See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/289545316

Simulation of 3D gravity anomaly of thin coal layer in

sedimentary environment and its delineation

Article in International Journal of Tomography and Simulation · January 2014

CITATIONS READS

3 3,541

1 author:

Hendra Grandis
Bandung Institute of Technology
71 PUBLICATIONS 366 CITATIONS

SEE PROFILE

All content following this page was uploaded by Hendra Grandis on 12 January 2016.

The user has requested enhancement of the downloaded file.

International Journal of Tomography and Simulation [ISSN 2319-3336];
Year: 2014, Volume: 26, Issue Number: 2;
[Formerly known as “International Journal of Tomography & Statistics”
(ISSN 0972-9976; 0973-7294)];
Copyright © 2014 by CESER PUBLICATIONS

Simulation of 3D Gravity Anomaly of Thin Coal Layer

in Sedimentary Environment and its Delineation

Hendra Grandis

Faculty of Mining and Petroleum Engineering, Institut Teknologi Bandung,

Jalan Ganesha 10 Bandung (Indonesia)
Email : [email protected], [email protected]

ABSTRACT

We investigate the possibility of application of the gravity method to delineate a typical

coal-bed methane (CBM) bearing formation. We simulate gravity data corresponding to a
realistic 3D gravity model associated with a thin coal layer in a sedimentary environment.
The separate and combined gravity effects of the regional component and the coal layer
were calculated by using gravity forward modeling with FFT algorithm. Several filtering and
anomaly enhancement techniques were applied to the simulated gravity anomaly to
delineate the presence of the coal layer. Only lateral extent of the anomalous sources may
be detected. Difficulties in extracting detailed parameters of the anomalous sources are
mainly related to the anomaly magnitude that is very small.

Keywords: Coal, CBM, gravity anomaly, filtering, enhancement, trend surface fitting.

Mathematics Subject Classification: 86A20

1. INTRODUCTION

The ever increasing energy demand and the decline of oil and gas resources have led to interests in
exploration for coal-bed methane (CBM) as one of unconventional gas resources. Being deposited in
a sedimentary environment as also the case for oil and gas bearing formations, the seismic method
remains the primary method for exploring CBM. However, for preliminary and regional scale of
exploration, it is rationally to implement less expensive geophysical exploration techniques before
seismic detailing. The use of gravity method for basin delineation at a regional scale and particularly
in hydrocarbon exploration is well-known (e.g. Chapin and Ander, 1999; Oruç et al., 2013). The
gravity method may also be useful for prospecting of coal-bed methane at the same exploration stage.
However, it is interresting and challenging whether gravity method can also be employed not only for
regional scale but also at sub-regional and even at prospect scale (Murphy and Brewster, 2007;
Badmus et al., 2011).

In this paper we simulate both simple and also more realistic 3D gravity models that represent the
presence of a low density thin coal layer in a sedimentary environment. We used 3D gravity forward
modeling calculation to obtain the theoretical response of the simulated subsurface model. Several
enhancement and filtering techniques commonly performed in potential-field (gravity and magnetic)
data analysis were applied to the simulated gravity anomaly. They include spectral analysis based

www.ceser.in/ijts.html
www.ceserp.com/cp-jour
www.ceserpublications.com
 International Journal of Tomography and Simulation [ISSN 2319-3336]

filtering and conventional regional-residual anomaly separation using surface fitting. Only part of
information on the residual anomaly associated with the anomalous target can be obtained, i.e. the
lateral delineation of the coal spots.

2. SIMULATION USING SIMPLE BLOCK MODEL

The coal layer or CBM bearing formation generally exists in a sedimentary basin environment. The
regional gravity anomaly in such geological setting may be dominated by the basement depth
variation, i.e. deepening of the basement that generates low gravity anomaly. The coal layer is also
associated with low gravity anomaly, since it usually has lower density than its environment. Both low
anomalies due to basement and coal layer may have interference effect resulting in difficulties in
differentiating one from the other. To simulate that situation, a 3-D gravity forward modeling is
performed by calculating the gravity effect of a right rectangular prism representing basement low and
coal layer separately and in combination.

In 3-D gravity forward modeling, a vertical rectangular prism is used to represent an anomalous body
with a constant density contrast relative to the surrounding or regional environment. More complex
model is formed by a superposition of a number of prisms. The practical formula for calculation of
gravitational attraction of a vertical rectangular prism was proposed by Nagy (1966), Plouff (1975) and
implemented as a Fortran subroutine GBOX in Blakely (1995). The gravity attraction at an observation
point at the origin O(0, 0, 0) due to a vertical prism with a unit density is given by (Blakely, 1995),

A G ª
«¬
> > x ln ( y  ( x 2
 y 2  z 2 )1 / 2  y ln ( x  ( x 2  y 2  z 2 )1 / 2
(1)
@ xx12 @ y1 º»
y2 z2
 z arctan ( z ( x 2  y 2  z 2 )1 / 2 x 1 y 1 .
¼ z1

where G is the gravitational constant. The edges of the prism are parallel to the reference axes and x1,
x2, y1, y2, z1, z2 are limiting coordinates for its volume in x, y and z coordinates respectively.

For i-th observation point at Si(ui, vi, wi) and j-th prism then expression of A in equation (1) is an
element of the kernel matrix [Aij] with all limiting coordinates of the prisms (x, y, z)j are relative to (ui, vi,
wi). The gravity anomaly 'gi at the i-th station Si, i = 1, 2, ..., N produced by a model consisting of M
prisms with densities mj, j = 1, 2, ..., M is

M
'g i ¦ Aij m j . (2)
j 1

In a matrix and vector multiplication form, equation (2) becomes

g A m. (3)

A block with 500 m thick and 20 km by 20 km lateral extension at 2000 m depth was set and a
negative density contrast of -0.2 gram/cm3 was assigned to the block to simulate a basement low
(Figure 1a). The negative density contrast represents the difference between the upper layer density

2
 International Journal of Tomography and Simulation [ISSN 2319-3336]

of 2.5 gram/cm3 relative to 2.7 gram/cm3 for the basement. The domain of calculation was set to an
area of 40 km by 40 km. A negative anomaly with a maximum amplitude of -3.36 mGal is observed
(Figure 1b). A slightly higher or lower amplitude of negative anomaly would be observed for the
density contrast varying around -0.2 gram/cm3. A similar simulation was done for coal layer with 100
m thick and 10 km by 10 km lateral extension at 500 m depth (Figure 2a). A negative density contrast
of -0.5 gram/cm3 was assigned to the block, representing 1.7 gram/cm3 coal relative to 2.2 gram/cm3
for the surrounding layer. A negative anomaly with a maximum amplitude of -1.74 mGal is observed
(Figure 2b).

(a) (b)

Figure 1. Simple block model with a negative density contrast (-0.2 gram/cm3)
representing a basement low (a) and its calculated gravity anomaly (b).

(a) (b)

Figure 2. Simple block model with a negative density contrast (-0.5 gram/cm3)
representing a thin coal layer (a) and its calculated gravity anomaly (b).

The regional anomaly is gravity response of the deeper layer (basement) and also upper layers
considered as the "background" relative to more isolated anomalies associated with coal. From these
simulations, the anomaly from coal is only around 50% of the anomaly related to basement low.
Therefore, the anomaly from coal may be masked by the anomaly associated with the basement

3
 International Journal of Tomography and Simulation [ISSN 2319-3336]

depth variation. Nevertheless, the lateral extent of the basement and coal are usually different in scale.
In gravity the total anomaly is simply the sum of individual anomalies. The profile of gravity anomalies
of individual sources and their superposition is shown in Figure 3. It indicates that the spatial
frequency (or spatial wavelength) of anomalies from basement undulation and coal might be different.
In such case, filtering of the gravity data to separate different frequencies might enhance the
contribution of either the basement (low frequency) or coal (high frequency). The attempt of filtering
gravity data to enhance or isolate anomalies related to coal will be performed by using simulated
gravity data with more realistic regional model.

 1.0 
0.0
0.0

Gravity Anomaly (mGal)

Gravity Anomaly (mGal)

-1.0 -2.0

-2.0

-4.0

-3.0

-6.0
-4.0
0 10000 20000 30000 40000
0 10000 20000 30000 40000
X (meter)
X (meter)

(a) (b)

Figure 3. Separate basement low (red) and coal (blue) related gravity anomalies (a) and
their combined gravity anomaly (b).

3. SIMULATION USING REALISTIC MODEL

We used available real seismic data to construct more realistic regional environment for the coal-bed
methane bearing formation. The seismic data come from an area typical of CBM prospects. However,
the exact location of the area is irrelevant so that the coordinates of all subsequent maps are set as
arbitrary. Figure 4 shows the seismic section with interpreted horizons and densities of formations
estimated from well-log data. The sedimentary layers overlying the basement are simply identified as
layer-1 to layer-7. Isolated coal spots with different thicknesses, i.e. 20, 50 and 100 m were simulated
to exist within layer-3 at locations where the depth reaches approximately between 500 and 800 m.
Figure 5 shows the outlines of the coal spots and also contours of the basement depth with 250 m
interval. The basement depth variation is supposed to dominate the regional component of the gravity
anomaly in this area. The top of the coal layer coincides with the bottom of the layer-2 (see Figure 6).
3
The density of 1.7 gram/cm is assumed for the coal layer, such that the density contrast with its
surrounding, i.e. the layer-3 (2.16 gram/cm3), is -0.54 gram/cm3.

Horizons of sedimentary layers and basement were digitized at 250 by 250 m grids. The forward
modeling calculation to obtain the gravity anomaly associated with the model was performed in an
area with 50 by 50 km in size. We used Parker's (1973) formula to calculate the gravity anomaly
caused by an uneven, uniform layer of material by means of a series of Fourier transforms. The

4
 International Journal of Tomography and Simulation [ISSN 2319-3336]

relationship between the vertical gravity effect 'g and its causative mass topography h(r) in the
Fourier domain is given by (Blakely, 1995),
f
| k |n1
F ('g )  2S G U exp(  | k | z0 ) ¦ n!
F (h n (r )) (4)
n 1

where r denotes position in the x-y plane, k is the 2-D wave number, U is density. Equation (4) is
convergent in the whole k-plane for z0 > 0, where z0 is the mean depth of the layer or density interface.
The Fourier Transformation F(.) has to be performed in 2-D since the data are also 2-D, i.e. h(r) is the
depth of the density interface (positive downwards) as function of position (x, y). Equation (4) provides
an efficient way to calculate the gravity effect caused by a subsurface mass anomaly or topography.
Furthermore, it can be easily applied to multi-layer case as well.

NW SE
Surface

500

1000 1.50
2.00
2.16
1500
2.33
2.42
2000 2.35
2.40
Basement
2500 2.70

3000 10 km

Figure 4. Seismic section showing horizons of sedimentary layers from surface to

basement with density colour scale given in gram/cm3 and depth scale in meter.

(a) (b)

5
 International Journal of Tomography and Simulation [ISSN 2319-3336]

Figure 5. Outline of isolated spots of coal layer and the profile line along South-West to
North-East (a) and simplified depth contour of the basement (b).
surface
0
Layer-1

Layer-2

Layer-3
coal 1
coal 2

-1000 Layer-4
coal 3
depth (meters)

Layer-5

Layer-6

Layer-7

-2000
Basement

-3000
0 10000 20000 30000 40000 50000 60000 70000
distance (meters)

Figure 6. Profile along South-West to North-East direction (see Figure 5a for the position
of the profile on map) showing the layer interfaces from surface to basement.

The gravity anomaly maps associated with the realistic model without and with coal layer are shown in
Figure 7a and Figure 7b respectively. In both maps, we can observe the North – South direction of the
regional trend with the lower anomaly at the East reflecting the deepening of the basement. Having
almost the same range in gravity anomaly, i.e. from 438.7 to 469.7 mGal, the difference between
these two gravity anomaly maps is hardly noticeable. This is related to the fact that the effect of the
coal layer is very small due to their geometry (very thin with limited lateral extent) and the low density
contrast (-0.54 gram/cm3).

(a) (b)

Figure 7. Gravity anomaly maps associated with the realistic model without coal layer (a)
and with coal layer (b).

6
 International Journal of Tomography and Simulation [ISSN 2319-3336]

4. SPECTRAL ANALYSIS AND FILTERING

The gravity anomaly associated with sources from shallow depth typically has high frequency (short
wavelength) while the anomaly from deeper sources has low frequency (long wavelength). These
characteristics can be used to separate the gravity data into regional and residual anomalies. In the
spectral analysis, data are transformed to spatial frequency domain by using 2-D FFT algorithm. The
power spectrum is the squared magnitude of the complex coefficients that result from the FFT
operation. To normalize and to decrease the bandwidth of the power spectrum, the logarithm is
commonly applied to the magnitude. The power spectrum is a 2-D function relative to wave-number
and direction or azimuth. The radially averaged power spectrum is a function of wave-number only
and is calculated by averaging the energy at all direction for the same wave-number. The spectral
analysis can then be used to filter and enhance the gravity data to facilitate interpretation of the
gravity data.

The radially averaged power spectrum from synthetic gravity data associated with model containing
coal layer (see Figure 7b) is presented in Figure 8. The slope of the line segments in the power
spectrum is related to the depth of the anomalous sources. The depth estimates in Figure 8 were
obtained by calculating the slope of the spectrum at segments consisting of 5 point samples. It is
shown that anomalies with wavelength less than 2 km (wave-number = 0.5/1000 m) are associated
with sources at 1 km depth or less.

By taking into account the power spectrum, a filter can be designed to attenuate low frequencies
related to regional structures and to enhance the higher frequencies associated with local anomalies.
In this case, local or residual anomalies are due to the coal layer. The result of Cosine Roll-off filtering,
i.e. the residual anomaly, of simulated gravity data containing a coal layer (Figure 7b) is shown in
Figure 9a. Although the magnitude of the residual anomaly is extremely small, the high frequency
components of the anomaly are enhanced and coincide with the outline of anomalies related with the
coal layer. The regional component from this type of filtering is presented in Figure 9b. The similarity
in pattern and magnitude with the total gravity anomaly shows the dominant effect of the regional
structures compared to the effect of the coal layer.

Anomaly enhancement or filtering techniques based on spectral analysis are promising although their
function may be limited only to define laterally the interesting spots. For a preliminary or
reconnaissance survey covering a large area, the filtering technique based on the spectral analysis
may lead to a more specific target sub-region for further studies. Both spectral analysis and filtering
were performed by using Oasis Montaj software from Geosoft Inc. (Geosoft, 2005).

7
 International Journal of Tomography and Simulation [ISSN 2319-3336]

Figure 8. Radially averaged power spectrum from synthetic gravity data associated with
model containing a coal layer (see Figure 7b). The red dashed line represent
the transition from lower to higher frequencies associated with regional and
residual (or local) anomalies respectively.

(a) (b)

Figure 9. Residual anomaly from Cosine Roll-off filtering (a) and the regional anomaly
component (b).

8
 International Journal of Tomography and Simulation [ISSN 2319-3336]

5. SURFACE FITTING

The residual anomaly associated with the anomalous source of interest can be obtained by
substracting the regional component from the total or Bouguer gravity anomaly. The regional anomaly
component is not known and can only be estimated. In the so-called trend-surface analysis or surface
fitting, the regional anomaly is estimated by fitting a (usually low order) polynomial surface in spatial
coordinates (x and y) to the gravity anomaly. The generic polynomial equation of N-th order
representing the regional anomaly can be expressed as,

N m
g R ( x, y ) ¦ ¦ aij x m n y n . (5)
m 0 n 0

Fitting equation (5) of a particular order to the gravity data in a least-squares sense is in fact a simple
linear inverse problem resolution (Menke, 2012). Several generic surface plotting softwares include
such surface fitting in their algorithm.

By observing Figure 7, we assume that the regional anomaly is dominated by the basement and can
be represented by a first order polinomial surface. This is justified by the fact that the trend of both
basement depth and gravity anomaly coincides as can be seen from their first order polinomial
surface fits (Figure 10a). The comparison between the gravity anomaly and its first order trend surface
fit as the regional anomaly is shown in Figure 10b.

Figure 11a presents the residual anomaly, i.e. the result of substraction of the first order polinomial
surface fit from the total gravity anomaly. It appears that only the thickest coal spot can be identified
from the residual anomaly map. Other coal spots are less obvious due to their lower thickness and
hence lower anomaly magnitude. High residual anomalies in the North-West and in the South are the
result of innacurate representation of the planar regional anomaly for that area. For comparison, the
residual anomaly obtained by substracting the anomaly in Figure 7a from the anomaly in Figure 7b is
presented in Figure 11b. The form of lowest anomalies coincides almost exactly with the outline of the
coal layer, while their amplitudes are proportional to the coal thicknesses (20, 50 and 100 m from
South-West to North-East). The simulation presented in this section shows that we can obtain
meaningful information on the target from the residual anomaly, provided that the regional anomaly is
known with high accuracy. In the ideal case, the regional anomaly is exactly known, i.e. gravity
anomaly of the sedimentary layers without coal (Figure 7a).

From the surface fitting, the residual anomaly amplitude of the thickest coal layer is still much higher
(-7.3 mGal) than the real value (-1.5 mGal). This fact emphasizes the fundamental limitations of the
surface fitting technique for regional-residual anomaly separation. The choice of the correct order of
the polynomial surface is difficult. For our simulated case, higher order polynomial surface will result in
better-fitting the gravity anomaly such that the magnitude of the residual anomaly would be lower.
However, over-fitting the gravity anomaly will result in diminishing the residual anomaly. There is also
a problem with the choice of the correct constant level of the regional. The zeroth order constant is
simply the mean value of the gravity anomaly while in principle there is still uncertainty of such
constant in repesenting the correct regional anomaly.

9
 International Journal of Tomography and Simulation [ISSN 2319-3336]

(a) (b)

Figure 10. (a) Coincidence of first order polinomial surface fit of basement depth (blue line
in meter) and gravity anomaly (red line in miliGal) and (b) Gravity anomaly and
its regional component from first order polinomial surface fit.

(a) (b)

Figure 11. Residual anomaly from trend surface analysis using first order polinomial fit for
the regional anomaly (a) and from exact regional anomaly substraction (b).

10
 International Journal of Tomography and Simulation [ISSN 2319-3336]

6. CONCLUSION

The simulation presented in this paper shows that identifying signatures of a coal layer in a
sedimentary environment by using gravity data is a difficult task. The gravity anomaly associated with
the presence of the coal layer is typically very small and is almost at the same order of the anomaly
caused by the regional element, i.e. depth variation of the basement. However, the gravity anomaly
produced by superficial or shallow coal layer and deeper basement are typically different in their
spatial extent, regardless their relative amplitudes. Enhancement of residual anomaly can be obtained
by using the spectral analysis and filtering, alhough the resulting residual anomaly can only delineate
the lateral extent of the target.

Further anomaly enhancement using trend-surface analysis or surface fitting may lead to residual
anomaly and its magnitude (or at least their relative magnitude) as well. For that purpose, the trend-
surface analysis should be performed with great cautions and also by using accurate and well defined
input or observed gravity data. A combination of techniques and their strategy of application
discussed in this paper may result in significant improvement of the gravity method capability in
absence of other equivalent method (in terms of availability, cost, practical aspect etc).

7. REFERENCES

Badmus, B.S., Sotona, N.K., Krieger, M., 2011, Gravity support for hydrocarbon exploration at the
prospect level, Journal of Emerging Trends in Engineering and Applied Sciences (JETEAS), 2, 1-6.

Blakely, R.J., 1995, Potential Theory in Gravity and Magnetic Applications, Cambridge University
Press.

Chapin, D.A., Ander, M.E., 1999, Applying gravity to petroleum exploration: Chapter 15 in Beaumont,
E.A., Foster, N.H., eds., American Association of Petroleum Geologists Treatise of Petroleum
Geology Handbook, Exploring for Oil and Gas Traps.

Geosoft Inc., 2005, Montaj MAGMAP Filtering Tutorial, www.geosoft.com.

Menke, W., 2012, Geophysical Data Analysis: Discrete Inverse Theory, 3rd Edition, Academic Press,
London.

Murphy, C.A., Brewster, J., 2007, Target delineation using full tensor gravity gradiometry data, ASEG
Extented abstracts.

Nagy, D., 1966, Gravitational attraction of a right rectangular prism, Geophysics, 31, 362-371.

Oruç, B., Sertçelik, I., Kafadar, Ö., Selim, H.H., 2013, Structural interpretation of the Erzurum Basin,
eastern Turkey, using curvature gravity gradient tensor and gravity inversion of basement relief,
Journal of Applied Geophysics, 88, 105-113.

Parker, R.L., 1973, The rapid calculation of potential anomalies, Geophys. J. Royal Astr. Soc., 31,
447-455.

Plouff, D., 1975, Derivation of formulas and FORTRAN programs to compute gravity anomalies of
prisms, National Technical Information Service No. PB-243-526, U.S. Department of Commerce.

11

View publication stats