Content uploaded by Sofia Davydycheva
Author content
All content in this area was uploaded by Sofia Davydycheva on Mar 28, 2016
Content may be subject to copyright.
435 The Leading Edge April 2012
SPECIAL SECTION: M arine and seabed technology
To CSEM or not to CSEM? Feasibility of 3D marine CSEM
for detecting small targets
The marine controlled-source electromagnetic (CSEM)
method is an important tool for offshore exploration. It
allows recovery of subsurface resistivity, a key hydrocarbon
(HC) indicator, using modeling- and inversion-based
interpretation of the EM data acquired on the seafloor.
Application of CSEM can reduce risk and optimize drilling
operations, as well as improve estimates of HC reserves.
e CSEM workflow consists of the following main steps:
feasibility study, survey design, data acquisition, processing,
and interpretation. Feasibility study and data interpretation
rely heavily on 3D EM modeling to properly select and plan
surveys and recover the 3D subsurface resistivity image.
Detection of deep small targets immersed in anisotropic
environments is a challenging task for the current CSEM
technology, making it essential to employ the best practices
in performing reliable feasibility studies. To manage this task,
a skilled modeler needs to take into account all available
geological and geophysical data to build a 3D Earth model
and utilize accurate 3D EM modeling software. e fast and
reliable 3D EM modeling that can accurately handle arbitrary
anisotropic resistive media with complex structural interfaces
and HC reservoirs would make a crucial contribution to the
success of CSEM technology.
To illustrate the importance of modeling in the marine
CSEM worklflow, we evaluate the feasibility of CSEM in a
challenging case represented by small hydrocarbon (resistive)
targets in the shallow and deep-water environments using both
synthetic and field case studies. We apply our accurate 3D
finite-difference modeling code developed to solve Maxwell’s
equations in the frequency domain for arbitrary anisotropic
subsurface media. e finite-difference scheme is solved using
the spectral Lanczos decomposition method and with appli-
cation of the optimal grid theory. is enables us to perform
calculations of multi-offset and wide-band multifrequency re-
sponses simultaneously and fast, at the cost of computing a
single-frequency and single-offset response. e code can be
used for any advanced application of 3D electromagnetics in a
borehole, on land, or in a marine environment.
e goal of this paper is to show why 3D EM modeling-
based feasibility studies should be performed prior to CSEM
surveys to determine the ability of CSEM technology to detect
and assess relatively small offshore reservoirs, thereby better
preparing geoscientists to answer the following fundamental
question: Given everything known about a target and the back-
ground, should a marine CSEM survey be performed?
e role of 3 D E M mo deling: “ To C SEM or not to CSE M? ”
Feasibility study, survey design, data acquisition, processing,
and interpretation are the five main steps of the marine CSEM
workflow. Each step of this workflow, especially feasibility
MICHAEL A. FRENKEL, SingularEM, LLC
SOFIA DAVYDYCHEVA, 3DEMSoft, LLC
study, survey design, and data interpretation, relies on 3D EM
modeling to properly select targets that are feasible for CSEM
surveys, optimize their data acquisition plans, and recover the
3D subsurface resistivity volumes.
Feasibility of 3D marine CSEM for detecting large HC
reservoirs/targets in absence of rough bathymetry and near-
surface resistors is not normally a challenging problem and of-
ten solvable using 2D or even 1D approximations. As offshore
exploration continues to expand into increasingly geologically
complex fields, geographically challenging locations, and/or
dealing with smaller targets, fast and reliable 3D EM modeling
that can accurately handle arbitrary anisotropic resistive media
with complex structural interfaces and HC reservoirs becomes
one of the most important core elements of the CSEM tech-
nology.
Several major oil companies, such as Chevron, ExxonMo-
bil, Shell, and Statoil, have managed to adopt advanced 3D
EM modeling tools and are among the industry leaders in ac-
tive use of CSEM in the E&P workflow. More than a half of
all marine CSEM surveys have been ordered by oil companies
armed with the full spectrum of in-house 3D EM modeling-
based survey planning and data interpretation tools. Although
the ability of CSEM to reduce the risk of exploration drilling
is undeniable, the relatively slow adoption and expansion of
CSEM technology is primarily related to the fact that the vast
majority of oil companies do not have in-house CSEM ex-
pertise and robust 3D EM modeling and interpretation tools.
We believe this situation will improve with time, and
more companies will start using marine CSEM routinely, just
as they use EM logging tools in each exploratory well. Mean-
while, those operators with limited in-house CSEM expertise
but with plans to acquire CSEM data to derisk their expensive
offshore drilling operations need to be actively involved at each
step and decision point of the CSEM workflow. Geoscientists
experienced in building 3D models, performing modeling,
and interpretation know that almost each field case study has
unique features that require special case-driven solutions as
well as the process in place for independent quality control.
It is especially important because interpretation of feasibil-
ity study modeling results (“To CSEM or not to CSEM?”) is
often based on the predetermined ambiguous criterion, e.g.,
it can be a threshold for the magnitude of the horizontal elec-
trical component Ex normalized by the background/reference
point signal (normalized magnitude ratio). In a 3D case, this
threshold is frequently set to 1.10 or to a slightly higher level
depending on the risk tolerance of an oil company and its
understanding of CSEM technology capabilities. However,
the majority of real 3D cases have one or a combination of
difficult-to-accurately-measure or model effects, such as rough
seafloor bathymetry, airwave, complex subsurface geology,
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
436 The Leading Edge April 2012
Marine and seabed technology
patchy near-surface resistors (e.g., gas hydrates), the overbur-
den anisotropy (normally, it is unknown prior to the survey),
relatively small deep HC reservoirs, etc., suggesting that the
1.10 normalized magnitude ratio threshold is way too low to
claim that a 3D CSEM survey is feasible and capable of add-
ing value to the HC exploration derisking process.
erefore, the availability of fast and accurate 3D CSEM
modeling becomes the game-changing factor by enabling the
testing of multiple possible scenarios and making accurate and
timely decisions regarding the target-specific need and added
value of a CSEM survey. In the next sections, we will present
our 3D EM modeling technology and show its application
to assess feasibility of 3D marine CSEM for detecting small
reservoirs in synthetic and field settings.
3D EM modeling technology
To solve Maxwell’s equations, we apply the accurate and fast
3D finite-difference (FD) modeling code (Davydycheva and
Druskin, 1999). e problem with respect to the EM field ex-
cited by a grounded electric or magnetic dipole is discretized
on an FD grid and solved iteratively using the multifrequency
spectral Lanczos decomposition method (SLDM–Druskin
and Knizhnerman, 1994). An FD scheme can be solved using
different iterative solvers; however, the SLDM solver allows
fast simultaneous computation of wide-band multifrequency
responses at the computing cost of a single-frequency run.
We use the Lebedev staggered FD grid approach that en-
ables efficient treatment of the arbitrary dipping anisotropy
of the electric conductivity. Such a grid gives the ability to
determine different components of the electric field, Ex, Ey,
and Ez, and the electric current density, Jx, Jy, and Jz at the same
spatial nodes. is allows application of the general anisotro-
pic Ohm’s law (connecting all components of E and J at the
same point) and enables efficient handling of dipping anisot-
ropy and dipping medium interfaces which are important for
accurate modeling of arbitrary seafloor bathymetry and com-
plex subsurface structures (e.g., Davydycheva and Rykhlinski,
2011) showed 3D modeling of dipping medium interfaces
and bathymetry effects).
e accuracy of an FD method is known to depend on
gridding. Our 3D modeling uses an equivalent medium ap-
proach, where the material averaging within inhomogeneous
grid boxes allows constructing the grid independently of the
medium model and handling thin high-contrast structures us-
ing moderately coarse grids. e optimal grid theory, which
allows constructing the same grid for a wide variety of medi-
um models and for wide spacing and frequency ranges, is also
applied in the numerical algorithm. Instead of minimizing the
global truncation error, which is typically done in many other
numerical approaches, we minimize the error at the receivers
and optimize the approximation of the boundary conditions
Figure 1. (a) XZ cross sections of 1D and (b) 2D/3D formation models at Y = 0. e bar represents resistivity range on log10 scale, and the model
consists of sea water (black 200-m layer), Earth below the seafloor, and target reservoir (yellow). e white dots display the positions of the X-directed
HED source (leftmost white dot) and a group of equally spaced receivers located on the seafloor. e source is towed 30 m above the seafloor.
Figure 2. Comparison of Ex magnitude versus offset (MvO) curves
for 1D model with and without the target reservoir (dotted and solid
curves, respectively) for frequencies f = 0.0 (DC case), 0.05, 0.25,
and 1 Hz. e EM field is excited by a 1 A × m HED source. e
anomalous reservoir response can be estimated as the difference between
the dotted and solid curves of the same color. e shallow (WD =
200 m) and the deep-water cases are shown on the top and bottom,
respectively. In Figures 2–8, the displayed frequencies are: f = 0.0 (DC
case = black curve), 0.05 (blue), 0.25 (green), and 1.0 Hz (red).
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
April 2012 The Leading Edge 437
Marine and seabed technology
at infinity (Davydycheva et al., 2003).
In our algorithm, the grid is attached to the tool trans-
mitter and receivers, not to the medium model. It is adjusted
to the tool depth of investigation at the minimal excitation
frequency and to the conductivity model through skin depth
in the most conductive part of the medium at the maximal
frequency. e grid is constructed automatically so that a step-
by-step grid refinement provides the convergence to correct
results at the receivers. Typically, it is easy to reach the accuracy
of 1–3% using model-independent gridding, but attempts to
overcome this accuracy level may become impractical for com-
plex models such as thin structures with high contrasts in con-
ductivity. However, 1–3% is normally enough for intended
CSEM applications. A compromise between the model com-
plexity, grid size, and accuracy should be found for each appli-
cation (Davydycheva, 2010). Our code was validated against
quasi-analytical solutions and 3D benchmark modeling (Fren-
kel and Davydycheva, 2009). e results of the 3D modeling
study presented in this paper were obtained on a single 2.2
GHz processor of an ordinary laptop computer with 6 GB
memory. It took less than one minute to calculate the EM
fields for a wide band of frequencies for one transmitter posi-
tion. Due to application of the optimal gridding technique,
the computational time required to perform calculations for
Figure 3. Modeling results for isotropic models in the shallow water (200 m) case. e target depth is 0.6 km. 2D and 3D cases are shown in
the top and bottom rows, respectively. Here, and in all the figures below, the columns from the left to right display Tx-Rx offsets from 1 to 8 km,
respectively. In Figures 3–8, the acquisition line is Y = 0 and the (X, Y) target center coordinates are X = Y = 0.
Figure 4. Modeling results for isotropic models in the deep-water case. e target depth is 0.6 km. 2D and 3D cases are shown in the top and
bottom rows, respectively.
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
April 2012 The Leading Edge 439
Marine and seabed technology
much larger resistivity models does not increase rapidly.
Wavelength and skin-depth in marine CSEM
Spatial resolution of CSEM measurements is primarily deter-
mined by the EM field wavelength (h) or by the source-receiver
(Tx−Rx) offset, whichever is shorter. In the CSEM low-frequency
range (when the displacement currents in the earth and water
can be neglected), h is related to the plane-wave skin-depth, b:
h = 2π × b, b = (l/(μ0πf))1/2, where l, f, and μ0 are the resistivity,
excitation frequency, and magnetic permeability of the Earth,
respectively. Table 1 shows b and h values for frequencies from
0.05 to 4.0 Hz and for a constant formation resistivity l = 1 Ω
× m.
e spatial resolution of CSEM for short and moderate
source-receiver offsets (< 6 km) at the fundamental CSEM fre-
quency f = 0.25 Hz or lower is determined and restricted by the
offset only. e higher frequencies of f >1 Hz allow somewhat
higher spatial resolution. However, the skin-depth, b, indicates
that their depth of investigation is restricted to a few hundred
meters.
Synthetic case study settings
e main purpose of this modeling study is to evaluate the
frequency-domain CSEM responses in shallow and deep-
water settings generated by relatively small resistive reservoirs
embedded in isotropic and anisotropic formations. We pres-
Figure 5. Modeling results for 3D isotropic models; the target depth is 1 km, and the shallow- (200 m) and deep-water cases are shown in the
top and bottom rows, respectively.
Figure 6. Modeling results for 2D anisotropic models; the target depth is 1 km, and the shallow- (200 m) and deep-water cases are shown in the
top and bottom rows, respectively.
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
440 The Leading Edge April 2012
Marine and seabed technology
ent simulation results and analysis for the most representative
cases.
Small 3D target benchmark model. e EM field is excited
by the horizontal electric dipole (HED) source towed at a fixed
30-m distance from the seafloor. We simulate the shallow-wa-
ter case with the water depth (WD) of 200 m and the deep
water case (WD > 2 km). e results of modeling allow the
determination of the optimal sets of frequency-offsets for ob-
taining the most pronounced reservoir responses.
Figure 1 shows the XZ cross section of the one-dimensional
(1D) and three-dimensional (3D) models used in this study.
e 1D model (Figure 1a) represents an infinite reservoir prop-
agated in the X and Y directions. e 3D reservoir (Figure 1b)
has dimensions (X × Y × Z) = (2 × 2 × 0.2) km, while in the 2D
case, the model properties are Y-invariant. In practice, however,
we model the 2D case by setting the reservoir length along the
Y direction to a large number (50 km). is enables us to use
the 3D program to model 2D cases (which some also refer to
as 2.5D modeling settings: 2D model excited by an arbitrary
3D source). e transverse resistance of the target reservoir or
a product of the reservoir thickness and its resistivity is 104 Ω ×
m2. It means that, for example, if the thickness of the reservoir
in the Z direction is 200 m, its resistivity is 50 Ω × m.
e considered target depths are 0.6, 1.0, and 2.0 km
Figure 7. Modeling results for 3D anisotropic models; the target depth is 1 km, and the shallow- (200 m) and deep-water cases are shown in the
top and bottom rows, respectively.
Table 1. CSEM skin depths and wavelengths given l = 1 Ω × m.
f (Hz) b (km) h (km)
0.05 2.25 14.14
0.25 1.01 6.33
1.00 0.50 3.16
2.00 0.36 2.24
4.00 0.25 1.58
below the seafloor (BSF). In the isotropic models, water and
background resistivities are 0.3 and 1.0 Ω × m, respectively. In
the anisotropic cases, we use constant values for the vertical and
horizontal background resistivities, Rv = 2 and Rh = 1 Ω × m,
respectively, so the anisotropy ratio is Rv/Rh = 2; however, the
target reservoir is assumed to be isotropic (Rv = Rh).
1D reservoir responses in shallow and deep water. Before dis-
cussing 2D and 3D cases, we would like to show simple 1D
simulations, which are normally performed as the first step of
feasibility studies. In the shallow-water case (WD = 200 m),
the anomalous reservoir response is weak at high frequencies
(Figure 2, top) because it is dominated by the airwave (the sig-
nal propagating through the air above the water surface), but it
becomes stronger at lower frequencies (f ≤ 0.05 Hz), at which
the airwave effect is weak. In the deep water case (Figure 2, bot-
tom), the anomalous response is stronger at higher frequencies
due to a strong nondecaying partially guided wave propagating
along the reservoir.
Due to the skin effect, the EM signal attenuates at higher
frequencies, and its level at longer offsets becomes closer to the
noise floor, which makes these data difficult to use in interpre-
tation. In this paper, for the electrical in-line component Ex and
the source moment of 1 A × m, we use 10–14 V/m threshold for
the noise floor.
Analysis of synthetic CSEM data
To determine optimal frequencies and offsets, we analyze the
CSEM synthetic data generated for small 2D and 3D resis-
tive targets in both shallow and deep water environments. We
display plots of the absolute and normalized magnitudes of the
horizontal electrical component (Ex) versus the offset: MvO
and NMvO, respectively. e normalization is done using a
measurement performed far from the reservoir. e MvO and
NMvO are plotted for the transmitter-receiver (Tx−Rx) mid-
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
April 2012 The Leading Edge 441
Marine and seabed technology
Figure 8. Modeling results for 2D isotropic models in deep water; the target depth is 2 km.
point positions for an acquisition line Y = 0; the (X, Y) target
center coordinates are X = Y = 0. Typical formats for displaying
modeling results for shallow and deep water cases are shown
in Figures 3–4.
e optimal frequencies are those that produce the larg-
est reservoir response. Because these are affected by the water
depth, we present the MvO and NMvO graphs for both shal-
low and deep-water cases. It is essential to jointly analyze the
MvO and NMvO curves. For example, the NMvO values in
the third and fourth columns of Figure 4 display rather high
normalized reservoir responses at f = 1.0 Hz at 4 and 8 km
offsets. However, in these cases, the absolute signal magnitude
is below the noise floor of 10−14 V/m, which makes this CSEM
data not usable for imaging the subsurface resistivity structure
in practical applications.
Target at 0.6 km below seafloor
In this section, we present the results of modeling for the case
when the target is located relatively shallow at 0.6 km BSF.
We show the MvO and NMvO plots of Ex component for the
shallow and deep water cases and compare 2D and 3D results
for these two respective cases (Figures 3–4).
e modeling data indicate that at 2 km offset 2D and 3D
anomalous effects are strong (NMvO > 1.5), and there is no
significant difference between 2D and 3D NMvO responses
at both high and low frequencies. e eyeball analysis suggests
that in this case the lateral resolution of CSEM does not much
depend on frequency. Indeed, the “width” of the reservoir re-
sponse (approximately estimated by the distance between the
inflection points on the NMvO curve flanks) is roughly the
same for all the frequencies and is determined primarily by the
offset. e reason is that the signal wavelength (h) below the
seafloor is equal or greater than the offset at the considered
range of frequencies.
As a result of the EM signal attenuation, the reservoir lateral
size should be estimated by the low-frequency CSEM data be-
low or equal to 0.05 Hz at 2 km offset in shallow water and the
frequencies ≥ 0.25 Hz at the same offset in the deep-water case.
Due to a small lateral site of the target under study (~2 km),
the horizontal resolution of CSEM deteriorates at 4 km offset,
so the indication of the lateral size of the target gets smeared.
As one can see from the fourth columns of Figures 3–4, at 8
km offsets, the effect of the target becomes hardly detectable.
e effect of the small target on the long-offset CSEM re-
sponse is not local; i.e., we observed two “horns” on the curves.
e first anomaly occurs when the receiver is located above the
resistive target, and the second one—when the transmitter is
situated right above it. Due to such a complexity of the CSEM
responses, the application of 3D synthetic data inversion (e.g.,
Abubakar, et al., 2009; Frenkel, 2010) is required to properly
assess the usefulness of the wide-range offsets.
Target at 1.0 km below seafloor
In this section, we present CSEM modeling results for small
2D and 3D resistive targets embedded into isotropic and
anisotropic background cross sections in the presence of the
shallow (200 m) and deep water (Figures 5–7).
Isotropic background model. e 3D results in Figure 5 show
the sharpest lateral resolution at 2 km offset; the low-frequency
data (f ≤ 0.05 Hz) at offsets 2 and 4 km provide the strongest
MvO and NMvO responses. In the deep-water case (the bot-
tom row), the 3D NMvO at f = 0.25 Hz is ~1.25 at 2 km offset
and > 1.5 at 4 km offset; however, because of low levels of the
absolute magnitude of the electromagnetic signal, 3D data at
frequencies f > 0.25 Hz are not suitable for unconstrained 3D
inversion-based interpretation.
Interpreting the phase-shift data in addition to the ampli-
tude can increase the sensitivity to the resistive targets. We ob-
serve that the phase shift is the most sensitive to the reservoir in
shallow water at low frequencies. However, it may also be more
sensitive to the near-surface disturbing effects as well, e.g., shal-
low resistive gas hydrates or local seafloor heterogeneities.
Anisotropic background model. 2D and 3D anisotropic
modeling results for multiple frequencies are presented in
Figures 6 and 7, respectively. Inline CSEM measurements are
mostly affected by the vertical resistivity. In these anisotropic
cases, the contrast Rtarget/Rv halves, which results in reduction of
anomalous effects. Hobbs et al. (2009) argue that in the case of
formation anisotropy, the time-domain CSEM method typi-
cally requires using both short- and long-offset data to reliably
invert for the reservoir properties. We have drawn a similar
conclusion for frequency-domain CSEM.
Our modeling results suggest the following. Data at offset 2
km display a very low anomaly (second columns from left in
Figures 6–7). At 8 km offset, they exhibit a poor lateral resolu-
tion and NMvO ~1.25 at f = 0.05 Hz in the shallow water case
and a low level of signal for f ≥ 0.25 Hz in both the shallow
and deep water cases. e modeling shows that the most useful
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
April 2012 The Leading Edge 443
Marine and seabed technology
information about the subsurface can be extracted from CSEM
data at offsets around 4 km. At this offset, the NMvO curves
are ~1.3 and 1.2 in 2D and 3D cases, respectively. Typically,
the frequency-domain CSEM amplitude measurement dem-
onstrates elevated responses above the resistive targets. Howev-
er, at higher frequencies and longer offsets, a reversed behavior
is observed because of strong spatial oscillations of the electric
field due to wave phenomena. e same behavior of CSEM
responses was observed by Sasaki and Meju (2009) and other
authors. e results of modeling indicate ~20% anomalous ef-
fect drop at low frequencies due to the anisotropy. e drop of
the NMvO level due to the anisotropy is more significant at the
higher frequencies in deep water, e.g., the NMvO drops from
~1.7 to 1.2 at 4 km offset and f = 0.25 Hz.
Target at 2.0 km below seafloor
In this section, we present CSEM modeling results only for 2D
targets embedded in isotropic background cross sections at 2.0
km BSF (Figure 8). e modeling results for the deep-water
case show NMvO~1.2 at f = 0.25 Hz and 4 km offset. e
anomalous NMvO level is low at 2 km offset, and the absolute
signal level is low at 8 km offset. It should be mentioned that
responses from a 2-km-deep small resistive target in shallow
water environments are relatively weak for all the frequencies
and offsets. us, the feasibility of resolving such or deeper
targets should be further investigated using synthetic data in-
version. When the reservoir depth is equal to or greater than
its diameter, its anomalous response at all frequencies/offsets
is typically on the borderline of detectability. Because of low
NMvO levels even in the isotropic 2D cases, there is no point
to display here 2D/3D simulations for small targets located at
depths ≥ 2 km in the anisotropic backgrounds.
Case studies
To demonstrate the applicability of CSEM for detecting hydro-
carbon-charged targets in different real marine environments,
we present two case studies. e first case study is for the West
Mediterranean region—Abu Sir Field (Apache, 2003); and the
second one is for the North-West shelf offshore Western Aus-
tralia—the Vincent Van Gogh structure (DMP, 2008). In both
cases, the wells have shown commercial accumulations of HC.
Our 3D CSEM modeling-based feasibility studies were done
retrospectively after drilling, using seismic and log data-driven
subsurface resistivity models. In this section, we present key
results of these studies.
e Mediterranean case study. e Abu Sir-2X, in water
depth of 1028 m, was drilled to a total depth of 4012 m
and penetrated HC-bearing intervals in both Pliocene- and
Miocene-aged sands. It was the fifth well in Apache’s deep-
water Egyptian program and the first appraisal well after
four successful exploratory tests. e shallower pay interval
is stratigraphically equivalent to the Pliocene pay sands in the
Abu Sir-2X discovery well. Logs and pressure data identified
a 41-m column between 2036 m and 2077 m with 25.6 net
meters of gas pay (Apache, 2003). Roberts et al. (2005) pres-
ents the rigorous rock physics analysis based on interpretation
of the seismic and well-logging data acquired over the Abu
Sir field, including the anisotropic logs from well Abu Sir-2X.
is well was drilled with the expectation of finding com-
mercial HC throughout the entire high-amplitude interval.
However, the analysis of well logging data showed high gas
saturation in the upper sand, but low gas saturation in the
lower much thicker sand. It is a representative example dem-
onstrating that the seismic data alone can provide misleading
Figure 9. e Mediterranean case study. (a) e top (X × Y) view
of the target model. (b) CSEM 3D synthetic data for target reservoirs
of 20 and 30 m thickness simulated along the red X axis of (a), f =
0.25 Hz and offset 7.5 km. e results of 3D modeling show that
the CSEM survey is feasible. ese results are in line with the CSEM
survey data (c) for the same frequency and offset from a line crossing
the Abu Sir-2X well (omsen et al., 2007). e arrow indicates the
towing direction; the purple and grey boxes are for the in-towing and
out-towing measurements, respectively.
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
444 The Leading Edge April 2012
Marine and seabed technology
Figure 10. e Offshore Australia case study. (a) Geological cross section at the drilling location of the discovery well Vincent-1 (left) and
the results of petrophysical interpretation for the interval 1300–1350 m (right). e target reservoir under study is the Lower Barrow Group
sand. e red and blue curves in Tracks 2, 4, and 5 show the true formation resistivity Rt (track 2), the water saturation Sw (track 4), and
the hydrocarbon volumetrics (track 5) derived using conventional resistivity interpretation method and the near real-time resistivity log data
inversion, respectively (Frenkel et al., 2007). e results of 3D CSEM modeling of multiple frequencies for the isotropic (b) and anisotropic (c)
backgrounds exhibit relatively low NMvO levels.
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
April 2012 The Leading Edge 445
Marine and seabed technology
information regarding pay zones; therefore, additional physi-
cal measurements are needed to accurately assess pay zone siz-
es and the amount of HC in the ground. e marine CSEM
survey over the Abu Sir Field was performed by EMGS after
the drilling. Some results of this survey were presented by
omsen, et al. (2007). e above sources provide us with
sufficient data to perform 3D CSEM modeling-based analy-
sis, the summary of which is presented below.
e model parameters are as follows. e horizontal resis-
tivity of the background is Rh = 0.8 Ωm and the anisotropy
ratio Rv/Rh = 2. e lateral target dimensions are shown in
Figure 9a; the target top is located 1.0 km BSF, and its resis-
tivity is Rt = 40 Ωm. In Figure 9b, we present the results of
3D EM modeling for the target thicknesses of 20 and 30 m.
e results of modeling show a strong CSEM normalized sig-
nal, NMvO > 1.5, for both cases. It should be underlined that
the results of this forward modeling run are in line with the
results of the marine CSEM survey for the SW-NE line cross-
ing the Abu Sir-2X well (omsen et al., 2007). e field
NMvO has a curve shape similar to the synthetic curve and
max out at about 2.0 level, which indicates that the reservoir
thickness should be close to 30 m. is study gives the green
light for surveying and affirmatively suggests “To CSEM.”
e offshore Australia case study. In this case study, we used
well log data from a vertical exploration well Vincent-1 from
the North West Shelf region of offshore Western Australia.
e Vincent-1 discovery, drilled in November 1999, en-
countered a 8.5-m gas column and a 19-m oil column in the
Lower Cretaceous Barrow Sandstones (DMP, 2008). In Fig-
ure 10a, we present the results of resistivitiy log data inversion
and subsequent petrophysical interpretation (Frenkel et al.,
2007). e subsurface resistivity model parameters are as fol-
lows. e water depth is 380 m, the formation resistivity was
recovered by the log data inversion (Figure10a, second track).
On average the horizontal resistivities of the overburden, tar-
get, and underburden are 1.2, 10.0, and 0.9 Ωm, respectively.
e resistivity anisotropy of the background was not known;
to simulate an anisotropic case, we set the anisotropy ratio of
the overburden Rv/Rh = 2. e target thickness is 28 m, and
its top is at 930 m BSF.
Because the transverse resistance of the target reservoir is
low (~280 Ω × m2), we modeled several scenarios with increas-
ing lateral dimensions of the target: (X × Y) = 2 × 2, 4 × 4, 6 ×
6, and 8 × 8 km to assess the minimum reservoir size produc-
ing normalized magnitude ratio above the 1.1 threshold. In
Figures 10 b-c, we present the results of 3D CSEM modeling
for 6 km offset and three frequencies f = 0.05, 0.25, and 1.0
Hz for the reservoir size of 8 × 8 km. Modeling indicates a
relatively low CSEM normalized signal, NMvO = 1.12 and
1.07, for isotropic (Figure 10b) and anisotropic (Figure 10c)
backgrounds, respectively. is 3D modeling study gives the
red light for surveying under these environmental conditions,
anisotropic overburden, and target parameters, and suggests
“Not to CSEM.”
Recommendations and conclusions
We performed a frequency-domain CSEM benchmark mod-
eling-based study for small 2D/3D resistive HC targets in
shallow and deep-water environments. e spatial resolution
of CSEM for short and moderate Tx-Rx offsets (e.g., <6 km) at
the fundamental frequency of 0.25 Hz and at lower frequen-
cies is primarily determined by the offset. e higher frequen-
cies of 1–2 Hz provide a higher spatial resolution. However,
the depth of investigation of the EM field in this frequency
range is restricted by the signal attenuation due to the skin
effect.
e selection of optimal CSEM excitation frequencies de-
pends on the target and background formation parameters, in-
cluding but not limited to the target depth, its transverse resis-
tance, the background resistivity and anisotropy, and the water
depth. Based on the modeling study, we make the following
recommendations:
t To detect small and moderate size resistive targets in iso-
tropic background formations, the offsets should be equal
or slightly longer than the target diameter. Offsets should
be roughly doubled for anisotropic background formations
with the anisotropy ratio Rv/Rh = 2.
t When the reservoir depth is equal to or greater than its di-
ameter, its anomalous response at all frequencies/offsets is
typically on the borderline of detectability.
t In the shallow water (WD ~ 200 m) environments, the op-
timal frequency choices are: f = 0.25–1 Hz for offsets 2–4
km and f = 0.05–0.25 Hz for longer offsets (>4 km).
t A similar recommendation is applicable to the deep-water
environments: the optimal frequency range is 0.25–1.0 Hz
for short offsets (2–4 km) and 0.05–0.25 Hz for longer off-
sets (4–8 km). e fundamental CSEM frequency of 0.25
Hz and higher frequencies do not provide a strong enough
signal-to-noise-ratio in deep water for longer offsets due to
strong skin effect.
In this study, we considered small HC targets embedded
in a homogeneous isotropic and anisotropic half-space. is
3D approach can be used for fast screening of a large number
of potential prospects to be explored. Once individual pros-
pects for CSEM surveying and well drilling locations have been
selected, a more comprehensive 3D modeling- and inversion-
based synthetic data feasibility study should be performed us-
ing real bathymetry and subsurface geological models, as well
as data and model uncertainties to accurately assess feasibility
of a marine CSEM survey and then to optimize its design.
e 3D EM modeling we applied utilizes the automatic
optimal gridding algorithm enabling accurate and rapid sim-
ulation of EM fields in arbitrary 3D anisotropic media and
can, therefore, be efficiently used throughout the entire CSEM
workflow. In the synthetic case presented here, the simultane-
ous multifrequency simulation of the EM fields for one trans-
mitter position took less than a minute on a single 2.2 GHz
processor.
We effectively tested our fast 3D CSEM modeling tech-
nology on two field data-based case studies using the ground
truth information about the subsurface resistivity—accurately
interpreted logs from the wells that penetrated the reservoirs.
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/
446 The Leading Edge April 2012
Marine and seabed technology
Often in exploration settings, well-logging data is not avail-
able or available from distant wells. is makes accurate 3D
EM modeling even more essential for integrating all known
seismic and other geological and geophysical data to determine
the potential value of CSEM surveying, optimizing the survey
design, and answering the key question: To CSEM or not to
CSEM?
References
Abubakar, A., T. M. Habashy, M. Li, and J. Liu, 2009, Inversion algo-
rithms for large-scale geophysical electromagnetic measurements:
Inverse Problems, 25, doi:10.1088/0266-5611/25/12/123012.
Apache Logs 2nd Success on Abu Sir Field Offshore Egypt, 2003:
www.rigzone.com/news/article.asp?a_id=5808 (last viewed on
12.15.2011).
Davydycheva, S. and V. Druskin, 1999, Staggered grid for Maxwell’s
equations in arbitrary 3-D inhomogeneous anisotropic media, in M.
Oristaglio and B. Spies, eds., ree-dimensional electromagnetics:
SEG, 119–137.
Davydycheva, S., V. Druskin, and T. Habashy, 2003, An efficient finite-
difference scheme for electromagnetic logging in 3D anisotropic
inhomogeneous media: Geophysics, 68, no. 5, 1525–1536, http://
dx.doi.org/10.1190/1.1620626.
Davydycheva, S., 2010, Separation of azimuthal effects for new-gener-
ation resistivity logging tools—Part I: Geophysics, 75, no. 1, E31–
E40, http://dx.doi.org/10.1190/1.3269974.
Davydycheva, S., and N. I. Rykhlinski, 2011, Focused–source elec-
tromagnetic survey versus standard CSEM: 3D modeling in com-
plex geometries: Geophysics, 76, no. 1, F27–F41, http://dx.doi.
org /10.1190/1.3511353.
Department of Mines and Petroleum (DMP) of Western Australia,
2008 Western Australian Oil and Gas Review: http://www.dmp.
wa.gov.au/documents/OilandGasReview2008t(1).pdf (last viewed
on 3/5/2012).
Druskin, V. and L. Knizhnerman, 1994, Spectral approach to solv-
ing three-dimensional Maxwell’s equations in the time and fre-
quency domains: Radio Science, 29, no. 4, 937–953, http://dx.doi.
org/10.1029/94RS00747.
Frenkel, M. A., M. Benefield, M. Gonfalini, and N. Mendybaev, 2007,
Near real-time petrophysical analysis using multidimensional for-
mation models: 48th Annual Logging Symposium.
Frenkel, M. A. and S. Davydycheva, 2009, A modeling study of low-
frequency CSEM in shallow water: 71st EAGE Conference and Ex-
hibition, Extended Abstracts.
Frenkel, M.A., 2010, Inversion of 3D marine CSEM data using seed-
type initial models: Progress in Electromagnetic Research Sympo-
sium (PI E R S), 613 – 616.
Hobbs, B., D. Werthmüller, and F. Engelmark, 2009, e effect of resis-
tivity anisotropy on transient electromagnetic earth responses: 71st
EAGE Conference and Exhibition, Extended Abstracts.
Sasaki, Y. and M. A. Meju, 2009, Useful characteristics of shallow
and deep-water marine CSEM responses inferred from 3D finite-
difference modeling: Geophysics, 74, no. 3, F67–F76, http://dx.doi.
org/10.1190/1.3168616.
omsen L., D. Meaux, S. Li, C. Weiss, A. Sharma, N. Allegar, and
K. Strack, 2007, Novel marine electromagnetics: from deep into
shallow water: www.kmstechnologies.com/Files/Technology_Li-
brary/Presentations/omsen_etal.pdf (last viewed on 12.15.2011),
presented at the SEG Annual Meeting Special Session “Recent Ad-
vances and the Road Ahead”.
Corresponding author: michael@singularem.com
Downloaded 12 Apr 2012 to 69.151.154.25. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/