sep
SEG DISC: Making a Dierence
with 4D: Practical Applications of
Time-Lapse Seismic Data, Rio de
Janeiro, Brazil, www.seg.org/disc
3 Oct
The CPS/SEG Beijing 2014 International Geophysical Conference and
Exposition will be held in Beijing, China, 2124 April 2014. CPS/SEG
Beijing 2014 organizers invite you to submit abstracts to be considered
for oral or E-poster presentation at this upcoming meeting. Submissions
must conform to SEG formats described in the abstract kit, be written
in acceptable English, and contain high-quality graphics.
II and gas axpIoraIIon and davaIopmanI
L&F gaosrIanras rhaIIangas In 6hIna
8aIsmIr daIa arquIsIIIon IarhnIquas
8ImuIIanaous sourras
8roadband saIsmIr IarhnIquas
8aIsmIr prorassIng
8aIsmIr ImagIng
FuII WavaHaId InvarsIon & modaIIng
VaIorIIy modaI buIIdIng
8aIsmIr modaIIng and InvarsIon
8aIsmIr InIarpraIaIIon
8aIsmIr aIIrIbuIa anaIysIs and
InIarpraIaIIon
asarvoIr rhararIarItaIIon
LIarIrIraI and aIarIromagnaIIr maIhods
FoIanIIaI HaId maIhodssurh as ravIIy,
LIarIrIraI and MagnaIIr (LM)
8orahoIa gaophysIrs
ork physIrs
4lIIma Iapsa rasarvoIr monIIorIng
nronvanIIonaI rasourras davaIopmanI
MIrrosaIsmIr and passIva saIsmIr
FrarIurad rasarvoIrs
haar surIara gaophysIrs
LnvIronmanIaI and angInaarIng
gaophysIrs
MInIng gaophysIrs
aomarhanIrs & pora prassura anaIysIs
aorhamIsIry
as hydraIas
Isk anaIysIs In L&F
aophysIraI InsIrumanIaIIon and
aquIpmanI
InIagraIad InIarpraIaIIons sIudIas
6asa hIsIorIas
aIa managamanI
MuIIIWava and muIIIromponanI
saIsmoIogy
6oaI and 68M gaophysIraI axpIoraIIon
Ihar raIaIad raIagorIas
Call for Papers
Primary categories for the CPS/SEG Beijing 2014 Technical
Program are as follows:
Abstract kits will be available on
the SEG Web site in May.
Hardcopies can be requested
from the SEG Business Office.
DEADLINE FOR
ABSTRACT SUBMISSION:
18 October 2013,
5 P.M. U.S. CENTRAL
STANDARD TIME.
Coorganized by:
Chinese Petroleum Society
Society of Exploration Geophysicists
Do not miss this opportunity.
Make plans to submit your abstract
beginning on 1 August 2013.
CPS/SEG Beijing 2014
Geophysical Conference
and Exposition
P.O. Box 702740
Tulsa, OK 74170-2740
meetings@seg.org
www.seg.org/meetings/Beijing2014
For more information please contact:
Kristi Smith, CMP
Programs and Events Manager
meetings@seg.org
+1-918-497-5564
www.seg.org/meetings/Beijing2014
716 The Leading Edge June 2013
Applications for Active membership have been received
from the candidates listed below. Tis publication does not
constitute election but places the names before the member-
ship at large in accordance with SEGs Bylaws, Article III,
Section 5. If any member has information bearing on the
qualifcations of these candidates, it should be sent to the
president within 30 days. Te list can be viewed online at
membership.seg.org/applicants/.
For Active membership
Abdel Motagally, Ahmad (CGGVeritas, Nasr City,
Egypt, Arab Republic)
Ajewole, Tomas Olanrewaju (Petronas Carigali SDN.
BHD., Kuala Lumpur, Malaysia)
Bongiovanni, Nicolas (Denbury Resources, Plano, USA)
Denny, Stuart (Dolphin Geophysical, Houston, USA)
Fritz, Richard (SM Energy, Tulsa, USA)
Geck Alvarez, Amanda (Rock Solid Images, Houston,
USA)
Gilstrap, Tatiana (Halliburton, Houston, USA)
Golob, Bruce (ION-GXT, Lakewood, USA)
Gonzalez, Zarko (OGX Petrleo e Gs Ltd., Bogota,
Colombia)
Juergens, Chris (Petrominerales Ltd., Calgary, Canada)
Lane, Andrew (Nuvista Energy Ltd., Calgary, Canada)
McConnell, Douglas (DMT Geosciences Ltd., Calgary,
Canada)
Minehardt, Todd (Enthought, Inc., Austin, USA)
Reyes, Felix (Paradigm Geophysical, Houston, USA)
Stephenson, Paul (Petrominerales Ltd., Calgary, Canada)
Tiwari, Dhananjay (TGS, Houston, USA)
Udofia, Udeme (Schlumberger Technology Corporation,
Houston, USA)
Uzoh, Echezona (Statoil, Cypress, USA)
For reinstate to Active membership
Al-Khelaiwi, Adil (BGP-Arabia , Alkhobar, Saudi Arabia)
Brock, Joseph (Agile Seismic, Sugar Land, USA)
El-Desoky, Hany (Shell, New Orleans, USA)
Graber, Stuart (Marathon Oil Company, Cypress, USA)
Kayes, Douglas (KG Consulting LLC, Te Woodlands,
USA)
Wang, Kexie (Jilin University, Changchun, China)
For transfer to Active membership
Bishop, Richard (consultant, Houston, USA)
Bogaards, Mark (Noble Energy Inc., Houston, USA)
Buonora, Marco Polo Pereira (Petrobras, Rio De Janeiro,
Brazil)
Chmela, William (Sekal, Houston, USA)
Ford, Sean (Lawrence Livermore National Lab, Livermore,
USA)
The Leading Edge
Membership
Requirements for Membership
Active: Eight years professional experience, partly in-
volving exercise of independent judgment.
Membership applications and details on other types
of membership, including Associate, Student, and Cor-
porate, may be obtained at http://membership.seg.org.
Freeman, Chuyler (Bureau of Ocean Energy Management
(BOEM), New Orleans, USA)
Hinton, Douglas (Marathon Oil Company, Houston,
USA)
Jonke, Katarina (CGGVeritas, Houston, USA)
Nourollah, Hadi (3D-Geo Pty. Ltd., Melbourne,
Australia)
Reveron, Jorge (PDVSA Intevep, Los Teques, Miranda,
Venezuela)
Roxis, Nikolaos (retired, Korinthos, Greece)
Sauve, Jefrey (consultant, Anchorage, USA)
Sparkman, Deane (Devon Energy, Norman, USA)
Terlikoski, Louis (Anadarko Petroleum Corporation, Te
Woodlands, USA)
Torry, Bradley (Arcis Seismic Solutions, Calgary, Canada)
Travassos, Jandyr (Wavefront Consulting, Rio de Janeiro,
Brazil)
For reinstate and transfer to Active Membership
Edivri, Edafe (Shell Petroleum Development Company,
Part Harcourt, Nigeria)
Lopez, Crucelis (ExxonMobil, Houston, USA)
Sharma, Ritesh (Arcis Seismic Solutions, Calgary, Canada)
Dr. James Irving
Sr. Lecturer at University of Lausanne
Member Since 1998
I am consistently impressed
with the steps that the SEG has taken,
and continues to take, in its evolution
towards being the worldwide home
of applied geophysics, as opposed
to a society narrowly focused on
petroleum and mining industry
applications. As recent President of
the Near Surface Geophysics Section
(NSGS), and member of SEG for over
10 years, I have seen support for our
section grow tremendously, along with
a marked increase in the amount of
SEG research focused on near-surface
hydrological, environmental, and
engineering problems. This can only
be expected to continue to grow as
issues involving clean water and the
effects of climate change become
increasingly important.