Accepted Manuscript

GCH_gravinv: A MATLAB-based program for inverting gravity anomalies over

sedimentary basins

Luan Thanh Pham, Erdinc Oksum, Thanh Duc Do

PII: S0098-3004(18)30173-0
DOI: 10.1016/j.cageo.2018.07.009
Reference: CAGEO 4161

To appear in: Computers and Geosciences

Received Date: 23 February 2018

Revised Date: 24 July 2018
Accepted Date: 29 July 2018

Please cite this article as: Pham, L.T., Oksum, E., Do, T.D., GCH_gravinv: A MATLAB-based program
for inverting gravity anomalies over sedimentary basins, Computers and Geosciences (2018), doi:
10.1016/j.cageo.2018.07.009.

This is a PDF file 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 final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
1

2 GCH_gravinv: a MATLAB-based program for inverting gravity anomalies over

3 sedimentary basins.

4 Luan Thanh Phama, Erdinc Oksumb,*, Thanh Duc Doa

a
5 VNU University of Science, Faculty of Physics, Department of Geophysics, 334 Nguyen Trai, Thanh

PT
6 Xuan, Ha Noi, Vietnam.
b
7 Süleyman Demirel University, Engineering Faculty, Dep.of Geophyisical Eng., Isparta, Turkey.
8

RI
9 *Corresponding author. Tel.: +902462111355 E-mail adress: eroksum@gmail.com

10 Abstract

SC
11 This paper presents a Matlab-based program GCH_gravinv including an easy-to-use graphical
12 user interface (GUI) for determining the depth to the basement of a sedimentary basin derived

U
13 from inverting its gravity anomalies by an iterative procedure. The developed code uses an
AN
14 advanced iterative rapid algorithm based on the combination of the FFT-based algorithm of
15 Granser and the space domain technique of Cordell and Henderson. As an advantage, this
16 combination performs a fast computation with high precision and the inversion scheme does
M

17 not require a mean depth or low-pass filtering. Given the required density contrast varying
18 exponentially with depth and a pre-assigned criterion for the termination of the iterative
D

19 procedure, the GUI allows the user an instant view of the convergence between the observed
TE

20 and inverted anomalies. The feasibility of the proposed algorithm is demonstrated on

21 synthetic data from a 3D model where the obtained results coincide well with the actual
EP

22 depths. The inversion of a real gravity anomaly over Chintalapudi subbasin (India) is
23 presented to compute the basement depths as a practical example. The obtained results from
24 the real data application are in agreement with published information available for the study
C

25 area.
AC

26 Keywords: gravity inversion, matlab, basement depth, sedimentary basin, Chintalapudi.i

27
28
29
30

i
Luan Thanh Pham contributed to establish the basis of the algorithm.
Erdinc Oksum contributed to the creation of the main code of the program linked to the algorithm.
Thanh Duc Do contributed to the article writing and to its scientific development.

1
 ACCEPTED MANUSCRIPT
31 1. Introduction
32 The aim of this paper is to determine the configuration of a sedimentary basin from the

33 observed gravity anomalies. Most existing algorithms employ either the polygonal model of

34 Talwani et al. (1959) or the stacked rectangular prism model of Bott (1960) to solve this

35 problem. The algorithms developed by Cordell and Henderson (1968); Murthy and Rao

PT
36 (1989); Barbosa et al. (1997); Barbosa et al. (1999); Rao et al. (1990); Mendonca (2004);

37 Pallero et al. (2015) are based on the assumption that the density of sedimentary rocks above

RI
38 the basement interface is uniform. However, the density contrast between the sediments and

SC
39 the basement is rarely uniform in nature, so this assumption is often unrealistic. It is well

40 known that the density of the sediments increases progressively with depth of burial. Several

41
U
other authors have presented different algorithms to compute the depth to basement using
AN
42 variable density contrast functions. Murthy and Rao (1979) used a linear density function to

43 model the gravity anomalies of 2D structures. Chakravarthi and Sundararajan (2005, 2007);
M

44 Silva et al. (2014); Silva and Santos (2017); developed methods using a parabolic density
D

45 function in gravity modeling of 2.5D and 3D sedimentary basins. Rao (1986, 1990); Rao et al.
TE

46 (1990) simulated the decrease in density contrast of sedimentary rocks with depth by a

47 quadratic function. Litinsky (1989); Silva et al. (2006, 2010) used a hyperbolic density
EP

48 function. Zhou (2009, 2013) developed interpretation methods using density contrast

49 variations in both horizontal and vertical directions. Cordell (1973); Chai and Hinze (1988);
C

50 Chakravarthi et al. (2013a, 2013b, 2016) assumed an exponential variation in density contrast
AC

51 of the sediments. According to these algorithms, the gravity anomaly for each rectangular

52 prism (or each side of the polygon) is calculated and then, the total gravitational field is

53 determined by adding the anomaly of all prisms (or all sides of the polygon). Therefore, the

54 computer time required to perform all the computations is large leading to slow inversion

55 methods.

2
 ACCEPTED MANUSCRIPT
56 To solve this problem, Granser (1987a) developed a method for the rapid calculation of the

57 gravity anomalies of sedimentary basins, based on the FFT algorithm. This method is similar

58 to the one introduced by Parker (1972). According to Parker, using the FFT-based algorithm,

59 the computation time for a model grid containing N points is proportional to N ln(N). For the

60 same type of the model, using the space domain techniques, the computation time is

PT
61 proportional to N2. This difference is not so important for 2D and small 3D models. However,

RI
62 as the number of input and output points increases, the ratio of the computation time in the

63 space-domain approach compared to the computation time in the FFT-based approach

SC
64 increases rapidly.

U
65 Granser (1987a) rearranged the forward algorithm to obtain the depth to the basement of
AN
66 sedimentary basins. This method is similar to Oldenburg’s (1974) inversion procedure.

67 Granser (1987b) also applied another frequency inversion method based on the Schmidt-
M

68 Lichtenstein nonlinear approach for the case of a 3D sedimentary basin with uniform density

69 contrast. Based on the FFT-based algorithm, several authors developed computer programs to
D

70 determine the geometry of the density interface from the gravity anomaly (eg. Gomez–Ortiz
TE

71 and Agarwal, 2005; Nagendra et al., 1996a, 1996b; Shin et al., 2006). Although the FFT-

72 based algorithm reduces the computation time, it requires a given mean depth of the interface
EP

73 and a low-pass filter to achieve convergence of the inversion procedure (Granser, 1987b;
C

74 Oldenburg, 1974).
AC

75 On the other hand, Cordell and Henderson (1968) introduced another 3D inversion method

76 using the space-domain technique despite the disadvantages of the computational speed due to

77 the use of the stacked rectangular prism model. Cordell and Henderson’s approach is based on

78 the Bouguer slab formula where an initial guess is made for the bottom depth of each prism

79 by assuming each prism is a slab infinite in all horizontal dimensions. This method has been

3
 ACCEPTED MANUSCRIPT
80 implemented in the Fortran programs that have been published by Cordell (1970); Cordell et

81 al., (1992); Phillips (1997).

82 In the present paper, we combine both Granser (1987a) and Cordell & Henderson (1968)

83 methods in a model to solve their limitations. The proposed algorithm is designed in Matlab

84 environment (version R2013b) with an easy-to-use graphical interface (GUI:

PT
85 GCH_gravinv.m) allowing the user the control of input parameters and thereafter displaying

RI
86 the results of the convergence between observed and calculated anomalies during the ongoing

87 process. The GUI also provides additional options for the visualization of the output data in

SC
88 2D maps and cross-sectional view. The accuracy and effectiveness of the developed code is

U
89 demonstrated on a synthetically produced gravity dataset of a 3D basin model. For a real data
AN
90 case, we also apply the method to determine the basement surface of the Chintalapudi

91 subbasin in India.
M

92 2. Theory
93 We consider a sedimentary basin model with depth to the basement h(x, y) The decrease of
D

94 the density contrast with the depth in the sedimentary basin could be approximated by an
TE

95 exponential function ∆ρ = ∆ߩ଴ ݁ ିఒ௭ (where z is positive vertically downward in units of

EP

96 length, ߂ߩ଴ is the density contrast observed at the ground surface, and ߣ is the decay

97 constant).
C

98 Establishing a three-dimensional Cartesian coordinate system with positive z-axis pointing

AC

99 downwards and horizontal x-axis and y-axis pointing eastwards and northwards respectively,

100 the gravity anomaly due to the sedimentary basin is given by (Granser, 1987a):

  ( −k) 
n
∞
1
101 
∆g = F 2πγ∆ρ0
−1
 F [1 − e ] − ∑
− λ h
F e h   
−λh n
(1)
 k +λ  n =1 n ! 
  

4
 ACCEPTED MANUSCRIPT
102 where F[ ] denotes the Fourier transform, F-1[ ] denotes the inverse Fourier transform, ߛ is the

103 universal gravitational constant, k is the wavenumber.

104 In the case of the gravity effect of an infinite horizontal slab (Bouguer slab) with a top at

105 surface and a slab bottom at a depth h, Equation (1) is reduced to (Granser, 1987a):

PT
2πγ∆ρ0
106 ∆g = (1 − e − λ h ) (2)
λ

RI
107 Equation (1) has meaning only when the Fourier series converges for any reasonable

SC
108 topographic relief function h(x, y) and, moreover, rapidity of convergence is really important

109 in practical computation. To improve the convergence speed of the forward algorithm in

U
110 Equation (1), Granser (1987a) suggested using the mean depth of the basement interface as a
AN
111 reference level z0 in the sense that the depth of the interface is described by h = z0 + ∆h.

112 Following Granser (1987a), the use of a reference level z0 requires an additional Bouguer slab
M

113 term, and performing an upward continuation from the level z0 to the surface. Thus, Equation
D

114 (1) becomes (Granser, 1987a):

TE

2πγ∆ρ0
115 ∆g =
λ
(1 − e ) + 2πγ∆ρ e
− λ z0
0
− λ z0
EP

 − k z0  ( −k) 
n
∞
e ( − λ∆h )
116 ×F −1
 F [1 − e ]− ∑ F e − λ∆h ∆h n    (3)
 k +λ  n =1 n! 
  
C
AC

117 Here, the e − k z0 term represents the upward continuation.

118 Essentially, Granser’s method is composed of computations involving Fourier transforms,

119 which are achieved using the FFT algorithm. It thus provides a useful tool to compute the

120 gravity effect of sedimentary basins where the density contrast varies exponentially with

121 depth.

5
 ACCEPTED MANUSCRIPT
122 Granser (1987a) rearranged Equation (1) to determine the basin configuration from the

123 gravity anomaly profile. However, this computation requires a given mean depth z0 of the

124 basement interface and a suitable low-pass filter of the gravity data to achieve convergence of

125 the inversion procedure. Moreover, using a single regularizing filter for the entire model is not

126 completely satisfactory due to the oversmoothing of shallow interfaces of the sedimentary

PT
127 basin.

RI
128 Cordell and Henderson (1968) developed another iterative inversion algorithm based on the

129 Bouguer slab formula. This algorithm can easily be computed in the space domain. In the

SC
130 algorithm of Cordell and Henderson (1968), density contrast is assumed to be uniform. In

U
131 our algorithm it can vary with depth. AN
132 Iterations begin with an initial guess for the depth of the interface h(x, y). It is assumed that

133 the observed anomaly at each grid point (i, j) is being generated by an infinite slab. Following
M

134 Cordell (1973), the first approximation of the depth to basement with exponential density
D

135 contrast variation is obtained from the equation for the infinite horizontal slab (Equation 2):
TE

1  λ∆gobs (i, j ) 
h((i,) j ) = − ln 1 −
1
 (4)
λ  2πγ∆ρ0 
136
EP

137 where, ∆gobs (i, j ) is the observed gravity anomaly at grid point (i, j).
C

138 Using the initial depth estimates as obtained from Equation (4), the gravity anomaly of the
AC

139 sedimentary basin are then calculated at each observation point by the Granser (1987a)

140 algorithm since the formula for calculating the gravity effect is much simpler and consumes

141 less computing time than the algorithm of Cordell and Henderson (1968), which is based on

142 the division of the basin into prisms. Cordell and Henderson (1968) suggested the following

143 relationship for the next modification or improvement of the model:

6
 ACCEPTED MANUSCRIPT
∆g obs (i , j )
h((i , j )) = h((i ,) j )
t +1 t
144 (5)
∆g (t )
calc ( i , j )

145 where, t stands for the number of iterations, ∆gcalc(i, j ) is the calculated gravity anomaly at grid

146 point (i, j). The above ratio is used to revise the bottom depths of prisms rather than using

PT
147 again an infinite-slab approximation as used in the method of Bott (1960).

148 After every thickness element is iterated, the gravity anomalies of the iterated model are

RI
149 recalculated, and another iteration is performed. Note that the mean depth of the basement

SC
150 interface is redefined after each iteration. The iterative process in gravity inversion tries to

151 reduce the difference between the calculated and observed anomaly. The measure of the

U
152 difference is given by the root mean square (RMS) error:
AN
∑∑ ( ∆g − ∆g calc ( i , j ) )
M N
2
obs ( i , j ) (6)
i =1 j =1
RMS =
M

153
M ×N
D

154 where N and M are the number of grid nodes along the x-axis (east) and y-axis (north)
TE

155 respectively.

156 This procedure is repeated until the model is satisfied when a desired fit between ∆gobs (i, j ) and
EP

157 ∆gcalc(i, j ) is achieved for every grid point (i, j).

C

158 As the proposed algorithm combines forward modeling in the frequency domain and iterative
AC

159 inversion in the space domain, it provides fast and accurate computation while ensuring

160 convergence.

161 3. Overview of the GCH_gravinv GUI

162 By running the GCH_gravinv program a simple graphical interface pops up covering the

163 entire screen. The configuration of the main GUI window is illustrated in Fig. 1. Here the left-

164 hand side of the window includes the main control panel divided into two parts. The upper
7
 ACCEPTED MANUSCRIPT
165 part enables the settings for the initial parameters and the criterion for the termination of the

166 iterative procedure whereas the lower part allows the user to track between the resulting data

167 and setting their visualization style. The remaining part to the right of the panel is the display

168 area for the input/output maps or graphical plots.

169 After the first run of the program, loading the input residual gravity data set is provided by the

PT
170 interactive “Import Data” menu. This allows the user two options for the format of the data; a

RI
171 grid [*.grd : (Surfer 6 text grid or Surfer 7 Binary grid format)] file ; or an equal spaced xyz

172 column wise data [*.dat ; (x:east direction, y: north direction)] where as both are compatible

SC
173 with Golden Software Surfer formats. The number of grid nodes in the east and north

U
174 direction can be both odd and even. The grid interval is required to be equal the east and north
AN
175 directions. The code stores the input data as well as its grid information to a temporary *.mat

176 file and retrieves them at every start of an iterative process or by a call for its display. After
M

177 successful data loading the information about the active input grid is shown in the uppermost

178 box of the control panel. The other required parameters such as the density contrast and the
D

179 decay rate also need to be entered in their related edit boxes. The units of the gravity
TE

180 anomalies are mGal, the densities in g/cm3 and distances are in km. Before starting the

181 algorithm, the user has the option for defining the behavior of the termination of the iteration
EP

182 procedure. One option is the divergence-mode in which the iterative procedure terminates
C

183 when the root mean square error (RMS) between the observed and calculated data at any step
AC

184 of the iteration has increased relative to the previous step. The other option is the

185 convergence-mode in which the iteration stops when the RMS is below a predefined threshold

186 value. In all cases, however, the iterative procedure ends when the user-defined maximum

187 number of iterations has been accomplished. An instant plot of the obtained RMS values

188 during the ongoing iteration steps can be displayed by checking the related radio button

189 available in the GUI. As an optional preference, to avoid edge effects, some near-edge data

8
 ACCEPTED MANUSCRIPT
190 may be cut off by editing the “ignoring from edges” field box. Then the RMS is calculated

191 ignoring data close to the edges. Ignoring edge nodes is not always necessary due to very

192 small edge effects. Using an input gravity data set larger than the area of interest could be an

193 alternative reasonable solution to avoid such effects.

194 After the user confirms the inputs, the code first calculates the initial approximations of the

PT
195 depths to basement with exponential density contrast variation using Equation (4) and goes

RI
196 through with the iterative procedure. During this, two independent background functions are

197 called consecutively of which one is for the forward FFT based computation of the gravity

SC
198 effect achieved using Equation (3) and the other for the next updating of the model depths

U
199 using Equation (5) in the space domain but also for the control of the convergence and to
AN
200 enforce the termination when the user defined criterion is achieved.

201 Finally, the function stores the output data of the inverted basement depth, the gravity
M

202 response due to the calculated model depths, the differences between the observed and
D

203 computed gravity and the RMS error calculated at each iteration step. Plotting of the outputs
TE

204 can be guided from the lower part of the control panel of the GUI with some optional view

205 styles. The colormap tool enables the user to alter the color zones interactively, by setting the
EP

206 color limits of the map to specified minimum and maximum values. All of the output data can

207 be stored by a user defined name either to a *.mat file comprising the complete outputs or to
C

208 seperated files of *.grd format for the maps and *.dat format for the RMS vector. Exporting of
AC

209 a current figure is also available. The program also allows the user to forward calculate

210 gravity from a defined depth grid which might be used as a part of the inversion process, e.g.

211 for removing regional effects of deep layers.

212

213

214

9
 ACCEPTED MANUSCRIPT
215 4. Synthetic data application
216 In this section, we present the application of the proposed algorithm on synthetic data

217 produced by a known basin model. The parameters of the synthetic model are as follows: the

218 total number of columns and rows are 134 × 106; the square grid interval is 1 km; the depth

219 structure of the basin is shown in Fig. 2a, the density contrast of sediments within the basin

PT
220 varies according to the equation: ∆ρ = −0.57e−0.25 z g/cm3.

RI
221 The gravity anomaly due to the model obtained from the application of the Granser method is

222 shown in Fig. 2b which is used as the observed anomaly. Using this input data, we performed

SC
223 the inverse procedure by the above described algorithm. As an optional preference, to avoid

224 edge effects, three data nodes from edges were ignored during the calculations.

225
U
Using threshold value of the convergence criteria 0.01 mGal, the inversion scheme required
AN
226 twentythree iterations for the convergence of the computed anomalies to the observed
M

227 anomalies. The inverted depth results are shown in Fig. 2c. Comparison of Fig. 2a and Fig. 2c

228 shows that the inverted depth coincides well with the theoretical depth. The error between
D

229 these depths is shown in Fig. 2e and the RMS error between them is only 0.009 km.
TE

230 Moreover, as shown in Fig. 2b and Fig. 2d, the gravity anomaly from the inverted depth fits

231 well with the observed anomaly. The difference between them is shown in Fig. 2f. It can be
EP

232 observed that these differences are very small and are in the range of -0.02 - +0.04 mGal.

233 After the initial depth estimate by Eq. 4, the value of the RMS error between observed and
C
AC

234 calculated anomalies is 2.57 mGal and reaches 0.009 mGal at the last iteration (Fig. 2g). We

235 note here that the calculation of the basement depth from the observed gravity anomalies on a

236 2-D grid of 134 × 106 data, took only about 2.1 sec, which is a very short time for this kind of

237 calculation. For a comparison (on computation time and precision of our proposed method),

238 the synthetic gravity anomaly was also inverted by the algorithm of Chai and Hinze (1988),

239 based on the stacked prism approach. Although the inverted depth result compares very

10
 ACCEPTED MANUSCRIPT
240 favorably with the theoretical depth (Fig. 2h), the time taken to invert this model grid on the

241 same PC is about 1143.6 sec. Using this algorithm, the inversion scheme performed sixteen

242 iterations and RMS error between the inverted and theoretical depth is 0.011 km.

243 In order to demonstrate the effectiveness of the proposed algorithm, random noise with

244 amplitude equal to 1% of the anomaly amplitude has been added to the synthetically observed

PT
245 anomaly shown in Fig. 2b. The observed anomaly data with the noise (Fig. 3a) is given by

∆g obs _ noise = ∆g obs + 0.01× ∆g obs × rand (106,134)

RI
246
247
248 Inverted depth results are shown in Fig. 3b. It can be observed that the inverted depth

SC
249 compares reasonably with the model depth. Fig. 3c shows the map of difference between

250 them. In this noisy case, the algorithm performed twelve iterations (Fig. 3d). The RMS error

251
U
between the calculated and actual depths is 0.12 km.
AN
252
M

253 5. Real data application

254 The practical applicability of the algorithm is demonstrated with the interpretation of gravity
D

255 anomalies observed over the Chintalapudi subbasin, close to the Eastern Continental Margin
TE

256 of India (Fig. 4). It is one of the subbasins that compose Godavari basin, and is situated in the
EP

257 south-eastern part of the Godavari basin (Mishra et al., 1989). According to Mishra et al.

258 (1989), the Chintalapudi subbasin is a full graben with well-developed rift tectonics. This
C

259 subbasin has been interpreted previously by several other studies (e.g. Chakravarthi and
AC

260 Sundararajan, 2004, 2007; Silva et al., 2014; Santos et al., 2015; Silva and Santos, 2017) with

261 the aim of obtaining the basement relief of the basin.

262 The size of the study area is 60 km × 40 km involving observed gravity data grids with 1 km

263 interval. From the borehole density measurements in Chintalapudi basin, the sediment density

264 is known to vary with depth. Chakravarthi and Sundararajan (2007) have modeled the

265 density-depth dependence of the basin by a parabolic density function. In this study, the

11
 ACCEPTED MANUSCRIPT
266 density-depth dependence of the basin was approximated by a least-squares fit of an

267 exponential function to the measured data reported by Chakravarthi (2003). Fig. 5 shows the

268 approximation of density versus depth data by the exponential function. The calculated

269 parameters of the exponential function related with the constants ∆ρ and λ are -0.5246 g/cm3

270 and 0.5771 g/cm3, respectively.

PT
271 The observed gravity anomaly of the Chintalapudi subbasin has been inverted previously by

RI
272 Chakravarthi and Sundararajan (2007), Silva et al. (2014), Santos et al. (2015) and Silva and

273 Santos (2017) without any adjustment of the regional field since it appears to be zero and the

SC
274 data can be interpreted immediately. However, a removal of the regional field is required in

U
275 general. Fig. 6a shows the observed gravity anomaly over the Chintalapudi subbasin digitized
AN
276 from Chakravarthi and Sundararajan (2007). Here, this anomaly was inverted using the above

277 described algorithm. Fig. 6b shows the inversion result of the anomaly map of the basin. The
M

278 geometry of the inverted interface compares favorably with those obtained by Chakravarthi

279 and Sundararajan (2007), Silva et al. (2014), Santos et al. (2015), and Silva and Santos (2017)
D

280 (Fig. 6f). The mean depth of the subbasin floor obtained from the present method is
TE

281 approximately 3 km that is within the interval of 3–5 km reported by Bastia et al. (2010) for
EP

282 the onshore depressions in the Eastern Continental Margin of India, and it agrees well with

283 the results given by other authors (Silva et al., 2014; Santos et al., 2015; Silva and Santos,
C

284 2017). This value is also in good agreement with the depth of 2.935 km obtained in a borehole
AC

285 reported by Chakravarthi and Sundararajan (2007). Fig. 6c shows the gravity anomaly map

286 obtained from the application of Equation (3) to the depth to basement map obtained. By

287 comparing Fig. 6a and Fig. 6c, it is clear that the depth to basement of the Chintalapudi basin

288 obtained from the method produces a gravity anomaly very similar in shape to the original

289 residual gravity. Fig. 6d shows the difference between the residual and computed anomaly

12
 ACCEPTED MANUSCRIPT
290 maps comprising a range of -0.7 – +0.2 mGal which can be accepted as to be small when

291 comparing with the range of the overall observed data.

292 Convergence of the algorithm has been achieved at the twelfth iteration, where the

293 computation time was only 0.45 sec. The variation of RMS errors calculated during the

294 iterative procedure versus the iteration number is shown in Fig. 6e. The RMS error

PT
295 corresponding to the initial structure is 1.1852 mGal and reaches 0.0565 mGal at the optimum

RI
296 solution.

SC
297 6. Conclusion

298 We have presented a Matlab-based code, GCH_gravinv including an easy-to-use graphical

299
U
user interface for determining the depth to the basement of sedimentary basins derived from
AN
300 inverting gravity anomalies by an iterative procedure. The code is compatible with the case

301 where density contrast varies exponentially with depth. Based on Granser (1987a) and Cordell
M

302 and Henderson (1968) algorithms, the code combines the advantages of both the frequency
D

303 domain and space domain techniques. Thus, it not only reduces the computation time, but also
TE

304 retains the required accuracy in the computations. The accuracy and effectiveness of the

305 developed code were tested on a synthetically produced gravity data set. Considering the
EP

306 known model depths, the inverted results are in good agreement with the actual ones. The

307 code was also evaluated by reanalysing real data set from the Chintalapudi subbasin. By using
C

308 the prior information of the densities from an available drill-hole study, the calculated depth
AC

309 results after an optimum convergence between the observed and inverted gravity anomalies

310 yield a good agreement with the results of previous studies. Consequently, the developed code

311 was proven useful in interpreting gravity anomalies observed over sedimentary basins. The

312 code may be an efficient tool for estimation of the basement surface of sedimentary basins.

313

13
 ACCEPTED MANUSCRIPT
314 Acknowledgements

315 The authors sincerely thank David Gómez-Ortiz from King Juan Carlos University, Spain for

316 his very useful suggestions to improve the manuscript. The authors record with pleasure their

317 sincere thanks to the editor Dario Grana and the four anonymous reviewers for their very

318 constructive suggestions to improve the manuscript.

PT
319 Computer Code Availability

RI
320 The GCH_gravinv program is an open source code and can be obtained from the web site

SC
321 [http://www.iamg.org] or from the corresponding author for use in many geophysical and

322 tectonic problems.

323
U
AN
324 References
M

325 Barbosa, V.C.F., Silva, J.B.C., Medeiros, W.E., 1997. Gravity inversion of basement relief
D

326 using approximate equality constraints on depths. Geophysics 62(6), 1745–1757.

TE

327 Barbosa, V.C.F., Silva, J.B.C., Medeiros, W.E., 1999. Gravity inversion of a discontinuous

328 relief stabilized by weighted smoothness constraints on depth. Geophysics 64(5), 1429–
EP

329 1437.

330 Bastia, R., Radhakrishna, M., Srinivas, T., Nayak, S., Nathaniel, D.M., Biswal, T. K., 2010.
C

331 Structural and tectonic interpretation of geophysical data along the Eastern Continental
AC

332 Margin of India with special reference to the deep water petroliferous basins. Journal of

333 Asian Earth Sciences 39, 608–619.

334 Bott, M.H.P., 1960. The use of rapid digital computing methods for direct gravity

335 interpretation of sedimentary basins. Geophysical Journal of the Royal Astronomical

336 Society 3, 63–67.

14
 ACCEPTED MANUSCRIPT
337 Chai, Y., Hinze, W.J., 1988. Gravity inversion of an interface above which the density

338 contrast varies exponentially with depth. Geophysics 53, 837–845.

339 Chakravarthi, V., 2003. Digitally implemented method for automatic optimization of gravity

340 fields obtained from three-dimensional density interfaces using depth-dependent density.

341 U. S. Patent 6,615,139.

PT
342 Chakravarthi, V., Sundararajan, N., 2004. Automatic 3D gravity modeling of sedimentary

RI
343 basins with density contrast varying parabolically with depth. Computers and Geosciences

344 30, 601–607.

SC
345 Chakravarthi, V., Sundararajan, N., 2005. Gravity modeling of 2.5-D sedimentary basins – a

346 case of variable density. Computers and Geosciences 31, 820–827.

347
U
Chakravarthi, V., Sundararajan, N., 2007. 3D gravity inversion of basement relief – a depth
AN
348 dependent density approach. Geophysics 72(2), 23–32.
M

349 Chakravarthi, V., Ramamma, B., Venkat Reddy, T., 2013a. Gravity anomaly modeling of

350 sedimentary basins by means of multiple structures and exponential density contrast-depth
D

351 variations: a space domain approach. Journal of the Geological Society of India 82(5),
TE

352 561–569.

353 Chakravarthi, V., Rajeswara Sastry, S., Ramamma, B., 2013b. MODTOHAFSD–a GUI based
EP

354 JAVA code for gravity analysis of strike limited sedimentary basins by means of growing

355 bodies with exponential density contrast–depth variation: a space domain approach.
C
AC

356 Computers and Geosciences 56, 131–141.

357 Chakravarthi, V., Pramod Kumar, M., Ramamma, B., Rajeswara Sastry, S., 2016. Automatic

358 gravity modeling of sedimentary basins by means of polygonal source geometry and

359 exponential density contrast variation: Two space domain based algorithms. Journal of

360 Applied Geophysics 124, 54–61.

15
 ACCEPTED MANUSCRIPT
361 Cordell, L., Henderson, R.G., 1968. Iterative three-dimensional solution of gravity anomaly

362 data using a digital computer. Geophysics 33, 596–601.

363 Cordell, L., 1970. Iterative Three-Dimensional Solution of Gravity Anomaly Data. National

364 Technical Information Service PB-196-979, U.S. Department of Commerce.

365 Cordell, L., 1973. Gravity anomalies using an exponential density-depth function - San

PT
366 Jacinto Graben, California. Geophysics 38, 684–690.

RI
367 Cordell, L., Phillips, J.D., Godson, R.H., 1992. U.S. Geological Survey Potential-Field

368 Software Version 2.0, Open-File Report 92-18, U.S. Geological Survey.

SC
369 Gomez-Ortiz, D., Agarwal B.N.P., 2005. 3DINVER.M: a MATLAB program to invert

370 gravity anomaly over a 3-D horizontal density interface by Parker-Oldenburg’s algorithm.

371 Computers and Geosciences 31, 513–520.

U
AN
372 Granser, H., 1987a. Three - dimensional interpretation of gravity data from sedimentary
M

373 basins using an exponential density – depth function. Geophysical Prospecting 35, 1030–

374 1041.
D

375 Granser, H., 1987b. Nonlinear inversion of gravity data using the Schmidt-Lichtenstein
TE

376 approach. Geophysics 52, 88–93.

377 Litinsky, V. A., 1989. Concept of effective density: Key to gravity depth de-terminations for
EP

378 sedimentary basins. Geophysics 54, 1474–1482.

379 Mendonca, C.A., 2004. Inversion of gravity field inclination to map the basement relief of
C
AC

380 sedimentary basins. Geophysics 69, 1240–1251.

381 Mishra, D. C., Gupta, S. B., Venkatarayudu, M., 1989. Godavari rift and its extension towards

382 the east coast of India. Earth and Planetary Science Letters 94, 344–352.

383 Murthy, I.V.R., Rao, D.B., 1979. Gravity anomalies of two-dimensional bodies of irregular

384 cross-section with density contrast varying with depth. Geophysics 44, 1525–1530.

16
 ACCEPTED MANUSCRIPT
385 Murthy, I.V.R., Rao, S.J., 1989. A Fortran 77 program for inverting gravity anomalies of two-

386 dimensional basement structures. Computers and Geosciences 15, 1149–1156.

387 Nagendra, R., Prasad, P.V.S., Bhimasankaram, V.L.S., 1996a. Forward and inverse computer

388 modeling of gravity field resulting from a density interface using Parker-Oldenburg

389 method. Computers and Geosciences 22, 227–231.

PT
390 Nagendra, R., Prasad, P.V.S., Bhimasankaram, V.L.S., 1996b. FORTRAN program based on

RI
391 Granser's algorithm for inverting a gravity field resulting from a density interface.

392 Computers and Geosciences 2, 219–225.

SC
393 Oldenburg, D.W., 1974. The inversion and interpretation of gravity anomalies. Geophysics

394 39(4), 526–536.

395
U
Pallero, J.L.G., Fernandez-Martinez, J.L., Bonvalot, S., Fudym, O., 2015. Gravity inversion
AN
396 and uncertainty assessment of basement relief via Particle Swarm Optimization. Journal of
M

397 Applied Geophysics 116, 180–191.

398 Parker, R.L., 1972. The rapid calculation of potential anomalies. Geophysical Journal of the
D

399 Royal Astronomical Society 31, 447–455.

TE

400 Phillips, J.D, 1997. Potential-field geophysical software for the PC, version 2.2, Open-File

401 Report 97-725, U.S. Geological Survey.

EP

402 Rao, D.B., 1986. Modeling of sedimentary basins from gravity anomalies with variable

403 density contrast. Geophysical Journal of the Royal Astronomical Society 84, 207–212.
C
AC

404 Rao, D.B., 1990. Analysis of gravity anomalies of sedimentary basins by an asymmetrical

405 trapezoidal model with quadratic density function. Geophysics 55, 226–231.

406 Rao, D.B., Prakash, M. J., Ramesh Babu, N., 1990. 3-D and 2 1/2-D modeling of gravity

407 anomalies with variable density contrast. Geophysical Prospecting 38, 411-422.

17
 ACCEPTED MANUSCRIPT
408 Santos, D. F., Silva, J.B.C., Martins, C.M., dos Santos, R.D.C.S., Ramos, L. C., de Araújo,

409 A.C.M., 2015. Efficient gravity inversion of discontinuous basement relief. Geophysics

410 80(4), 95–106.

411 Silva, J.B.C., Costa, D.C.L., Barbosa, V.C.F., 2006. Gravity inversion of basement relief and

412 estimation of density contrast variation with depth. Geophysics 71, 51–58.

PT
413 Silva, J.B.C., Oliveira, A.S., Barbosa, V.C.F., 2010. Gravity inversion of 2D basement relief

RI
414 using entropic regularization. Geophysics 75(3), 29–35.

415 Silva, J.B.C., Santos, D.F., Gomes, K.P., 2014. Fast basement relief inversion. Geophysics

SC
416 79(5), 79–91.

417 Silva, J.B.C., Santos, D.F., 2017. Efficient gravity inversion of basement relief using a

418
U
versatile modeling algorithm. Geophysics 82, 23–34.
AN
419 Shin, Y.H., Choi, K.S., Xu, H., 2006. Three-dimensional forward and inverse models for
M

420 gravity fields based on the Fast Fourier Transform. Computers and Geosciences 32, 727–

421 738.
D

422 Talwani, M., Worzel, J., Ladisman, M., 1959. Rapid gravity computations for two
TE

423 dimensional bodies with application to the Mendocino submarine fracture zone. Journal of

424 Geophysical Research 64, 49–59.

EP

425 Zhou, X., 2009. 3D vector gravity potential and line integrals for the gravity anomaly of a

426 rectangular prism with 3D variable density contrast. Geophysics 74(6), 43–53.
C
AC

427 Zhou, X., 2013. Gravity inversion of 2D bedrock topography for heterogeneous sedimentary

428 basins based on line integral and maximum difference reduction methods. Geophysical

429 Prospecting 61(1), 220–234.

430

431

432

18
 ACCEPTED MANUSCRIPT
433 Figures and Captions

PT
RI
U SC
AN
M
D
TE

434
EP

435 Figure 1. Screenshot views of the GCH_gravinv GUI.

C
AC

19
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP

436
AC

437 Figure 2. Synthetic example. (a) depth of the basement interface, (b) gravity anomalies due to

438 the basin with exponential density contrast, (c) the inverted depth of the model, (d) gravity

439 anomaly from the inverted depth, (e) the difference between the actual and inverted depth, (f)

440 the difference between observed and computed gravity anomaly, (g) variation of RMS error

441 versus iteration number, (h) the inverted depth using the algorithm of Chai and Hinze (1988).

20
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M

442

443 Figure 3. Synthetic example with 1% random noise. (a) gravity anomalies with random noise,
D

444 (b) calculated depth of the basement interface, (c) the difference between the actual and
TE

445 inverted depth, (d) variation of RMS error versus iteration number.
EP

446
C
AC

21
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
447

448 Figure 4. Location of Chintalapudi subbasin.

M

449
D
TE
C EP
AC

22
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE

450

451 Figure 5. The fit of the exponential function to the Chintalapudi subbasin sediment density-
EP

452 depth data (the measured density contrast was digitized after Chakravarthi, 2003).
453
C

454
AC

23
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE

455
EP

456 Figure 6. Chintalapudi subbasin. (a) Observed gravity anomaly map of the basin (digitized
C

457 after Chakravarthi and Sundararajan (2007), (b) basement surface of Chintalapudi subbasin
AC

458 obtained by the proposed algorithm, (c) recalculated gravity anomaly based on the estimated

459 depth to basement, (d) difference between observed gravity (Fig. 6a) and the recalculated

460 gravity (Fig. 6c), (e) plot of RMS error versus iteration number obtained during the iterative

461 procedure, (f) The basement topographic map of the Chintalapudi subbasin (after Silva and

462 Santos, 2017).

463

24
 ACCEPTED MANUSCRIPT
GCH_gravinv: a MATLAB-based program for inverting gravity anomalies over sedimentary basins.

Highlights

• We developed a MATLAB program to invert gravity anomalies over sedimentary

basins

• The algorithm combines the advantages of both the FFT-based and space domain

PT
techniques.

RI
• It can be applied to the case where density contrast varies exponentially with depth.

• The inversion scheme is independent of mean depth and low-pass filtering

SC
• The code provided good result over 3D synthetic model as well as on real gravity data

U
AN
M
D
TE
C EP
AC