special topic
first break volume 30, October 2012
Data Processing
Analysis of a broadband processing technology
applicable to conventional streamer data
Zhengzheng Zhou,1* Milos Cvetkovic,1 Bing Xu1 and Philip Fontana2 provide a case study to
show that processing techniques can realize broadband marine seismic results from conventional streamer acquisition data without the need for the emerging built-for-purpose broadband streamer technology.
T
he vast majority of marine seismic data is acquired
with conventional towed streamer cables which are
equipped only with hydrophones and are all typically towed at the same constant depth for any single
survey. The receiver ghost zeros out the spectral response of
the recorded data at notch frequencies equal to any integer
divided by the ghost delay time, limiting both the top end of
the usable spectrum as well as attenuating the low frequency
response. The source ghost has a similar effect. The combined
effect of the ghosts results in greatly reduced resolution in the
subsurface image. A conventional ‘solution’ for this problem
is to tow the sources and streamers shallow (at less than 7 m
of depth below the surface) to obtain higher frequencies at
the expense of attenuating low frequencies and exposing the
sensors to a noisier environment near the sea surface.
Over the past few years, several distinct technological
advances have emerged that make it possible to mitigate the
effects of the ghosts and obtain broadband images with greatly
enhanced resolution. These methods combine two or more
distinct measurements to remove the ghosts. The combinations
can be either hydrophone plus geophone (Carlson, 2007),
or hydrophones on streamers towed at different depths
(Posthumus, 1993), or hydrophones at offset-variable depths
(Soubaras, 2010). In each case, the de-ghosting process is not a
simple sum of different measurements in x-t domain, but rather
it often requires sophisticated processing flows to accomplish.
These methods require non-conventional data acquisition,
either with streamers equipped with multi-component sensors,
or with streamers configured in unconventional geometries such
as over-and-under or slanted configurations. Such methods are
not applicable to conventional ‘flat’ streamer data. Therefore,
it is beneficial to have a purely processing-based broadband
technique that can recover the high frequency data beyond the
ghost notches and enhance the low frequency signal attenuated
by the ghosts, as such a processing method can greatly extend
the resolution of conventional marine seismic data.
We have developed an effective broadband processing
method based on a new de-ghosting technique. It can remove
most of the ghost effects from conventional streamer data.
In this paper, we will refer to this method as WiBand. It is
designed to address both the amplitude attenuation and the
phase distortion introduced by the ghosts to obtain nearly
flat spectral response in the typical range of 4 Hz to 150 Hz,
as well as a compact, well focused seismic wavelet. In order
to validate our method, we carried out an experiment in
which multiple streamers were towed at different depths and
evaluated the phase reconstruction fidelity of the algorithm
by comparing the WiBand result from the deep tow data with
the standard processing result from the shallow tow data.
It should be noted that there are many other effects that
limit the resolution of seismic images such as the absorption
and scattering of high frequencies by the earth, the difficulty
of removing high frequency multiples, and the difficulty of
obtaining a velocity model good enough to focus high
frequencies under complex overburden. We will not address
these issues here.
De-ghosting methods
Up-going seismic waves are reflected downward by the sea
surface with near -1 reflection coefficient. At any given depth
below the surface, the up and down going waves interfere
destructively at certain frequencies producing notches in the
power spectrum of the data recorded at that depth. There
have been many attempts at solving this problem (Robertsson,
2002). The standard operator for removing the receiver ghost
is an x-y-t domain pseudo-differential operator:
,
This operator is to be applied to shot records. Here, r is the
surface reflection coefficient which is close to -1; c is the near
surface sound velocity and is close to 1500m/s; z is the receiver depth. This operator is typically applied in the f-k or f-p
domains. However, stable application of this operator is very
difficult due to its near singularity, the variability of r and z,
ION GX Technology.
Polarcus.
*
Corresponding author, E-mail: Zheng.Zhou@iongeo.com
1
2
© 2012 EAGE www.firstbreak.org
77
special topic
first break volume 30, October 2012
Data Processing
and the difficulty of transforming streamer data into f-k or
f-p domain without end point artifacts. For instance, when
the streamer depth varies up to +/-1m, this operator is no
longer stationary along the cable and additional corrections
are required. Furthermore, because the operator is non-linear
in the key parameters r and z, obtaining these parameters
directly from the input data requires effective non-linear
optimization methods and carefully selected objective functions (Baan, 2008; Mo, 2009). Nevertheless, the advantage is
that the operator is described by a very small set of variables
that makes such non-linear problems solvable. Typically,
only a 2D version of this operator can be applied.
Another line of attack on the de-ghosting problem
uses variations of traditional blind deconvolution methods
(Zhang, 2011). In order to overcome the mixed phase nature
of ghosted data, the L2-norm objective function of standard
Weiner-Robinson deconvolution must be replaced with nonquadrilinear measures. A subtle difficulty with this approach
is that the convolutional operators required are fairly long
when expressed in the time domain, so a direct time domain
approach to solve the non-linear optimization problem
faces the issue of a large number of degrees of freedom.
This approach also requires special techniques to extend its
application to earth models that are not one-dimensional.
The WiBand workflow combines the strengths of these
two types of de-ghosting approaches, and data-adaptively
derives the most stable operator to remove the effects of the
source and receiver ghosts from the data prior to migration.
As such, it is not a data creation method, but a method to
recover the signal weakened by the ghosts and relies on the
presence of usable signal at and near the ghost frequencies
in the raw data.
Figures 1(a) and (b) compare a standard processing
result (a) with a WiBand result (b) from an offshore Gabon
dataset acquired in 2011 using a streamer towed at 15 m
below sea surface. The receiver ghost notch associated with
this depth is about 50 Hz and in the standard processing
flow a high-cut filter was applied to remove energy above
55 Hz from the data. No high-cut filtering was applied in the
WiBand processing flow. It is apparent from this comparison
that the WiBand flow has managed to greatly enhance the
resolution of the image. Both low and high frequency signals
have been enhanced and many structural details invisible in
the standard product are well resolved in the WiBand image.
Figure 1(a) Standard 4 ms PSTM image.
Figure 1(b) WiBand 2 ms PSTM image.
Deep tow versus shallow tow test
Figure 2 Streamers were towed in a fan pattern. The blue line just above
the dark green line represents streamer 4 (deep tow). The orange line above
streamer 4 represents streamer 5 (shallow tow).
In order to validate the WiBand results, we need to closely
examine the phase of the WiBand output, particularly the
phase of the high frequency part of the signal above the
first receiver ghost notch frequency. One way of performing
this analysis is to compare WiBand output from a deep tow
dataset with a raw shallow tow dataset which has near zerophase signal beyond the ghost notch of the deep tow data.
For this purpose, we acquired a new dataset in 2012
in the same general area offshore Gabon. In this new
experiment, eight streamers equipped with lateral position
controllers were deployed in a fan pattern (Figure 2) and
towed at different depths. Two air gun arrays were towed
at different depths and were fired in alternating order. For
78
www.firstbreak.org © 2012 EAGE
special topic
first break volume 30, October 2012
Data Processing
Figure 3(a) Shallow tow data raw stack.
this paper, we focus on the data associated with the 5 m
source and recorded by the two central streamers, which
are streamer 4, towed at 12 m depth, and streamer 5, towed
at 7 m depth. We refer to streamer 4 data as the deep tow
line and streamer 5 data as the shallow tow line.
As can be observed from Figure 2, streamer 5 has a much
larger feathering angle relative to the sailing direction than
streamer 4. This feathering results in considerable mid-point
dispersal when we bin the data according to nominal 2D
geometry. Therefore, for the purpose of comparing the two
lines, we focus on the first second or so of data below the
water bottom. For the shallow tow line, the extent of midpoint dispersal within a single CDP bin is about 200 m over
the range of offsets contributing to the stack at the water
bottom level. For the deep tow line, the relevant dispersal
extent is 50 m. For the same CDP bin, the average midpoint
locations for the two lines are about 200 m apart.
In Figures 3(a) and (b), we show the raw stacks from
the shallow tow line (a) and the deep tow line (b). The two
sections display dramatically different wavelet characteristics.
This difference stems from the different ghost delays in the two
datasets. Notice the dramatic difference in the seismic wavelet
in these two sections. Figure 4 shows the effect of a ghost on
the amplitude and phase of recorded data. The ghost introduces
notch frequencies at which the amplitude is greatly attenuated
due to destructive interference. More importantly, across each
of these notch frequencies, the phase response undergoes a rapid
1800 degree rotation, putting the data beyond the first notch in
opposite polarity to the main band of data below that notch.
In our case, the deep tow data has a notch and a polarity
flip at 63 Hz which is very close to the peak frequency of
the shallow tow data (see Figure 6), leading to the dramatic
difference in the appearance of the two sections. It is therefore
© 2012 EAGE www.firstbreak.org
Figure 3(b) Deep tow data raw stack.
Figure 4 Phase (blue) and amplitude (red) spectra of the operator representing a ghost at 10 ms delay with a -0.95 surface reflection coeffecient.
very instructive to use the shallow tow data as a standard to
analyze the phase characteristics of the WiBand result from
the deep tow data.
Analysis of results
We first compare the WiBand results from both datasets. It
is apparent from Figures 5(a) and (b) that, after WiBand,
the two datasets are no longer dramatically different in
their wavelet characteristics. The spectral plots in Figure 6
shows that the WiBand process has largely filled in the ghost
notches in the power spectra, and the resultant power spectra
from both the deep and the shallow tow data have similar
spectral slope over a very broad frequency range. Because the
WiBand process is applied independently to these two lines,
the common spectral shape post-WiBand suggests that the
de-ghosting process worked in a consistent fashion.
To analyze the phase fidelity of the WiBand result, we
will use the 5–105 Hz band of the raw shallow tow data as
79
special topic
first break volume 30, October 2012
Data Processing
the standard, as we know from Figure 4 that the shallow
data has essentially linear phase within this band modulo
phase of the source wavelet, which is common to both the
Figure 5(a) WiBand result from the shallow tow line.
shallow and the deep tow lines. So if the WiBand result from
the deep tow line matches in phase the raw shallow tow data
in this band, we can conclude that the WiBand process has
correctly reversed the deep tow data’s polarity flip at 63 Hz.
In order to use the shallow tow data as the standard, we
first apply a zero-phase high-cut filter to it to remove the
phased-distorted part of its signal beyond its notch frequency
of 110 Hz. The power spectra of the shallow tow data before
and after this filter are plotted as the blue and red curves in
Figure 7. To focus on the phase characteristics of the deep tow
WiBand results when comparing it to the filtered shallow tow
data, we apply a zero-phase shaping filter to match its power
spectrum to that of the filtered shallow tow data. The spectra
of the deep tow WiBand data before and after this shapefiltering are plotted as the green and purple curves in Figure 7.
We display the high-cut filtered shallow tow data in
Figure 8, and the shape-filtered deep tow WiBand data in
Figure 9. By comparing Figures 8 and 9, we determine that
the two images match well in phase. That we do not obtain
perfect match can be attributed to the disparities in the
feathering geometries of the two lines discussed earlier.
In Figure 10, we plot two traces, one from each line, from
the marked locations in Figures 8 and 9. The traces have the
afore-mentioned filters applied. A very good match is observed
between the two traces. The peaks and troughs of the major
reflection events line up very well. In particular, the phases of
the main events match remarkably well considering distance
between the two lines (indeed, some of the minor mismatches
can be attributed to diffractions which occur at different
locations in these two lines.) For the deep tow WiBand trace
displayed here, the recovered signal around the ghost notch at
63 Hz contributes very significantly to its waveform. Therefore,
a good match to the shallow tow data cannot be obtained unless
the phase of the recovered signal near and above the notch frequency has been reconstructed correctly by the WiBand process.
Signal-to-noise ratio of raw data
Figure 5(b) WiBand result from the deep tow line.
In order for any de-ghosting process to work properly and
recover useful signal at and beyond the ghost notch, the raw
Figure 6 Spectra of the deep tow and shallow tow datasets before and after
WiBand processing.
Figure 7 Spectra of the shallow tow data, raw in blue and after zero-phase
high-cut filtering in red, and spectra of the deep tow WiBand data in green,
and after zero-phase shaper-filtering in purple.
80
www.firstbreak.org © 2012 EAGE
special topic
first break volume 30, October 2012
Data Processing
Figure 11 Signal and noise spectra of raw single channel sections.
Figure 8 Raw shallow tow data with only a zero-phase high-cut filter to
remove energy above its receiver notch frequency of 110 Hz.
Figure 9 WiBand result from the deep tow data with zero-phase shaping filter
to match its power spectrum to the data in Figure 8.
spectra are measured from a 600 ms data window below
direct arrivals and above water bottom reflections. The signal
spectra are taken from a 600 ms data window that starts
600 ms below the water bottom reflection. The source volume
is 2950 cu in. The noise above 20 Hz is well below the signal
strength, even for ghost notch frequencies. This supports our
view that the raw data inherently supports de-ghosting.
Also worth noting is that the higher-frequency noise
above 20 Hz is dominated by streamer-born modes and
towing depth has little impact on them, whereas the tow
depth does strongly influence the strength of the swell noise
the receivers are exposed to. To avoid the swell noise, one
would wish to tow the streamers very deep; however, too
great a towing depth would mean that the first-order receiver
ghost would be pushed lower and closer to the dominant frequency of the streamer-born noise, which is not reduced by
the deeper towing, reducing the signal-to-noise ratio at the
notch frequency and limiting the ability of processing-based
solutions to recover that data near the notch frequency.
This analysis suggests that one should measure the relative
strengths of the swell noise and the streamer-borne noises if
possible and design the survey parameters to optimize the
signal-to-noise ratio over the whole target frequency range.
Conclusion and discussion
Figure 10 Central traces from sections in Figures 8 (red) and 9 (blue).
data must have useful signal at and around the notch frequency. It is our view that modern marine streamer acquisition equipment has advanced sufficiently so that the various
modes of streamer-borne noises are much lower than even
the very weak signal present at the notch frequencies.
In Figure 11, we plot signal and noise spectra of the data
recorded by the first channels from the shallow tow cable at
7 m depth and the deep tow cable at 12m depth. The noise
© 2012 EAGE www.firstbreak.org
We have successfully recovered broadband signal from deep
tow streamer data and obtained high-resolution images. For the
deep tow data, the prominent receiver ghost notch at 63 Hz is
filled. An essentially flat power spectrum is recovered between
4 Hz and 150 Hz. We have validated the result by comparing it
to shallow tow data which is known to have a near-linear-phase
wavelet below 110 Hz. The comparison shows that the phase
of the recovered broadband data matches that of the shallow
tow data indicating that the strong phase distortion at 63 Hz
originally present in the deep tow data has been corrected.
We believe that a processing-based broadband solution
can provide an effective alternative to the solutions based
on acquisition and processing that require non-conventional
streamer acquisition. A processing-based solution can be
applied to existing conventional streamer data to extract
the maximum value out from the data. Much of our current
81
special topic
first break volume 30, October 2012
Data Processing
datasets have been processed through workflows that start
with a resampling to 3 or 4 ms and contain high-cut filters
that remove most of the energy above the first order receiver
ghost notch frequency. We have demonstrated that there is a
great amount of information present in the raw data that is
lost through such standard processing flows, and that such
information can be retained and utilized through a new
type of workflow that carefully preserves all the signal and
properly compensates for the effects of the ghosts.
Our method can also be used in the context of new
acquisition and remove the constraints on receiver depth
posed by the requirements for high frequency signal. Since
the ghost notch can be effectively filled, a new acquisition
no longer has to have the receivers shallow in order to
record frequencies above approximately 90 Hz. Rather, the
acquisition parameters can now be chosen to both improve
operational efficiency and optimize signal-to-noise ratio over
the whole desired frequency band by considering factors
such as swell noise and avoiding positioning the ghost notch
at a dominant frequency of streamer-borne noise.
data in Ophir’s block offshore Gabon. We would also like
to thank the Polarcus acquisition crew and management for
making this experiment possible.
References
Baan, M. and Pham, D. [2008] Robust wavelet estimation and
blind deconvolution of noisy surface seismics: Geophysics, 73(5)
V37–V46.
Carlson, D. [2007] Increased resolution of seismic data from a dual
sensor streamer cable. 76th SEG Annual InternationalMeeting,
Expanded Abstracts.
Fontana, P., Cheriyan, T. and Saxton, L. [2011] Why not narrowband.
81st SEG Annual International Meeting, Expanded Abstracts.
Posthumus, B. J. [1993] Deghosting using a twin streamer configuration.
Geophysical Prospecting, 41, 267–286.
Robertsson, J. and Kragh, E. 2002, Rough-sea deghosting using a single
streamer and a pressure gradient approximation: Geophysics, 67(6),
2005–2011.
Soubaras, R. [2010] Deghosting by joint deconvolution of a migration
and a mirror migration. 80th SEG Annual International Meeting,
Expanded Abstracts, 3406–3410.
Acknowledgements
Zhang, Y. and J. Claerbout, 2011, A new bidirectional deconvolution
We would like to thank Ophir Energy and the government
of Gabon for allowing the acquisition of the experimental
Don’t forget,
it’s time to...
method that overcomes the minimum phase assumption: Stanford
Exploration Project Report No. 142.
Renew your EAGE
membership for 2013
Pay before 31 December 2012 and continue receiving the following benefits:
45 (print)
1569-44 04 (online)
issn
1873-06
issn
FIRST
UniQ System Sales
BREAK
NeAr
© 2012 Schlumberger.
First Break
Geoscience
and Engin
eering
10 ·
9 – SEPTEMBER
VoluMe
2012
NuMBer
4 · AuGuST
es: what
do we really
.org
2012
08-05-12
Microseism
ic bootcamp
Wet suits
on the way
for Arab
Plate Workshop
!
11:27
understan
d?
12-07-12
15-08-12
1-3
ware.indd
r.indd
Sensor&Soft
2_aug_cove
A037
CC00957-M
14539-NSG1
15:25
ysics
eoph
aceg
.nearsurf
www
02-07-12
Geological 3D
models
ional computer
of three-dimens
MethThe problems
elds are discussed.
oil and gas fi
to model
of geological
approaches
technological
are deodological and
of modelling
different stages
maturity,
creation in the
the exploration
dealgeointo account
excellent books but
scribed taking
geological and
There are many
Geophysics
of the initial
i.e. aimed
ing with Environmental
amount and quality
INWARD facing, of Geothey are all
number of at Geophysicists and studentsto stakephysical data.
with a large
and not accessibleof Environmodels are explained numerous hydrocarbon physics
purchasers a serious
holders and
for
This is
of building the
mental Geophysics.
building the models
Different aspects
much Environmental
The results of
problem becausecommissioned by Civil
is
illustrative material. software systems are presented.
Nuclear
Geophysics
Water Engineers,and lack
fields in the different
amongst others lead to ill
involved in build- Engineers,
can
of the companies
and Engineers
not (often
of clear communication
to professionals
which may
for the students
be interesting
specified projects
fields and also
The book will
and oil and gas
ned and inappropriatelyinformation.
anddefigeology.
models of oil
in petroleum don’t!) deliver the desired
need to understand
ing geological
who
specialization
a
with
book for people they regularly encounpre- solve
He facing
an OUTWARD
the problems
high school graduates
solution, proved
ThisatisTNK-BP.
can
because it in optimal geotechnical investigation. It
ModellingGeophysics
PhD their applied, intrusive geophysical
of Geological
hastoadeliver
Hehelp
be
and
ter and
is the Manager
at LUKOIL.
but last to
use to students kindly provided
by the essential,
of theofI.M.
significant
Konstantin Zakrevsky Geological Modelling Centre
faculty
case studies
course, be
a wealth of
in the
of the geologicalmay, of inasthe
it contains
prob- collaborators.
viously worked
practitioners
is a graduate
and
engaged
a great
mineralogy and
Konstantin isby friends, colleagues
do not require but do
documents
geology and
of Oil and Gas.
Russian state whom this book is addressed
State University
basis of geophysics
can
the authors of The people toControl”.
Gubkin Russian
what information and
and is one of
Quality of the mathematical
principles,
Models:
modelling
understanding
physical
“3D
interpretation
the
of
book
to
lems of geological
and boundariesin an integrated way
author of the
need to understand
the limitations
fields and is the
be gleaned, we can combine techniques which one technique
on modelling
how
critical problems
A new issue of First Break every
month. Discount on additional
EAGE journals.
13840- cover
zakrevskiy
21x28.indd
Modellingy
Attractive bookshop offers
K.E. Zakrevsk
sics
ntal Geoedphy
(needed!)
Environme
you ever want
d to ask!
Everything
were afrai
to know but
educat
EAGE members get attractive discount
on books from the EAGE (online) Bookshop.
Visit www.eage.org/bookshop for more
information.
series
ion tour
7
especially
complex and
solve subtle,
solve.
alone cannot
14:19
09:44
1
Geophysics
Reservoir
30 – ISSUE
take the lead
m/uniqsales
September.indd
Geco.indd
1
1-3
2012
Environmental
UniQ – Time to
www.slb.co
14744-Cover
CC02994-MA016
FirstWestern
Break 9
30 – Issue
9 – September
VOLUME
The industry-leadin
from WesternGeco g UniQ* integrated
is now available point-receiver
and energy
land seismic
companies.
for lease
system
or purchase
UniQ provides
to both service
in the industry. the most efficient
Recent awards execution of high-density
■ 200,000
and achievements
onshore surveys
live-channel
include
capacity of
■ Over
continuous
20,000 vibrator
point-receiver
points per
■ Reliability
day with high-productivit acquisition
and in-built
redundancy
y source techniques
Included
for over 98%
with the
average system
data management,system hardware
uptime
is
Flexible support, and point-receiver integrated software
for survey
noise
training and
planning,
maintenance reduction and signal
enhancement.
options are
available.
Volume
2012
sics
4 · August
Geophy
10 · Number
Marine
on Applied
Issue
nearSpecial
of a shallow
d
ation
Norway te,
Forewor
y investigfjord, Vannes Haflidason
263
ciplinare, Finneid ux, M.
g, H.
Multidis
Forsber
L’Heure
267 shore landslid
C. F.
J.-S.
Vardy,, A. Steiner,
shallow
M.E.
en
marine
h to
O. Longva
and
J. Brendry
and
ry approac
A. Gaillot
sciplina ent
N. Sultan,
A multi-di assessm ,
near279 geohazard t, Y. Thomas
3D seismic :
B. Marsse
studiesFault, Sea
olution
n
Y. Stephan
ir-scale
of high-res
Anatolia
to reservo
ution
Grall,
North
ok, C. Saritas
Contrib imaging
the active
291 seafloor
Westbroand H.
ion to
G.K.
applicat a Marsset, A. Rochat
of Marmar , B.
seabed
G. Çifçi,
n of
Thomas
Y.
P. Henry,
erisatio Sea
L. Géli,
charact North
ce
central Diaz-Do
tomated
on the
in the and D.
Semi-au
incision and
303 pockmarks, D. Long
gully Channel
nce of
hetland
J. Gafeira
cture
and significa
Faroe–S
of the■ r Technical
infrastru
The timing
Special
317 eastern flank
Articles
on seafloo
Topic
D. Long
and Orthogona
the middle
its impact
from
Stewart
lSea
wide azimuth
H.A.
surface North
lded The
surveys:
Nelson
Central
acquisition
full field
of the
An ice-mou
ene
■
model:
and imaging
Technolo
333 Pleistoc
using iterative
editors
gy Feature
F. Buckley
quantitativ
the guest
ofTechnical
e improvem
presentati
hies
ents
ons in geoscienc
Biograp ■ EAGE
347
News
Volume
Modelling
Schlumberger.
iCS
*Mark of
GeoPhyS
12-se-0015
SurFACe
Now avail
lease or able for
purchase
Geological 3D
ce
Surfa
Near hysics
Geop
1
Discount on events
Benefit from discount on
all EAGE events worldwide.
EAGE’s Annual conference & exhibition will take place
in London (UK), from 10-13 June 2013.
and
practiced environmental
postgradutaught and
Styles has
to undergraduates,
Professor Peter
for 35 years nationally and internationally.
geophysics
engineering
training courses,
Geophysics
ates, and professional
by a PhD in
posts at Swanin Physics followed
from Oxford
held academic
Research
He graduated
Tyne. He has
as Professorial published
University
from Newcastle-uponHe has
and now Keele
Geophysics.
all aspects
sea, Liverpool and Environmental
reports on
Society,
professional
Fellow in Applied
and many
of the Geological Miner90 papers
He is a Fellow
more than
of Materials, ning
Geophysics.
the Institute
defi
of Environmental
Society and
committee
a memthe UK government
the Royal Astronomical
waste. He was Nuclear
He chaired
radioactive
of
als and Mining.
disposal of
on Non-Proliferation
geological
Nuclear
criteria for
Committee
Panel of the report
Royal Society
DECC
Characterization
ber of the
the Geosphere was author of the recent
He
Weapons and
hydrofracturing.
Gas
Authority.
with Shale
Decommissioning
associated
Seismicity
delivered by
on Induced
a one-day courseAccompanied
(EET) offers
Education Tour various locations globally. and others
The EAGE’s
at
members
it provides
renowned geoscientists
course book, key topics in the Geosciences.
by a comprehensive developments in its launch in 2006.
latest
success since
access to the
been a great
The Tour has
14347-Cover
Boek EET
VII.indd
14-02-2011
14:35:19
Peter Styles
30-07-12
08:57
1-3
Access to EarthDoc
Unlimited access to EAGE’s online
geoscience database with around
45,000 event papers and scientific
articles.
Renew online via www.eage.org!
Step 1
Step 2
Step 3
Go to www.eage.org - Login to My Account
Log in with your membership number (or login name) and password
Click on My EAGE, then select ‘My Services - Renew Membership’
More benefits?
Visit our website: www.eage.org
EAGE online services: easy, fast and secure
MEM13-V2H.indd 1
82
03-09-12 16:44
www.firstbreak.org © 2012 EAGE