You are on page 1of 37

 

 
Theoretical assessment of 3-D magnetotelluric method for oil and gas exploration: Synthetic examples
Kun Zhang, Wenbo Wei, Qingtian Lu, Hao Dong, Yanqing Li
PII:
DOI:
Reference:

S0926-9851(14)00104-9
doi: 10.1016/j.jappgeo.2014.04.003
APPGEO 2474

To appear in:

Journal of Applied Geophysics

Received date:
Revised date:
Accepted date:

23 July 2013
3 April 2014
4 April 2014

Please cite this article as: Zhang, Kun, Wei, Wenbo, Lu, Qingtian, Dong, Hao,
Li, Yanqing, Theoretical assessment of 3-D magnetotelluric method for oil and
gas exploration: Synthetic examples, Journal of Applied Geophysics (2014),
doi:
10.1016/j.jappgeo.2014.04.003

This is a PDF le of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its nal form. Please note that during the production process
errors may be discovered which could aect the content, and all legal disclaimers that
apply to the journal pertain.

ACCEPTED MANUSCRIPT
Theoretical assessment of 3-D magnetotelluric method for oil and gas

Institute of Mineral Resources Chinese Academy of Geological Sciences, MLR Key Laboratory of

SC
R

IP

a,

Kun Zhang *, Wenbo Wei , Qingtian Lu , Hao Dong , Yanqing Li

exploration: Synthetic examples

Metallogeny and Mineral Assessment, Beijing 100037China

School of Geophysics and Information Technology, China University of Geosciences, Beijing, PR

NU

China
c

MA

Geophysical R&D institute of China Oilfield Service Limited, Tanggu, Tianjin 300451,China

* Corresponding author.

TE

E-mail address: zhangkun1010@163.com (K. Zhang).

CE
P

Abstract

In petroleum explorations, seismic reflection technique has been almost always the preferred

AC

method for its high exploration depth and resolution. However, with the development of three
dimensional (3D) inversion and interpretation schemes, much potential has been shown in MT
method dealing with complex geological structures as in oil and gas exploration. In this study,
synthetic geophysical models of petroleum reservoir structures are modelled and utilized to
demonstrate that feasibility of 3-D MT technique for hydrocarbon exploration. A series of typical
reservoir structure models are constructed and used to generate synthetic MT and seismic data
to test the capabilities of 2-D / 3-D MT and 2-D seismic inversion techniques. According to the
inversion comparison, in addition to correctly retrieves the original forward model, the 3-D MT
method also has some advantages over the reflective seismology method, which suffered from

ACCEPTED MANUSCRIPT
the lack of reflection wave and multiple wave problems. With the presented 3-D high resolution
MT inversion method, MT techniques should be employed as one of the first choices for

IP

petroleum explorations.

SC
R

Keywords
Magnetotelliric; 3-D inversion; Oil exploration; Synthetic example

NU

1. Introduction

MA

In petroleum explorations, the magnetotelluric (MT) technique is often used as a complement


method to the primary technique of reflective seismology method (From Wikipedia, the free

encyclopedia). With the diffusive nature of electromagnetic (EM) wave propagation, the vertical

TE

resolution of EM exploration methods usually cannot match those of reflective seismology in

CE
P

cases such as distinguishing deep strata. Therefore MT is often employed as a backup method
where seismic reflection method performs poorly in complex terrain and geological conditions

AC

(Sun et al., 2003; Constable et al., 1998), for example, in situations where carbonate outcrops
produce strong reflections that interfere with accurate seismic detection.
The MT method, however, possesses several important advantages over other methods: it is
unaffected by high-resistivity layers, capable of great exploration depth and low-cost (Chen and
Wang, 1990; Wei, 2002). It is also more environment-friendly because of its light-weight
equipment and reduced hazards comparing to seismic reflection methods (e.g. no drills, no
explosives). In most cases, MT data are collected along 2-D profiles across geoelectrical strikes
with 2-D assumptions that the layers and bodies underground extent unlimitedly in strike
direction. In some instances, accurate interpretation of 3-D geology can be achieved by careful
2

ACCEPTED MANUSCRIPT
application of 2-D modeling and inversion (Wannamaker, 1984). But, in other cases, 2-D inversion
may not get reasonable models, especially in a complex geological environment. Besides the

IP

different physical response parameters, the vertical resolution of 2-D MT may be not as good as

SC
R

seismic reflection methods. However, 3-D MT modeling does not require the 2-D underground
assumption nor need approximate calculations like strike rotation. Theoretically, 3-D methods can
model any 3-D geological body, although might be restricted by the algorithms of forward

NU

modeling in reality. And the location and uniform limitation (grids and sites) in 2-D method

MA

(Avdeev et al., 2000; Chen and Zhao, 2009) is rarely appeared in 3-D (the method used in this
paper), because each site corresponds to each grid contained it which means the relative position

CE
P

TE

inversion results (Zhang, 2013).

of the site in the grid is not critical and the relative size of the grids will not influence the

A number of 3-D MT inversion methods have been proposed, including rapid relaxation inversion

AC

(Smith and Booker, 1991; Tan et al., 2003), conjugate gradient inversion (Mackie and Madden,
1993; Lin, 2011), quasi-linear approximate inversion (Zhdanov, 2000), Bayesian statistical
inversion (Spichak et al., 1995) and nonlinear conjugate gradient (NLCG) inversion (Newman and
Alumbaugh, 2000; Mackie et al., 2001; Zhang et al, 2013). Although each of these methods has its
own advantages and disadvantages (see comparisons by Zhang et al., 2013), we consider the
NLCG method to be the ideal 3-D MT inversion method with the advantage of high resolution,
efficiency and precision. There are many applications of crust and ore exploration using 3-D MT
survey and inversion, including Coso field exploration for large geothermal reservoirs (Newman et
al, 2005), Xinjiang field AMT exploration for study the electrical structure of Cu-Ni mining (Xiao et
3

ACCEPTED MANUSCRIPT
al., 2010), Iceland field exploration for understanding the complex geothermal systems of the
Hengill and Krafla volcanic complexes (Gasperikova et al, 2011), Gulf of Mexico exploration for

IP

demonstrating the capability of imaging a sea-bottom resistivity structure (Zhdanov et al, 2011),

SC
R

Krysuvik exploration in SW Iceland for understanding the electrical structure of the high
temperature areas (Hersir et al., 2013), Central California field exploration for studying the
electrical conductivity structure of the San Andreas Fault (Tietze and Ritter, 2013) and so on.

NU

After effective results and interpretations are obtained from 3-D MT inversion models, the

MA

advantages of 3-D inversion are manifested on large real-world data set and resolution of

complex structure (Tietze and Ritter, 2013) and so on (Newman et al, 2005; Xiao et al., 2010).

TE

MT sounding can reflect the underground electrical interfaces where conductivity differences

CE
P

exist. An oil reservoir usually requires an impermeable cover, connected reservoir and
impermeable base, so the gas, oil or water in reservoir is difficult to escape. The electrical

AC

difference between cover and reservoir are usually great and can be easily detected by MT
method. In the case of an anticline fold oil trap, which is a strong manifestation, electrical
difference between the anticline and background rock can also be great and can be distinguished
by MT method. Although there are some successful examples of oil and gas exploration using MT
included the Ashili region (Wu, 2012) and Gui Depression (Xia et al., 2012) in China, and some
areas of Russia (Berdichevsky et al., 2010) and Japan (Khalil and Ushijima, 2003), MT is still not
the usual method in the hydrocarbon exploration. And only 2-D inversion methods are used in
those interpretations. So in our opinion, the applications of 3-D MT in petroleum exploration are
not well-informed.
4

ACCEPTED MANUSCRIPT

Utilizing the 3-D MT inversion code presented by Zhang (2012) and Zhang et al. (2013), our

IP

primary goal of this study is to demonstrate the value of 3-D MT technique in oil and gas

SC
R

exploration. A series of 3-D models of typical reservoir structures are constructed and used to
generate synthetic data of MT and seismic reflection method. Both 2-D and 3-D modeling and
inversion were carried out with the synthetic data, along with 2-D reflective seismology

NU

inversions to demonstrate the validity of those methods in oil and gas exploration. The result

MA

comparison shows that the 3-D method has a better resolution and accuracy then the 2-D MT
method. Accurate 3-D MT inversion may also compensate for seismic waveform errors such as

TE

multiple waves and waveform distortions generated by undulating interfaces.

CE
P

2. Model of typical oil-bearing structure and synthetic data


Some sedimentary rocks contained oil and gas deformed due to the enormous pressure, then oil

AC

flowed into anticline and formed rich region named oil-bearing structure on petroleum geology.
Typically, gas is at the top of the anticline due to the minimum density, and oil is in the middle
covering the water (Sun et al., 2011).
Typical reservoir construction usually contains two elements: reservoir and trap. And the
electrical impedance variation caused by resistivity differences is the basic to distinguish the
layers, in other words, (3-D) MT exploration can reflect every reservoir and trap layer
theoretically because their physical parameter (resistivity) is different (Batzle and Wang, 1992;
Chave and Jones, 2012). Reservoir, porosity and permeability are the keys that oil and gas need
interconnected voids in permeable rocks. Trap can gather the oil and gas for a certain time
5

ACCEPTED MANUSCRIPT
interval, and can be formed by deposition, faults and folds. So oil and gas must be collected in the
trap with permeable and porous rocks (Tissot et al., 2003). And they typically occur in the core

IP

unit of an anticline where, along with groundwater, they collect in pores and fissures in

SC
R

permeable reservoir rock, which is overlain by impermeable rocks (e.g., shale or limestone). Such
a representative geological model is shown in Fig. 1(Sun et al., 2011). And the petrophysical

NU

properties of the components are listed in Table 1.

MA

The geophysical model (Fig. 2) consists of five layers (from top downwards): shale, limestone, gas,
oil, groundwater and limestone overlying a basement rock. The material properties to be used in

TE

the model are extracted from Table 1, and are shown in Table 2.

CE
P

In 3-D models, identify the horizontal directions as x and y, and vertical direction as z. But in 2-D
models, only x and z directions exist. The length of the central section along the y-axis is 55 km

AC

(along the y-axis, we only use 10-65 km section from the synthetic model), and the depth of z
direction is 10 km (along the z-axis, we only use 0-10 km section from the synthetic model). The
layers are arranged as a symmetrical anticline; the bottom depths of layers 15 at the highest (i.e.,
central, or core) point are respectively 1800, 3200, 3600, 4400 and 4800 m, and the bottom
depths of layers 1, 2, 4 and 5 at the lowest point on either side are respectively 2400, 3200, 4000,
4800 and 6100 m (the gas layer, layer 3, does not extend to the sides of the structure). The layers
of gas, oil and groundwater (layers 35) are thinner than the other strata. The layer representing
groundwater, layer 5, is thicker at the sides than at the core. In the x-axis of 3-D model, the
structure is not the same as y. Although layer 3 only exist in the center in 3-D model, layer 4 and
6

ACCEPTED MANUSCRIPT
5 are thinning from center to both sides of x-axis and the transverse range reduces at y-axis. The

IP

resistivity of layer 1-5 is 30m, 100m, 100000000m, 10000m and 10m.

SC
R

3. MT response of synthetic model

Utilizing the above synthetic material properties, the 2-D and 3-D impedance, apparent resistivity

NU

and impedance phase responses are determined by forward calculation.

MA

3.1 2-D forwarding

A 2-D finite element forwarding program (Zhang et al., 2008) is used to compute responses in the

electric field parallel polarization (TE) and (magnetic field parallel polarization (TM) modes for the

TE

profile at 12 sites and for 12 periods in the range 0.011000 s. Noise is not specifically included in

CE
P

the synthetic data. The boundary extents are assumed to be 200 km horizontally and 100 km

AC

vertically. The background resistivity is set at 100 m.

In Fig. 3 the 2-D forwarding results are different for the TE and TM modes. In TM mode, high
resistivity anomalies are reflected more strongly (apparent resistivity > 90 m; phase < 42).
Base on the synthetic model, the groundwater layer below and around the gas and oil layers is
shown as middle-low apparent resistivity (30-50m) at middle frequency around the center high
apparent resistivity, and the apparent resistivity reaction is stronger in TM mode, and phase is
stronger in TE mode. In addition, there is a low-resistivity band (< 30 m) in the apparent
resistivity sections at frequencies around 10 Hz.

ACCEPTED MANUSCRIPT
3.2 3-D forwarding
For the 3-D model, the 3-D staggered-grid finite difference forwarding program of Zhang (2012) is

IP

used to compute the impedance tensor responses generated at 144 sites and 12 periods in the

SC
R

range 0.011000 s. The boundaries and background resistivity are set as above for 2-D
forwarding.

NU

Fig. 4 shows sections of the apparent resistivity and impedance phase of XY and YX modes. About

MA

three abnormal parts are in the sounding sections, low apparent resistivity and high phase part in
high frequency, middle apparent resistivity and middle phase part in middle frequency, high

TE

apparent resistivity and low phase part in low frequency.

CE
P

3.3 Comparison and analysis of 2-D and 3-D responses


The pseudo-sections in Fig. 5 show the 2-D and 3-D apparent resistivity and phase section of the

AC

model (Fig. 2-b). Although 2-D and 3-D responses are accurate enough, the low resistivity layer
which exists in the 2-D forwarding but not in 3-D forwarding and the synthetic model shows that
3-D responses are more faithful to the model. The apparent resistivity is < 30 m for the
low-resistivity anomaly, and > 95 m for the high-resistivity anomaly which reaction is stronger
than 2-D. And the phase responses are similar.

4. Inversion of typical MT synthetic modeling response


Our goal in this paper is to demonstrate the applicability of the 3-D MT technique to oil and gas
exploration, taking into account accuracy and resolution issues in both 2-D and 3-D inversions.
8

ACCEPTED MANUSCRIPT
We apply 2-D inversion to the central profile using apparent resistivity and phase data, and 3-D
inversion to 3-D model using complex impedance tensor. And only 3-D forward modeling data are

SC
R

IP

used in both of 2-D and 3-D inversions. A discussion of the advantages of 3-D inversion follows.

4.1 The used MT inversion methods

For the inversion method, we used the non-linear conjugate gradient (NLCG) method described

NU

by Rodi and Mackie (2001) for 2-D inversion, and our NLCG code from Zhang (2012, 2013) for 3-D

MA

inversion.

The objection function and its gradient can be expressed as:

[(d F (m)) / ]2 mTW TWm

TE

(1)

CE
P

d / m 2 Re ((d F (m)) / ) F / m
where is the object function d is the collected data (apparent resistivity and phase;

AC

impedance); is data error; F (m) is forwarding factor; is a regularization factor; m is model; and

W is the model covariance matrix.

The search length and direction can be expressed as follows:

p0 C0 (0 / m0 )
pk Ck (k / mk ) k pk 1 , k 1, 2,

(k / mk )T Ck ((k / mk ) (k 1 / mk 1 ))
(k 1 / mk 1 )T Ck 1 (k 1 / mk 1 )

mk ,l mk k ,l pk , l 0,1, 2,

k ,l 1 k ,l

(k ,l / mk ,l )T pk
pkT H k ,l pk

(2)
9

ACCEPTED MANUSCRIPT
Where p is the search direction, is the search length, C is a preconditioning factor, l is the line

IP

search number, and H is a quadratic function of the object function.

SC
R

In this 3-D algorithm, the preconditioning method (detail) of the NLCG optimization method is
improved:

Cl (i, j, k ) [l (i, j, k ) I W TW ]1

/ ( Z

obs

1
Z ) m(i , j , k ) 2

(4)

NU

(i , j , k )

MA

C is a preconditioning factor, is a non-independent coefficient to cell (i,j,k). C approximates


the Hessian matrix, thus avoiding four pseudo-forwarding calculations in each iteration ,as a

TE

CE
P

compute p , and

parameter matrix related with the current resistivity model and data error and it is used to

For NLCG 3-D inversion, We make four changes to improve the efficiency and practicality of the

AC

previous algorithm (Zhang, 2012): the preconditioning method is improved using a parameter
matrix related to the current resistivity model and data error as the preconditioning factor to
approximate the Hessian matrix, the OPENMP parallel API is utilized to establish an efficient
parallel inversion structure based on frequency which means the data are distributed
frequency-wise to different CPU threads in one computer using shared memory (general PC is
enough for the 3-D inversion, and recommendation configuration is 6 cores and 16 GB memory
for large-scale computing), a lower number of forward and pseudo-forward iterations is used to
reduce the computation time, positioning storage is used to store sparse matrices to save
memory, the terrain is added into the inversion algorithm. For approximately 200 survey sites and
10

ACCEPTED MANUSCRIPT
12 periods, the compute time is less than 20 s per iteration per period, and less than 30 min for

IP

the whole inversion process, so the inversion algorithm and code are fast and low-loss.

SC
R

For the 2-D and 3-D inversion parameters, all inversions are commenced from a 100 m
homogeneous half-space with the same grid setting in center profile, and the regulation factor is
1, and error floor is 10%, and gradient method is used in preconditioning process in inversion.

NU

Data errors are assumed to be 10%, as used for inversion. The grid set is 32 (horizontal)*38

MA

(vertical) for 2-D inversion, and 32 (x)* 32 (y)*38 (z) for 3-D.

4.2 Comparison and analysis of center cross-sectional results of MT

TE

4.2.1 Fitting degree of 2-D MT inversion

CE
P

Plotted results of the 2-D TE and TM joint inversion of the central profile are shown in Fig. 6,
showing the two sounding modes of the synthetic and inversion responses. The normalized root

AC

mean square (RMS) misfit is 0.24, so the sounding of the synthetic and inversion models are very
similar, and only small differences in the phase value are found (synthetic data: 45 to 35;
inversion sounding: 44 to 36).

4.2.2 Fitting degree of 3-D MT inversion (central profile)


Fig. 7 shows that inversion model sounding is very similar to the synthetic data, with differences
of 35 m and 42 in the synthetic sounding sections and 40 m and 43 in the inversion
occurring only at the sides of the y-x pseudo-section for the TM mode.

11

ACCEPTED MANUSCRIPT
4.2.3 Comparison and analysis of 2-D and 3-D MT inversion cross-section
Taking the location, depth and lateral extent of the model strata shown in Fig. 2 as benchmark

IP

values, Fig. 8 shows the electrical structure of the central profile. In the 2-D inversion profile, the

SC
R

resistivity of the shale is 2940 m, with a thin electrical transition band influenced by the
underlying high-resistivity limestone and oil layers. The resistivity of the limestone is 32107 m,
with a thick electrical transition band influenced by the low-resistivity shale and groundwater

NU

layers. The oil and gas layer of resistivity 40120 m contained a thick electrical transition band

MA

at both sides, influenced by groundwater having a resistivity of 3280 m; the performance


characteristic of this electrical transition is influenced by the presence of the underlying

TE

limestone.

CE
P

The number 1-5 represents the stratum of shale, limestone, gas, oil, water in the synthetic model.
Compared with the layer 1-5, the gas layer 3 and oil layer 4 cannot be distinguished in 2-D and

AC

3-D inversion result sections with the top depth of 3200 m and 3600 m, but the top boundary of
layer 2 is reflected clearly with the depth of 1800-2400 m. The boundary of layer 4 and 5 and
limestone below layer 5 of 3-D result is more clear and accurate than 2-D.

In the profile of the 3-D inversion model (vertical and horizontal scales adjusted to be consistent
with the 2-D profile, for easier comparison), the resistivity of the shale is 2940 m, contained
within a single thick electrical transition band; the resistivity of the limestone is 50100 m,
contained within a single thick electrical transition band; the resistivity of the oil and gas layer is
40135 m, contained within a single thick electrical transition band on each side; and the
12

ACCEPTED MANUSCRIPT
resistivity of the groundwater is 4063 m; the performance characteristic of this electrical

IP

transition is influenced by the presence of the underlying limestone.

SC
R

For the inversion results, the resistivity of every layer cannot reach to the value in the synthetic
model, because the reflection of inversion result is gather effect of the objection and the around
rock and layers, so a transition band is usually exist in the inversion results. From the comparison

NU

of synthetic, 2-D and 3-D inversion models, the resistivity of background limestone is 100 m

MA

(blue) in the synthetic model; it is lower (green) in the 2-D inversion model and the range becomes
small; the resistivity and range is more accurate in 3-D model without the influence of water layer.

The oil and gas layers are not distinguished both in 2-D and 3-D inversion models, but the electrical

TE

difference of the interface of oil and water, water and bottom limestone is clearer in 3-D model, and

CE
P

the anticlinal form of main layers is displayed, such as the range of top layer 1 and bottom high
resistivity layers. So the 3-D inversion results show a high degree of reduction in difference on the

AC

synthetic model.

4.3 Sounding comparison and analysis of 3-D inversion


Fig. 9 shows the plotted results of the 3-D inversion for the real and imaginary parts of the Zxx, Zxy,
Zyx and Zyy response components of the synthetic and inversions, at a frequency of 1.33 Hz. The
normalized RMS misfit is 0.19, so the soundings of the synthetic and inversion model are very
similar, and only small differences exist in the Zxx and Zyy components due to the errors, which are
of the same order of magnitude as the errors in Zxy and Zyx. In addition, the absolute magnitudes
of the errors are much low.
13

ACCEPTED MANUSCRIPT

Fig. 10 shows four superposed yz cross-sections on the x-axis of the synthetic and inversion

IP

models. From top to bottom, the stratum layers are shale, limestone, gas, oil, water and

SC
R

limestone in the synthetic model (Fig. 10-a). Compared with the synthetic model, the gas layer
and oil layer cannot be distinguished with the top depth of 3200 m and 3600 m, but the top
boundary of limestone layer is reflected clearly. The boundary of water layer and limestone below

MA

NU

it is clear.

Fig. 11 shows this information in different views from different perspectives. For the whole 3-D

model, the inversion model is very similar to the synthetic model, and the shape, lateral extent

TE

and depth of every layer is reflected clearly and accurately with the exception of the limestone-oil

CE
P

and oilgas interfaces, which differed considerably. Because the horizontal and vertical extent of
the oil and gas in the reservoir are relatively small (maximum width 60 km, maximum height 1 km,

AC

average thickness < 600 m) and the depth of the top of the oil and gas is 5 km below the surface,
the data anomalies are weak and difficult to interpret accurately. However, it does not affect the
interpretation of the high resistivity layers. So 3-D inversion has the potential to find the best
possible representation of the electrical conductivity structure of the subsurface (Hubert, 2012).

5. Comparison of MT and seismic results


We use 2-D high-order finite difference method (Tesseral software, Tesseral Technologies Inc) to
do the forward modeling of seismic wave. The comparative feasibilities of oil and gas exploration
using the MT and seismic methods are illustrated in Fig. 12, which shows the 3-D inversion results
14

ACCEPTED MANUSCRIPT
of MT and 2-D seismic forwarding waveform. Four accurate electrical layers are reflected in the
MT section. The seismic waveform reflects the interfaces of every layer overall, but the relative

IP

positions of the interface of the two-way travel time vary, and the interface between limestone

SC
R

and oil at the sides of the section is not reflected. In addition, the interface of oil and
groundwater appears longitudinally discontinuous, and multiple waves are clearly seen.

NU

Although signal-processing techniques are available for removing multiple wave signals, it is

MA

nevertheless difficult and may significantly affect subsequent interpretation. The longitudinal
faults can only be processed by the depth prestack method if the absolute velocity structure is

known precisely. Therefore, use of the seismic reflection method alone to interpret oil and gas

CE
P

TE

prospects may not necessarily give the true picture.

Although 3-D MT does not give a sufficiently precise vertical resolution, its superior horizontal

AC

resolution helps to delineate the location and extent of the objective layer. Thus, from the results
of this study, we consider that the MT exploration method should be used in oil and gas
prospecting.

6. Extensions
6.1 Additional typical oil-bearing structure
Based on the 3-D MT inversion resolution results, three additional, more comprehensive
geophysical models (Models 1, 2, 3 in Fig. 13, Fig. 14, Fig. 15) are used to assess the validity and
importance of the 3-D MT method for oil exploration.
15

ACCEPTED MANUSCRIPT

Model 1 (Fig. 13) consists of shale (layer 1; 30 m; bottom depth 800-1200 m), limestone (layer 2;

IP

100 m; bottom depth 1400-1800 m) and an oil layer (layer 3; 10 000 m; 50 km wide along

SC
R

x-axis and become narrow on both sides; bottom depth 2200-2600 m and 400-600 m thick). Model
2 (Fig. 14) contains the above shale (layer 1; 30 m; bottom depth 1800-2200 m), limestone (layer
2; 100 m; bottom depth 3200-4000 m) and oil (layer 3; 10 000 m; 20 km wide along x-axis;

NU

bottom depth 3600 m and 400 m thick), plus groundwater (layer 4; 10 m; bottom depth

MA

5200-6000 m and 1600-2000 m thick and become narrow on both sides of x-axis). Model 3 (Fig.

15) contains shale as model 2, plus oil-bearing limestone (layer 5; 1000 m; bottom depth 4400

m and 2600-3000 m thick; 50 km wide along x-axis and become narrow on both sides) and

TE

groundwater (layer 4; 10 m; bottom depth 5200-6000 m and 400-1200 m thick and become

CE
P

narrow on both sides of x-axis).

AC

Fig. 16-18 shows the 2-D and 3-D inversion models of the central profile. The number 1-5
represents the stratum of shale, limestone, oil, water and complex layer in the synthetic models
shown in Fig. 13-15. The 3-D inversion accurately reflects the electrical properties and the
thickness and lateral extent of main layers, including the bottom shale boundary depth
1800-2200 m in model 2 and 3 with resistivity values of 2530 m; the bottom limestone
boundary depth 1800-2200 m in model 1 and about 3000 m in model 2; the bottom oil boundary
depth 2200-3000 m in model 1 (200400 m); the bottom water boundary depth 5200-5800 m
in model 2 (1020 m) and about 4400-6000 in model 3 (4050 m); the bottom oil-bearing
limestone boundary depth about 4400 m in model 3 (100220 m).
16

ACCEPTED MANUSCRIPT

Fig. 19(a-c) shows the 3-D inversion resistivity for the yz sections in Models 1, 2 and 3. The

IP

sections accurately reflect the differences of stratums in electrical properties which contain shale,

SC
R

limestone, oil, groundwater, oil-bearing limestone. The electrical structure of inversion models
are consistent with the synthetic model in strata thickness and extended range, except the oil
layer in Model 1 (Fig. 19 a) because the electrical layer cannot be distinguished with the top

NU

depth of 1600-1800 m and bottom depth of 2200-2600 m. Comparisons between the 2-D and 3-D

MA

inversion models indicate that the 3-D MT results are closer to the synthetic model than 2-D with
analyzing both location and magnitudes of different media because of the data information (2-D

inversion only use Zxy and Zyx, but 3-D use all the components) and the basic theory of the

TE

inversion algorithms (2-D inversion base on a lot of theoretical hypothesis and 2-D theory, but 3-D

CE
P

has relatively few assumptions), and the resolution of the 3-D MT results is adequate for oil and

AC

gas exploration.

6.2 Vertically fractured oil-bearing structure


To emphasize the role of 3-D MT, we design the extended vertical fracture model shown in Fig. 20.
This model also contains an anticlinal structure.

Fig. 20(a) and (b) show the same fundamental electrical structure as Fig. 20 (c) and Fig. 20 (d). Fig.
20(d) clearly show the shape of the arc and vertical interface, but the waveform structure only
reflected the interface of the horizontal layers, as shown in Fig. 20(e). The vertical fracture cannot
be seen in Fig. 20(e), but the possibility of a fault is suggested. Thus we cannot determine
17

ACCEPTED MANUSCRIPT
whether it may represent a fault trap or not using only the seismic waveform method. As 3-D
inversion can be used to interpret large real-world and has advantages over 2-D approaches in

IP

complex subsurface situations (Tietze and Ritter, 2013), and can reflect the information what

SC
R

other exploration methods cannot, we consider that 3-D MT can play an important role in oil
exploration.

NU

7. Conclusions

MA

In this paper we have assessed the new 3-D MT technique and the resolution of inversion method
with synthetic models for oil and gas exploration. Five 2-D and 3-D geophysical synthetic models

of typical reservoir structure are developed, in which realistic physical and electrical properties

CE
P

TE

are assigned for shale, (oil-bearing) limestone, gas, oil and groundwater layers.

By contrast 2-D and 3-D MT forward modeling and inversion and 2-D seismic wave modeling, we

AC

get the following conclusions:

(1) The comparison of 2-D and 3-D MT forward modeling using synthetic models reveals that 3-D
responses have better performance in the details of apparent resistivity and phase vaule and
other aspects.
(2) The comparison of 2-D and 3-D MT inversion methods using synthetic data reveals that the
3-D inversion models are closer to the synthetic models. Through the four examples of MT
inversion using synthetic data of anticline models, 3-D inversion results has advantages in the
details (specifically in every middle-low resistivity and main high resistivity layer) of the
position and resistivity of the structures. But it should be noted that the high resistivity layer
18

ACCEPTED MANUSCRIPT
in middle-high resistivity background is difficult to be resolved in both 2-D and 3-D MT
inversion method.

IP

(3) The comparison of the 3-D MT and 2-D seismic results using synthetic data of anticline and

SC
R

fracture (vertical interface) models reveals that 3-D MT can make up for the deficiency of the
seismic results such as the fake sync-phase axis and multiple waves. In reflective seismology
result, phenomenon of dislocation in one layer is often appeared in time-axis, and multiple

NU

wave problems are also serious. The vertical interface cannot be reflected in the seismic

MA

result either. On the other hand, 3-D MT inversion results can reveal both of horizontal and
vertical electrical interfaces without the fake abnormally in seismic result.

So we suggest the 3-D MT inversion technique has more advantages than previous non-seismic

TE

methods, comparing with 2-D MT and seismic reflection methods, The 3-D MT inversion method

CE
P

has important potential for oil exploration and should be utilized more frequently. Real field data

AC

for oil exploration will be tested in further researches.

Acknowledgments

This work is supported by Magnetotelluric static shift mechanism analysis and correction
methods study K1318, the National Natural Science Foundation of China under Grant 40930418,
and 3-D inversion of gravity and magnetic constrained by sparse prior information 41104061.

REFERENCES
Avdeev, D.B., Kuvshinov, A.V., Pankratov, O.V., Newman, G.A., Rudyak, B.V., 2000. Modeling
induction log responses in 3D geometries using a fast integral equation approach.
19

ACCEPTED MANUSCRIPT
Three-dimensional electromagnetics, 4: 55-63
Batzle, M., and Wang, Z., 1992, Seismic properties of pore fluids; Geo. 57, 1396-1408.

IP

Berdichevsky, M., Fox, L., Jakovlev, A., Bubnov, V., Kulikov, V., Pushkarev, P., 2010. Russian oil and

SC
R

gas geoelectric surveys case history-phoenix in Russia. Industry Applications of EM


Exploration Abstract, Session EM2-1 (in Russian).

Chave, A.D., Jones, A.G., 2012. Magnetotelluric Method Theory and Practice. New York:

NU

Cambridge Unversity Press, 1-83

Chen, L.S., Wang, G.E., 1990. The Magnetotelluric Method. Beijing Geological Publisher, Beijing,

MA

309 pp (in Chinese).

Chen, X.B., Zhao, G.Z., 2009. Automatic construction of a Site-Centered Grid (SCG) for 2D

TE

magnetotelluric inversion. Chinese Journal of Geophysics, 52(6): 1564-1572 (in Chinese)


Constable, S.C., Orange, A.S., Hoversten, G.M., Morrison, H.F., 1998. Marine magnetotelluric for

CE
P

petrolum exploration part 1: A sea-floor equipment system. Geophusics, 63(3): 816-825


He, Z.X., Jia, J.D., Gou, L., 2001. The role of non-seismic technology in oil and gas exploration and

AC

development. Petroleum exploration and development 28(4), 7072 (in Chinese).


Gasperokova, E., Newman, G.A., Feucht, D., Arnason, K., 2011. 3-D MT Characterization of two
Geothermal Fields in Iceland. GRC Transactions, 35, 1667-1671
Hersir, G.P., Arnason, K., Vilhjalmsson, A.M., 2013. 3-D inversion of magnetotelluric (MT)
resistivity data from Krysuvik high temperature geothermal area SW Iceland, Proceedings,
Thirty-Eighth Workshop on Geothermal Reservoir Engineering Stanford University,
California, 11-13
Hubert, J., 2012. From 2D to 3D models of electrical conductivity based upon magnetotelluric
data: Expreiencs from two Case Studies. Acta Universitatis Upsaliensis. Digital
20

ACCEPTED MANUSCRIPT
Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and
Technology 890. 55pp. Uppsala

SC
R

Memoirs of the Faculty of Engineering 63, 87106.

IP

Khalil, A., Ushijima, K., 2003. Magnetotelluric soundings in Minami-Noshiro Oil Field, Japan.

Li, A.Y., Liu, J.X., Zhu, C.S., 2012. The application of magnetotelluric sounding to oil and gas
exploration in Central Guangxi Depression. Geophysical Geochemical Exploration 36(1),

NU

812 (in Chinese).

MA

Liu, G.D., 2002. Building the next great wall: The second round of oil and gas exploration of China.
Progress in Geophysics 17(2), 185190 (in Chinese).

Liu, G.D., Zhu, L.L., 2003. Latest proceeding of petroleum geophysics. Progress in Geophysics

TE

18(3), 363-367 (in Chinese).

CE
P

Mackie, R.L., Rodi, W., Watts, D., 2001, 3-D magnetotelluric inversion for resource exploration,
SEG Int1 Exposition and Annual Meeting, San Antonio, 9-14

AC

Newman, G.A., Hoversten, M., Gasperikova, E., Wannamaker, P.E., 2005. 3-D Magnetotelluric
Characterization of the Coso Geothermal Field. Thertieth Workshop on Geothermal
Reservoir Engineering, Stanford University, California, 2
Ou, D.X., 2007. Effect of computer precision and grid length on MT simulating using FEM. Journal
of Guilin University of Technology, 27(3): 329-334 (in Chinese)
Rodi, W.L., Mackie, R.L., 2001. Nonlinear conjugate gradients algorithmfor 2 - D magnetotelluric
inversion. Geophysics 66(1), 174187.
Sun, T.W., Lu, J.F., Liu, Z.B., Fu, X.F., Fu, G., 2011. Hydrocarbon migration and enrichment features
of the Fuyu oil layer to the east of the Daqing placanticline. Petroleum exploration and
21

ACCEPTED MANUSCRIPT
development 38(6), 700-708 (in Chinese)
Sun, W.B., Song, Q.H., Zheng, L., He, Z.X., 2003. The development of MT sounding technique and

IP

its application to the oil and gas exploration. Geology and Prospecting, 39: 3-9 (in Chinese)

SC
R

Tietze, K., Ritter, O., 2013. 3D magnetotelluric inversion in practice the electrical conductivity
structure of the San Andreas Fault in Central California. Geophysical Journal International,
195(1): 130-147

NU

Tissot, B.P., Wilt, D.H. (author), Xu, Y.Y., Xu, L., Hao, S.S.(translator), 2003. Formation and

MA

distribution of oil, Beijing: Petroleum industry publishing house (in Chinese)


Wang, Z., 1988, Wave velocities in hudrocarbons and hydrocarbon saturated rocks-with

applications to EOR monitoring: Ph. D. Thesis, Stanford University.

TE

Wang, Z., Nur, A., and Batzle, M. L., 1988, Acoutic velocity in petroleum oils: SPE paper 18163,

CE
P

Proc. 63 rd SPETC Formation Eval. Res. Geol. Section, 571-585.


Wannamaker, P.E., Ward, S.H., Hohmann, G.W. 1984. Magnetotelluric responses of

AC

three-dimensional bodies in layered earth. Geophysics 49, 1517-1533


Wei, W.B., 2002. New advance and prospect of magnetotellutic sounding (MT) in China. Progress
in Geophysics 17(2), 245254 (in Chinese).
Wu, W., 2012. The application of MT sounding to oil and gas exploration in Arshili depression.
Technique of Neijiang 4, 102 (in Chinese).
Xia, X.C., Wang, J.X., Liu, J.C., Zhang, J.G., Zhang, L., Xu, X.X., 2012. The application of
magnetotelluric sounding to lithofacies division in oil and gas exploration of northern
Nanpangjiang area. Geophysical Geochemical Exploration 36(2), 174179 (in Chinese).
Xiao, Q.B., Cai, X.P., Xu,vX.W., Liang, G.H., Zhang, B.L., 2010. Application of the 3D magnetotelluric
22

ACCEPTED MANUSCRIPT
inversion code in a geologically complex area. Geophysical Prospecting, 58: 1177-1192
Zhang, K., Wei, W.B., Ye, G.F., 2008. The actualization of finite-element magnetotelluric

IP

forwarding on Matlab. Seismological and Geomagnetic Observation and Research 29(5),

SC
R

83-88 (in Chinese)

Zhang, K., 2012. The nonlinear conjugate gradient 3-D magnetotelluric inversion program. State
Intellectual Property Office of the P.R.C: 2012SR057553

NU

Zhang, K., 2013. The research of improved 3-D nonlinear conjugate gradient inversion of

MA

magnetotelluric and parallel computing. China University of Geosciences (Beijing) (in


Chinses)

Zhang, K., Dong, H., Yan, J.Y., Lu, Q.T., Wei, W.B., He, Y.X., 2013. A NLCG inversion method of

CE
P

(in Chinese)

TE

magnetotellurics with parallel structure, Chinese journal of geophysics, 56(11), 3922-3931

Zhdanov, M.S., Wan, L., Gribenko, A., Cuma, M., Key, K., Constable, S., 2011. Large-scale 3-D

AC

inversion of marine magnetotelluric data: Case study from the Gemini prospect, Gulf of
Mexico, Geophysics 76(1), F77-F87
http://en.wikipedia.org/wiki/Magnetotellurics

Appendix
Table list
Table 1 Petrophysical properties of materials in oil-bearing structures

Property

Resistivity,

Density,
23

S-wave

P-wave

ACCEPTED MANUSCRIPT
3

velocity

velocity

VS (m/s)

VP (m/s)
13303970

101000

2.02.50

7802300

Limestone

1010 000

2.3

14503500

Oil

10 000

2.93

Gas

100 000 000

0.70.8

Groundwater

1100

11.02

SC
R

Shale

( 10 kg/m )

IP

(m)

25006000
2600

1400

2350

NU

1400

1450

2700

available

from

MA

(Wang, 1988; Wang et al., 1988; Batzle and Wang, 1992; Chave and Jones, 2012; and some data
http://wenku.baidu.com/view/2bff294ccf84b9d528ea7aea.html

TE

http://www.docin.com/p-52175997.html

AC

CE
P

Table 2 Synthetic material properties used in model

Resistivity,

P-wave

velocity

velocity

VS (m/s)

VP (m/s)

Density,

Property

(m)

S-wave

(10 kg/m )

Shale

30

2.0

1500

2000

Limestone

100

2.3

2000

3500

Oil

10 000

2.93

1400

2600

Gas

100 000 000

0.8

1400

2350

Groundwater

10

1.01

1450

2700

24

and

ACCEPTED MANUSCRIPT

NU

AC

CE
P

TE

MA

Fig. 1 Typical oil-bearing structure model

SC
R

IP

Figure list

Fig. 2 Synthetic geophysical model (a): 3-D model; (b): central section of 3-D model as 2-D model.
1: shale; 2: limestone; 3: gas; 4: oil; 5: groundwater

25

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

Fig. 3 2-D forwarding model responses for central profile, showing apparent resistivity and

AC

CE
P

TE

impedance phases for TE (xy polarization, left) and TM (yx polarization) modes

26

CE
P

TE

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

AC

Fig. 4 3-D forward modeling responses (apparent resistivity and impedance phase)

Fig. 5 Comparison of pseudo-sections for 3-D (left) and 2-D model results

27

SC
R

IP

ACCEPTED MANUSCRIPT

AC

CE
P

TE

MA

NU

Fig. 6 Comparison between 2-D sounding for synthetic and inversion model

Fig. 7 3-D sounding comparison of synthetic and inversion model (central profile)
28

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

TE

Fig. 8 MT inversion model results

AC

CE
P

Synthetic model 1: shale; 2: limestone; 3: gas; 4: oil; 5: groundwater

29

TE

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

AC

CE
P

Fig. 9 Comparison of synthetic and inversion soundings at 1.33 Hz

Fig. 10 Comparison of yz cross-sections (a): synthetic model; (b): inversion model

30

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

AC

CE
P

TE

Fig. 11 Slices through 3-D inversion model for same data as Fig. 10

Fig. 12 Comparison between 3-D MT inversion and 2-D waveform forwarding for central section
of synthetic model (a): MT inversion model; (b): Seismic forwarding model
Synthetic model 1: shale; 2: limestone; 3: gas; 4: oil; 5: groundwater
31

NU

SC
R

IP

ACCEPTED MANUSCRIPT

CE
P

TE

MA

Fig. 13 Extended model 1 with thick oil layer. Materials are: #1 shale; #2 limestone; #3 oil

AC

Fig. 14 Extended model 2 with thick water layer. Materials are: #1 shale; #2 limestone; #3 oil; #4
groundwater

Fig. 15 Extended model 3 with complex layer. Materials are: #1 shale; #2 limestone; #4
groundwater; #5 complex layer
32

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

TE

Fig. 16 Dimensions of extended MT inversion Model 1. Materials are: #1 shale; #2 limestone; #3

AC

CE
P

oil

Fig. 17 Dimensions of extended MT inversion Model 2. Materials are: #1 shale; #2 limestone; #3


oil; #4 groundwater
33

MA

NU

SC
R

IP

ACCEPTED MANUSCRIPT

AC

CE
P

TE

groundwater; #5 complex layer

Fig. 18 Dimensions of extended MT inversion Model 3. Materials are: #1 shale; #2 limestone; #4

34

ACCEPTED MANUSCRIPT

MA

NU

SC
R

IP

Fig. 19 Resistivity results for extended model 1-3

TE

Fig. 20 Comparison of 3-D MT and 2-D seismic waveform

CE
P

1: shale (VS = 1500 m/s; VP = 2000 m/s); 2 and 4: limestone ( VS = 2000 m/s; VP = 3500 m/s);

AC

3: oil (VS = 1400 m/s; VP = 2600 m/s)

35

ACCEPTED MANUSCRIPT

AC

CE
P

TE

MA

NU

SC
R

IP

Highlights
We make some electrical and velocity synthetic models of typical oil-bearing structure.
We compare 2D and 3D MT forwarding and inversion and 2D seismic forwarding results.
3D MT can make up the shortfall of seismic.
We consider 3D MT inversion technique has important potential for oil exploration.

36