Contributions to Geophysics and Geodesy Vol.

48/2, 2018 (133–160)

An optimal multiscale approach to

interpret gravity data using successive
decomposition of the Bouguer anomaly

Mohamed ARFAOUI1,2 , Mohamed Hedi INOUBLI2, Adnen AMIRI2

1
Office National des Mines,
24, Rue de l’Energie, 2035, La Charguia, Tunis, Tunisia; e-mail: [email protected]
2
Unité de Recherche de Géophysique Appliquée aux Matériaux et aux Minerais, Université
Tunis-El Manar, Faculté des Sciences de Tunis, 2092 El Manar, Tunis, Tunisia

Abstract: A multiscale interpretive approach is presented to identify multiscale discon-

tinuities and sources boundaries, and to obtain comprehensive pseudo-section density im-
ages of 2D structures at depth. This interpretive approach is based on the total horizontal
gradient and apparent density operator applied to the gravity anomalies of pseudo-depth
slices derived from the successive decomposition of the Bouguer anomaly. This approach
is applied to gravity data from the northwestern part of Tunisia, it provides the fault
system repartition at different depths and describes the density distribution within the
basins and trough. Thus, the NW–SE and NE–SW faults are the deepest whereas the
N–S and E–W faults are relatively superficial. The density pseudo-sections reveal the
vertical continuity of salt Triassic outcrops of Jebel Debadib as well as a diapiric model,
the presence of a Triassic salt dome in the Kalaa El Khasba through and dense series
under the Miopliocene and Quaternary deposits of the Mejarda basin.

Key words: Gravity anomalies, Tunisia, Multiscale discontinuities, Pseudo-depth slices,

Pseudo-section density

1. Introduction

Gravity surveys are designed to delineate geologic structures at depth by in-

terpretation of observed data. The essential question is to find a quantitative
and qualitative interpretive process for determining 3D locality and density
of causative sources. Generally interpretation of gravity data follows three
principle strategies: (1) anomaly separation, (2) identifying source bound-
aries and discontinuities, and (3) resolving sources at depth, which includes
depth determination and source density.

doi: 10.2478/congeo-2018-0005 133

Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Anomaly separation into regional and residual components by decompo-

sition is commonly used in gravity interpretation. Short wavelength compo-
nents associated with near surface geologic bodies are separated from long
wavelength features associated with deep seated structures. Decomposition
may be achieved by least squares fitting of polynomial surfaces (Simpson,
1954), wavenumber filters as an upward continuation (Gupta and Ramani,
1980), Gaussian function (Arfaoui et al., 2015), or bandpass Wiener opti-
mum filters (Pawlowsky and Hansen, 1990). Since the Bouguer anomaly
is the sum of several components caused by deep seated, intermediate and
near surface geologic bodies the gravity effect of those sources may be es-
timated from a successive decomposition process where the decomposition
“bootstraps” to greater depths with each iteration (Arfaoui et al., 2011).
Sources boundaries and discontinuities are usually defined by several dif-
ferent derivative operators. These discontinuities generally correspond with
high density contrast zones associated with geological contacts and faults.
Various derivative operators have been developed for enhancing the gravity
signal. They are designed using a combination of simple gravity gradients
and have been tested by many studies. One fundamental study was real-
ized by Cordell and Grauch (1982, 1985). They used the maximum total
horizontal gradient to map boundaries between domains of uniform density
and magnetic susceptibility. The use of this operator was improved and
automated (Blakely and Simpson, 1986) by assigning a quality factor to the
maximum gradient after comparing each point of the grid with its neigh-
bors. Miller and Singh (1994) and Verduzco et al. (2004) introduced the tilt
angle (TDR) and the total horizontal derivative of the tilt angle (THDR),
respectively, which are relatively powerful operators to detect boundaries
between areas having a density contrast. The high resolution standard-
ized operators such as the Theta map, horizontal tilt angle (TDX) and
E-Tilt, are elucidated by Wijns et al. (2005), Cooper and Cowan (2006)
and Muzaffer and Ünal (2013), respectively. Despite the high resolution of
the derivative operators, the discontinuities and source boundaries derived
from their application to the Bouguer anomaly, represent a resultant of dif-
ferent source components seated at different depths. Fedi and Florio (2001)
presented a multiscale derivative analysis (MDA) based on the enhanced
horizontal derivative (EHD). This multiscale analysis allows the determina-
tion of source boundaries and discontinuities at different depths.

134
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

The first two principal strategies concern themselves with resolving the
discontinuities and boundaries of geologic sources while the third is con-
cerned with source characterization. Numerous quantitative interpretations
of potential field anomalies are developed after calculation of simple source
effects by Nettleton (1942). Despite the complexity of the resolution pro-
cess caused by the nonlinearity, methods were developed to assess the source
parameters from the potential field anomalies as the Werner deconvolution
(Werner, 1953) and the Euler deconvolution (Thompson, 1982).
The latest application was developed by Reid et al. (1990) for the in-
terpretation of 2D data. Other applications of Euler deconvolution applied
to the analytic signal and the enhanced analytical signal are introduced by
Keating and Pilkington (2004) and Salem and Ravat (2003), respectively.
The source parameters imaging method (SPI) developed by Thurston and
Smith (1997) is also effective for the evaluation of source parameters of
potential field anomalies. The 2D forward modeling of density contrasts
at depth has occupied equally an important place in quantitative gravity
interpretation studies. The 2D modeling generates geometrical shape and
physical properties which provide the closest calculated anomaly to the ob-
served one (Talwani et al., 1959; Talwani and Ewing, 1960). However,
density contrasts with depth are generally determined using inversion tech-
niques (Last and Kubik, 1983; Mareschal, 1985; Murthy and Rao, 1993).
In this paper, we present a multiscale interpretive approach adopted for
the three gravity modeling techniques based on the successive decomposi-
tion of the Bouguer anomaly. This approach first determines discontinuities
and boundaries of sources at different depths in the first step. In the second
step, the imaging of density repartition at depth is resolved. This approach
was applied to gravity data from northwestern Tunisia with satisfactory
results.

2. Processing and methodology

Potential field anomalies are dissociated into residual and regional anomalies
using a method of separation. Several pairs of regional and residual anoma-
lies can be obtained from each separation method by changing the director
parameter. However, it is possible to separate the Bouguer anomaly into

135
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

several components, since it is the sum of the gravity effects of superim-

posed geological structures situated at different depths. Therefore, several
multi-component decomposition scenarios of Bouguer anomaly are possible
as the successive decomposition developed by Arfaoui et al. (2011).
Successive decomposition was used to decompose the gravity anomaly
with a pseudo-depth slicing scheme (Fig. 1). It was carried out in first
step to dissociate the Bouguer anomaly into several pairs of residual and re-
gional anomalies associated with different controlling parameters, which cor-
respond to different pseudo-depths of separation. The second step concerns
the dissociation of the residual anomaly corresponding to the pseudo-depth
Pi to two anomalies by subtracting the residual anomaly on the pseudo-
depth Pi−1 (Pi−1 < Pi ), and we obtained:
– The gravity anomaly of the pseudo-depth slice sited between the pseudo-
depths Pi and Pi−1 .
– Residual anomaly on the Pi−1 depth.

Fig. 1. Successive decomposition of the Bouguer anomaly.

136
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

The next steps consist to determining the gravity effects of the pseudo-
slice located between the pseudo-depths P i−1 and Pi−2 by the deduction
of the residual anomaly corresponding to the pseudo-depth P i−2 from the
residual anomaly associated to the pseudo-depth P i−1 . Thus, the successive
decomposition of the Bouguer anomaly using a given separation method,
leads to gravity anomalies of several pseudo-depth slices located between
the pseudo-depths Pi and Pi−1 , whose sum is equal to the Bouguer anomaly
(Fig. 1).
We decompose the gravity anomaly with a pseudo-depth slicing scheme
using the Gaussian filter as a separation method (Fig. 1). Spectral depth
analysis in the wavenumber domain (Spector and Grant, 1970; Negi et al.,
1986; Maus and Dimri, 1996; Bansal and Dimri, 2001) is then used for ap-
proximating pseudo-depths of pseudo-slice interfaces. We applied the spec-
tral depth analysis method to estimate depths of the causative sources of
anomalies in residual maps associated with standard deviation of Gaussian
filter equal 0.4, 0.3, 0.2, 0.1, 0.05 and 0.02 cycle km −1 , which approximate
the pseudo-depths of pseudo-slice interfaces. The logarithm of the radial
average of the energy spectrum of each residual map was plotted versus the
radial frequency (Fig. 5). The half slope of each linear segment of the spec-
trum corresponds to the mean depth of the causative sources of anomalies
in the residual maps.
The total horizontal gradient operator is applied to the gravity of pseudo-
depth slices derived from the successive decomposition of the Bouguer ano-
maly to produce horizontal gradient maps at different pseudo-depths. The
maxima of this operator correspond to the abrupt density variation zones
associated to geological contacts and faults.
The apparent density operator is applied equally to the gravity of pseudo-
depth slices to establish apparent density maps at different pseudo-depths
from which we evaluate pseudo-density sections. The apparent density op-
erator is performed in the wavenumber domain which assumes that a hori-
zontal layer of fixed thickness and density variable could explain the gravity
signal. The operator determines the contrast of the apparent density in the
wavenumber domain as expressed in the equation (1) given by Gupta and
Grant (1985):
r
L(r) = (1 + e−tr )−1 , (1)
2πG

137
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

where G is the gravitational constant, r is the wavenumber and t is the

thickness of the equivalent layer (pseudo-slice). The apparent density maps
were calculated by using a thickness equivalent layer equal to the thickness
of the pseudo-depth slices. Thus, the thickness of 490, 150, 350, 400, 1210
and 1950 m are used for the gravity anomaly of pseudo-slices 1, 2, 3, 4, 5
and 6 (Fig. 6), respectively. Taking into consideration density values of 350
surface samples and the repartition of outcrops, the background densities
of 2.25, 2.28, 2.3, 2.33, 2.38 and 2.42 g/cm 3 are adopted for the calculus of
apparent density maps of the pseudo-slices 1 to 6.
The six apparent density maps associated the maps of different slices
were used to established density pseudo-slice sections. The apparent density
data of each profile are then extracted from the six apparent density maps
profiles, combined and attached to pseudo-depths of the anomaly separation
interface. After, they are gridded by profile to generate apparent density
pseudo-section.

3. Application of the interpretive approach to field case

3.1. Geographic and Geological settings of the study area

The study area is located in the northwestern part of the Tunisian Atlas.
It extends from Kalaa El Kasba in the South to Jebel Hairech in the North
(Fig. 2).
The study area is located at the front of the nappe zone. It is charac-
terized by folding tectonic, Atlasic type deformation and salt structures. It
includes the western parts of three different geological domains: the Me-
jarda basin in the North, the diapir zone in the middle and the central
Tunisian Atlas in the South. Furthermore, the northern part of the study
area includes the Permo-Triassic, Triassic, Jurassic and Cretaceous struc-
tures, which generally constitute the Mejarda basin limits (Fig. 3). The
diapiric zone includes Triassic and Aptian anticlines, Early Eocene syn-
clines and Mio-Pliocene and Quaternary filled depressions. The most no-
table feature of the southern section is the Kalaa El Khasba through filled
with Mio-Plio-Quaternary deposits (Chihi, 1995) and the Aptian Recifal
limestone outcrops in the form of resistant structures that dominate the
relief (Mahjoubi, 1978; Smati, 1986). Northwest-southeast and northeast-

138
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

Fig. 2. Location of the study region.

southwest trending faults locally marked by Triassic material, exert control

on these structures that are also cut by north-south and east-west trending
transverse faults (Sainfeld, 1951; Burollet and Sainfeld, 1956b; Chikhaoui,
2002; Smati, 1986; Burollet, 1991; Zaier, 1999b; Ould Begga, 2003; Amiri,
2013). The post-Albian series varies laterally in thickness and facies in re-
sponse to geodynamic and depositional processes (Perthuisot, 1978). Thus,
the reduced carbonate series and gaps in the sedimentary sequences may be
identified close to Triassic outcrops (Fig. 3). These sequences confirm up-
shoaling strata at the emplacement of actual Triassic extrusions (Perthuisot,
1978; Perthuisot et al., 1998). Thick and complete sedimentary series with
a major marl component are deposited in rim-synclines (Mellègue’s dam,
Oued Bou Adila area, Jebel Bou Lahnèche and Jebel Jerissa). Since Late
Aptian time, the structural evolution of this region has been controlled by
successive extensional and compressive tectonic events related to halokinesis
(Perthuisot, 1978; Perthuisot et al., 1998; Chikhaoui, 2002). The structures
comprising basins and Cretaceous up-shoaling were strongly influenced by
Tertiary tectonics, thus initiating anticlines, synclines and collapsed troughs.

139
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Fig. 3. Geological map of the study region adapted from Sainfeld (1951), Gottis and
Sainfeld (1955), Burollet and Sainfeld (1956a and 1956b), Lehotsky (1979), Lehotsky and
Bujnowsky 1995, Ben Haj Ali (1997), Mahjoub (1997), Fakhraoui et al. (1998), Zaier
(1999a).

140
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

The main structures (Fig. 3) of the study area are:

– The Mejarda Basin filled by Mio-Pliocene-Quaternary deposits,
– The salt Triassic structures explained by the diapir and salt glacier mod-
els (Ghanmi et al., 2000; Perthuisot et al., 1998; Rouvier et al., 1998;
Vila et al., 1994; 1996),
– The Kef trough sited in the SW Tunisian furrow boundary, is filled by
Mio-Pliocene-Quaternary sediments,
– The Kalaa El Khasba trough is a part of the offset graben system of
Rohia, Kalaa El Khasba and Ouled Bou Ghanem. It is filled with Mio-
Pliocene-Quaternary deposits.

3.2. Gravity anomaly

The Bouguer anomaly represents all mass heterogeneities in depth. Usually

the most interesting result from gravity surveys occurs when the survey re-
veals lineaments that are not observed at the surface. The Bouguer anomaly
map also provides important information on the nature of geological struc-
tures, and structural directions. A total of 4852 gravity stations from five
surveys acquired at 1 km intervals are used in this paper. These gravity
surveys belong to a current Tunisia gravity project conducted by the Office
National des Mines (Tunisia). It was begun in 1996 as systematic and long
term gravity coverage with a 1 station per km 2 for all of Tunisia. Consis-
tent high quality standards in gravity data and elevation control have been
maintained since the start of the project. The gravity data acquisition and
processing and accuracy standards are described in more detail in previous
papers (Arfaoui et al., 2011; Arfaoui and Inoubli, 2013).
The Bouguer anomaly map of the study area shows southern and north-
ern domains occupied by long-wavelength negative and positive anomalies,
respectively. They are separated by an intermediary domain characterized
by short and medium wavelength anomalies (Fig. 4). The northern domain
includes the Medjerda Basin and Permo-Triassic, Jurassic and Cretaceous
structures of Jebel Hairech and Oued Maaden. It is dominated by long wave-
length positive anomalies oriented following E–W and NE–SW directions;
with amplitudes ranging between –20 and 13 mGal. Moreover, this domain
indicates near surface high density bodies under the Pliocene-Quaternary

141
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Fig. 4. Bouguer anomaly map.

142
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

cover (Fig. 4). The middle domain comprises the Ouargha- El Kef region
and its surroundings. It is characterized by short and medium wavelength
NE–SW and NW–SE anomalies. It is characterized by positive anomalies
coinciding with Quaternary deposits in El Kef and Zouarine plains. They
correspond to Cretaceous and/or Eocene structures. The southern domain
extends to the south of Tajerouine and is essentially defined by negative

Fig. 5. Radially averaged spectrum of residual maps. Sp Res Gaussian filter 0.3: radially
averaged spectrum of the residual map obtained by Residual/ Regional Gaussian filter
with standard deviation equal 0.3 cycle km−1 .

143
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

NW–SE anomalies with amplitudes reaching –55 mGal. The most impor-
tant one is the Kalaa El Khasba Trough (Fig. 4).
The NW–SE and NE–SW trending gravity lineaments observed on the
Bouguer anomaly map correspond to outcropping structures (Kalaa Khasba
and El Kef troughs, salt Triassic structures of Jebel Debadib- Ben Gasseur...).
They present interruptions and spatial shifts and high gradients along cer-
tain anomalies which reflect faults and their effects. The Mio-Plio-Quaterna-
ry deposits are associated with two types of anomalies: negative anomalies
corresponding to low density material filling collapsed zones and positive
anomalies comparable to those observed in the Triassic and Cretaceous se-
ries. The positive anomalies are not related to Quaternary deposits but
rather support the presence of mass excess. These are attributed to the
Triassic and/or Cretaceous and Eocene dense rocks bordering the previ-
ously defined collapsed zones. Thus, the positive anomalies situated to the
southeast of Jebel Hairech and in the plains of Jendouba and El Kef which
coincide with quaternary units are actually caused by denser rocks under
the quaternary cover.
In the north, the Jebel Hairech Permo-Triassic and Kef Agueb Jurassic
structures belong to a NE–SW positive gravity anomaly which extends to
the NE and SW crossing Quaternary units. This confirms the continuity
of these structures under the Quaternary overburden. Moreover, the Jebel
Hairech structure is bordered by relatively high gradient zones indicating
the presence of significant faulting. They appear to have played an impor-
tant role in the evolution of this structure (Fig. 4). The anomaly amplitudes
associated with geological series of same age increase from south to north.
Generally the Bouguer anomaly map explains the majority of mapped
outcrops, but it also reveals uncorrelated anomalies, that are probably
caused by deeper structures. Resolving the Bouguer anomaly into residual
and regional components can resolve the near surface and deeper sources.

3.3. Successive decomposition of the Bouguer anomaly

Successive decomposition of the Bouguer anomaly into several components,

using the Gaussian filter as a separation method, allows the review of mainly
structural features and the analysis of density variation in depth (Fig. 6).
The regional anomaly derived from the Gaussian filter of standard deviation

144
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

Fig. 6. Successive decomposition of Bouguer anomaly, using the Gaussian filter.

equal to 0.02 cycle km−1 , shows a general linear trend extending from NNW
to SSE. The northern zone is dominated by a general gravity high oriented
east-west while the southern zone is characterized by a gravity low oriented
NW–SE. Those zones are separated by a middle region characterized by an
arch shaped high gradient following the NE–SW direction (Fig. 6). The
analysis of the pseudo-depth sliced gravity shows an increase in the size of
the anomalies and an attenuation of certain anomalies with depth.

145
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

The pseudo-depth slices 1, 2 and 3 correlate with the pseudo-depths of

0–490, 490–640, and 640–990 m, respectively. Their gravity effects show
dominate anomaly directions: E–W in the north, NW–SE and NE–SW in
the center and NW–SE in the south (Fig. 6). They show equally:
– An anomaly shape smoothing with depth; thus, the positive anomaly of
Sidi Dkril located in the East of Jebel Hairech and between Oued Mejarda
and Oued Mellègue, is composed of two anomalies on the pseudo-depth
slices 1 and 2, which merge into one in the pseudo-depth slice 3. This
expresses the gravity effect of a near surface fault affecting the source of
this anomaly but the lack of its persistence with depth suggests it is not
a major fault.
– Good correspondence between gravity anomalies and outcropping series
despite the exception of the uncorrelated anomalies observed in the Mio-
Pliocene and quaternary deposits. This correspondence shows a gen-
eral continuity of those outcropping series in depth to a pseudo-depth of
1000 m.
– Negative anomalies are associated with the Triassic outcrops of Jebel
Aiate (Oued Bou Adila), Jebel Debadib- Ben Gasseur and Mellegue dam
sector while other Triassic outcrops in the study area are associated with
positive anomalies.
– Positive anomalies occur within Mio-Plio-Quaternary deposits of Mejarda
basin. They extend for several kilometers along the E–W and NE–SW
directions. They seem to be related to dense materials similar to Tri-
assic, Jurassic and Cretaceous outcropping series, under thin Mio-Plio-
Quaternary overburden. This suggests the Mejarda basin is more com-
plex than a simple structural collapse basin. Notice in particular that
positive anomalies attenuate significantly with pseudo-depth greater than
2 km (Slice 5 of Fig. 6).
The pseudo-depth slices 4, 5 and 6 correspond to the pseudo-depth
ranges: 990 to 1390, 1390 to 2600 and 2600 to 4550 m, respectively. The
gravity maps associated with these slices are characterized by NW–SE and
NE–SW anomalies which become more dominant at greater depths (Fig. 6).
The pseudo-depth slices 4, 5 and 6 show:
– All anomalies associated with outcrop structures are totally attenuated
from the slice 5 or a pseudo-depth equal to 2600 m. Thus, the negative

146
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

anomalies on the Triassic outcrops of Jebel Aiate (Oued Bou Adila) and
Mellegue dam sector disappear at the pseudo slice 4, indicating a verti-
cal continuity of Triassic outcrop materials to a pseudo-depth equal to
1400 m. The negative anomaly associated with salt Triassic outcrops of
Jebel Ben Debadib- Gasseur disappears completely at the pseudo-depth
slice 5. This suggests that these outcrops are rooted to a pseudo-depth
of 2600 m.
– A progressive attenuation with depth of NW–SE negative anomalies asso-
ciated to El Marja, Oued Er Rmel and Zouarin basins. By pseudo-depth
slice 5 they are reduced to one single anomaly. A similar effect is ob-
served in the El Kef Plain, where the succession of positive and negative
anomalies associated with high and collapsed zones continue to pseudo-
depth slice 4; this succession is represent by a negative anomaly in the
gravity map of slice 5.
– Positive anomalies associated with dense structures under Mio-Plio-Qua-
ternary deposits attenuate with depth until they disappear entirely.
– The NW–SE anomalies of the middle domain are continuous with depth.
For pseudo- depth slices 1 through 4, the NE–SW trends have more
continuity and they truncate the NW–SE trends. But for depth slices 5
and 6, the NW–SE trend truncate the NE–SW anomalies.

3.4. Multiscale discontinuities and boundary identification

Next, the total horizontal gradient filter is applied to the gravity anoma-
lies of pseudo-depth slices resulting from the successive decomposition of
Bouguer anomaly. This permits identification of abrupt density variation
zones with depth (Fig. 7). The total horizontal gradient of the gravity
anomaly of deeper slice (7) (regional anomaly determiner with the Gaussian
filter of standard deviation equal to 0.02 cycle km −1 ) shows discontinuities
of density for depths greater than 4550 m. It shows an important NW–SE
discontinuity which corresponds with the deep fault identified by Arfaoui et
al. (2015). This fault cuts a NE–SW fault to the east of El Kef. These are
the dominant structural directions at depth for this study area and show no
correlation with the Triassic outcrops at the surface. But for the pseudo-
depth slice 6 (2600 m) some of the surface faults begin to become apparent.

147
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Fig. 7. Multiscale discontinuities and boundaries derived from the total horizontal gradi-
ent. Triassic outcrops is indicate in red color.

For example, the border faults of El Kalaa Khasba trough, the north fault
oriented E–W that ties this trough to the Ouled Boughanem trough, the
two faults limiting the Jebel Hairech structure, and the faults constituting
the southern limit of the Mejarda basin all show some expression (Fig. 7).
Both areas of the high total horizontal gradient surround the Ben Gasseur

148
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

Triassic structure; they indicate zones of abrupt density variations at this

depth. Besides, they provide the Triassic series disturbance in this area. At
this pseudo-depth the NW–SE fault is interrupted at El Kef apparently as a
result of the Triassic materials rising. The NW–SE and NE–SW structural
directions are still dominant over the middle domain structures (Fig. 7).
The pseudo-depth slice 5 (1400 m) shows:
– The south fault between the Ouled Boughanem and El Kalaa Khasba
troughs. This south fault is relatively shallower than the northern fault,
whose direction is close to E–W.
– N–S faults affecting the NW–SE and NE–SW fault of the southern border
of Mejarda basin and the E–W faults that begin to appear in this basin.
– The NE–SW faults affect NW–SE faults in the middle domain. This is
more indication of the change in the direction of the dominant faults
mentioned previously.
– The strike-slip faults limiting the structures of Jebel Hairech, Kef El Jebel
Agab and Chouichia, which evolved particularly to the SE.
– The discontinuities bordering the Ben Guasseur Triassic structure extend
northwards towards Jebel Debadib. Extension of this area to the east is
highly probable including the Zag Et Tir sector. Similarly, the bounding
faults of the Sidi Dkril positive gravity anomaly in SE of Jebel Hairech
appear to extend to this depth.
At the pseudo-depth slice 4 (1000 m), NE–SW faults dominate struc-
tures in the middle domain. They cut the NW–SE faults, especially on the
emplacement of the deep NW–SE fault. The deepest NW–SE fault cuts a
NE–SW fault at depth (slice 7), but at shallower depths (slice 4, 3.) the
fault systems are changing. NE–SW discontinuities affect the NW–SE dis-
continuities suggesting a change of tectonic regime or a variation of its main
direction.
At this pseudo-depth (1000 m), Triassic outcrops of Jebel Debadib in-
cluding Ben Gasseur, the Mellègue Dam sector and Jebel Aiate (Oued Bou
Adila) are clearly outlined by zones of maximum total horizontal gradient
reflecting the vertical continuity of these structures. The two parallel E–W
discontinuities in the Mejarda basin extend to the east for a greater distance.
The bounding faults of the Jebel Hairech structure appear to be effected by
other NE–SW faults. NW–SE faults begin to manifest in El Kalaa Khasba

149
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

trough (Fig. 7).

Moving to pseudo-depth slices 3 and 2 (less than 650 m), we see an
increase in the number of discontinuities with a decrease of their lengths.
With the exception of the bounding faults of El Kalaa El Khasba trough,
all the faults discussed previously are cut by other faults. These include:

– The E–W Faults limiting Jebel Hairech that are affected by NE–SW and
NW–SE faults.
– The E–W faults of Mejarda Basin.
– The faults bordering mini-basins of El Merja, Oued Er Rmel and Zouar-
ine.
– The faults of E–W branch depression linking the El Kalaa El Khasba and
Ouled Boughanem troughs.

Thus, the middle domain is characterized by two fault types differenti-

ated by their directions: NE–SW and NW–SE. The NE–SW faults are the
most dominant and they affect the NW–SE ones. Compared to the middle
and southern domains, the northern domain is characterized by deep quiet
structures that influenced by the activity of the EW and NE–SW faults
at relatively superficial depth (Fig. 7). The multi-scale analysis using the
total horizontal gradient of gravity slices allows for delineation of faults or
geological contacts.

3.5. Apparent density pseudo-section derived from successive de-

composition
The apparent density operator is applied generally, to generate density maps
(Gupta and Grant, 1985; Granser et al., 1989; Gupta and Sutcliffe, 1990;
Keating, 1991; Singh et al., 2003; Arfaoui et al., 2011). In our case we ap-
plied this operator to the gravity anomalies of pseudo-depth slices resulting
from the successive decomposition of Bouguer anomaly, to generate density
maps of pseudo-depth slices.
The process used to evaluate pseudo-sections describing density distri-
bution at depth, consisted on the extraction of the apparent densities data
of each profile from the apparent density maps and the attribution of a
pseudo depth of the anomaly separation interface to these data. After, they
are gridded by profile to generate apparent density pseudo-section.

150
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

Five isolated profiles of apparent density pseudo-sections are selected to

examine important structural features (Fig. 8). The profiles adopted for this
approach cross the mainly geologic structures of the study area: Mejarda
Basin, salt structure of Jebel Debadib, El Marja-Oued Er Rmel and Zouar-
ine basins and The Kalaa El Khaba through. The pseudo-sections PI and
PII cross the positive and negative Bouguer anomalies of Mejarda basin.
They show the presence of dense structures under the Mio-Plio-Quaternary
overburden.
Indeed, the Sidi Dkril Bouguer anomaly is associated with a dense struc-
ture whose density varies between 2.38 and 2.48 g/cm 3 . It is covered
by deposits of 500 m thickness and a density ranging between 2.24 and
2.28 g/cm3 , (Fig. 9).
Two other high zones are located northwest of Oued Mliz and at Jebel
Bou Rbah. They are covered by 500 m of Mio-Plio-Quaternary. However,
the Oued Mliz and Oued Rarai areas and the southeastern part of Jebel Bou
Rbah are characterized by thick Pliocene-Quaternary deposits with 2000 m
of thickness (Fig. 10). The presence of Cretaceous and Eocene series at
the base of collapsed zones is possible, especially since the density ranges
between 2.34 to 2.36 g/cm3 . Directly below the high zones, the density
decreases between 2000 and 4000 m depth (Fig. 10).
The pseudo-section density passing through Jebel Debadib shows a low
density diapir shape with a rooting depth of 2000 m and a width of 4000 m
(Fig. 11). At surface the density decreases from west to east (2.30 g/cm 3 to
2.23 g/cm3 ), reflecting the heterogeneity of Triassic materials in this part
of the diapir compared to the basal of the diaper portion that is homoge-
neous with a density less than 2.23 g/cm 3 (Fig. 11). The basal part of the
diapir shows a Triassic series dip to the southeast which could explain the
reduction of the rooting depth (1000 m) revealed previously in the Trias-
sic outcrops of Jebel Aiate and Mellègue dam sector located further in the
north.
The pseudo-density section of profile IV shows the structures of succes-
sive mini-basins of El Merja, Oued Rmel and Zouarine. It indicates an
overburden of densities less than 2.28 g/cm 3 with thicknesses varying be-
tween 500 and 800 m. This overburden consisting of Mio-Pli-Quaternary
deposits hides relatively dense terrains (2.32 g/cm 3 ) that are attributed to
Eocene and Cretaceous rocks (Fig. 12).

151
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Fig. 8. Location of profiles used for the pseudo-section density imaging.

152
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

Fig. 9. Pseudo-section density of profile PI, Mejarda basin.

Fig. 10. Pseudo-section density of profile PII, Mejarda basin.

153
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Fig. 11. Pseudo-section density of profile PIII, salt structure of Jebel Debadib.

Fig. 12. Pseudo-section density of profile PIV, successive basins El Marja, Oued Er Rmel
and Zouarine.

154
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

The distribution of the density in El Kalaa Khasba trough shows terrains

of low density at the center (2.25 g/cm 3 ) traversed by relatively dense series
(2.35 g/cm3 ) at a depth of 1000 m (Fig. 13). This distribution could indicate
the presence of Cretaceous and Triassic materials at depth. The density
distribution also shows the bounding faults dip towards the center of the
trough and that they are stopped at 2500 m depth. At this point two
contacts appear to reverse toward the west and the east. These contacts
combined with the density distribution beyond a depth of 2500 m, provide
the appearance of a dome having two clearly distinguishable horizons defined
by their densities. The deeper horizon sited at a depth greater than 4000 m
has a density equal to 2.28 g/cm3 ; it is surmounted by a dense horizon
(2.34 g/cm3 ), between the depths of 3500 m and 2500 m (Fig. 13). The
dome seems to be Triassic, and agrees with the rise of Triassic materials
started during the primary stages of formation of the Kalaa El Khasba
trough (Lehotsky and Bujnowsky, 1995).

Fig. 13. Pseudo-section density of profile PV, Kalaa El Khasba.

4. Accuracy of the Results

The gravity of pseudo-slices available by the successive decomposition of
Bouguer anomaly can be adopted to identify the discontinuities and geo-

155
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

logic contacts at different depths and to image density repartition on depth.

These constitute a rapid interpretive approach improved by the total hor-
izontal gradient and the apparent density operators. The results obtained
correlate with the majority of surface geologic units and model the gravity
structure effect as a density distribution following pseudo-section density.
However, certain sources of error associated with this approach can cause
discrepancies in results, as the depths of pseudo-slices and the background
density used to calculate the apparent density maps. Depths of pseudo-slices
are approximated by the spectral depth analysis method which corresponds
to mean depths of the causative sources of anomalies in residual maps. How-
ever, the causative sources associated with the residual anomalies exist at
different depths.
The densities used for transforming the gravity maps of pseudo-slices to
apparent density maps are chosen from the values of measured density of
surface rocks since we have no other control data. Background densities
adopted are chosen as an increasing function with depth, but they remain
a source of error which affects only the values of densities in the pseudo-
depth sections. They do not affect the image of geologic structure in the
pseudo-depth section resulting from the repartition of density since as the
apparent density operator calculates a contrast of density independently of
the background density, which is added at the end of the process as an av-
erage density.
Certainly, the resolution of the pseud-sections density and witch of multi-
scale discontinuities are affected by errors caused by the assessment process
of depth and density. Taking into consideration the gravity survey scale
(1/50000) and the spacing inter stations (1 km), we think that the results
reflect regional model with an acceptable resolution accuracy ratio of the
interpretative approach.

5. Conclusion
The multiscale interpretive approach based on the pseudo-depth slice gravity
anomalies derived from the successive decomposition of Bouguer anomaly
generally provided acceptable results compared to structural geological mod-
els in the study area. The use of depth spectral analysis makes this inter-
pretive approach more capable of revealing the evolution of fault systems

156
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

and causative source boundaries with depth and equally to establish 2D

pseudo-section density. The multiscale approach reveals deeper NW–SE
and NE–SW faults compared to the relatively superficial N–S and E–W
faults. It indicate that the excess of mass anomaly presented by the basins
of the study correspond to Triassic, Cretaceous and/or Eocene dense rocks
sited at relatively shallow depth under the Mio-Pliocene and Quaternary de-
posits. The salt Triassic outcrops of Jebel Debadib provide a diapir model
with a homogeneous basal layer dipped to southeast. The basement under
the Kalaa El Khasba trough is a Triassic dome.

Acknowledgements. We would like to express our thanks to the Office National

des Mines (Tunisia) for permission to use the gravity data of the study region. Many
thanks also to the technical editor and reviewer for their constrictive suggestions that
improved the final version of this manuscript substantially.

References
Amiri A., 2013: Apport de la géophysique appliquée à la modélisation géodynamique de
la moyenne valée de Mejarda. PhD. Thesis., Université Tunis El Manar, Faculté
des sciences de Tunis (in French).
Arfaoui M., Inoubli M. H., Tlig S., Alouani R., 2011: Gravity analysis of salt struc-
tures. An example from the El Kef-Ouargha region (northern Tunisia). Geophysical
Prospecting, 59, 576–591.
Arfaoui M., Inoubli M. H., 2013: Advantages of using the kriging interpolator to estimate
the gravity surface, comparison and spatial variability of gravity data in the El Kef-
Ouargha region (northern Tunisia). Arabian Journal Geosciences, 6, 3139–3147.
Arfaoui M., Reid A., Inoubli M. H., 2015: Evidence for a new regional NW–SE fault and
crustal structure in Tunisia derived from gravity data. Geophysical Prospecting,
63, 1272–1283.
Bansal A. R., Dimri V., 2001: Depth estimation from the scaling power spectral density
of nonstationary gravity profile. Pure and Applied Geophysics, 158, 799–812.
Ben Haj Ali M., 1997: Carte géologique au 1/50000 de Jendouba. Feuille No 32. Office
National des Mines. Tunisie.
Blakely R. J., Simpson R. W., 1986: Approximating edges of source bodies from magnetic
or gravity anomalies. Geophysics, 51, 1494–1498.
Burollet P. F., 1991: Structures and tectonics of Tunisia. Tectonophysics, 195, 359–369.
Burollet P. F., Sainfeld P., 1956a: Carte géologique au 1/50 000 d’El Kef. Direction des
Mines et de la Géologie. Tunisie.
Burollet P. F., Sainfeld P., 1956b: Carte géologique au 1/50 000 de Tajerouine. Feuille
No 50. Direction des Mines et de la Géologie. Tunisie.

157
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Chihi L., 1995: Les fossés néogènes à quaternaires de la Tunisie de la mer pélagienne: leur
signification dans le cadre géodynamique de la méditerranée centrale. PhD. Thesis.,
Université de Tunis El Manar (in French).
Chikhaoui M., 2002: La zone des diapirs en Tunisie: Cadre structural et évolution
géodynamique de la sédimentation méso-cénozoı̈que et géométrie des corps tri-
asiques. PhD. Thesis., Université de Tunis el Manar (in French).
Cooper G. R. J., Cowan D. R., 2006: Enhancing potential field data using filters based
on the local phase. Computers & Geosciences, 32, 1585–1591.
Cordell L., Grauch V. J. S., 1982: Mapping basement magnetization zones from aero-
magnetic data in the San Juan Basin New Mexico. Presented at the Snd. Ann.
Internat. Mtg. Sot. Explor. Geophys. Dallas, abstracts and biographies, 246–247.
Cordell L., Grauch V. J. S., 1985: Mapping basement magnetization zones from aeromag-
netic data in the San Juan basin New Mexico. In Hinze W. J. Ed. The utility of
regional gravity and magnetic anomaly maps Sot. Explor. Geophys, 181–197.
Fakhraoui M., Ghanmi M., Hatira N., 1998: Carte géologique au 1/50.000 De Nebeur.
Feuille No 39. Office National des Mines. Tunisie.
Fedi M., Florio G., 2001: Detection of potential fields source boundaries by enhanced
horizontal derivative method. Geophysical Prospecting, 49, 40–58.
Ghanmi M., Vila J. M., Ben Youssef M., Jouirou M., Ben Kherouf F., 2000: Le matériel
triasique interstratifié dans l’Albien de l’anticlinal autochtone atlasique du Jebel
Takrona (Tunisie): Stratigraphie Arguments gravimétriques signification dans la
transversale N–S des confins algéro-tunisiens (Maghreb du Nord-Est). Bulletin Sci-
ences Histoires Naturelles, 136, 19–27 (in French).
Gupta V. K., Ramani N., 1980: Some aspects of regional-residual separation of gravity
anomalies in a Precambrian terrain. Geophysics, 45, 1412–1426.
Gottis C., Sainfeld P., 1955: Carte géologique au 1/50.000 de Ghardimaou. Feuille No
31. Direction des Mines et de la Géologie. Tunisie.
Gupta V. K., Grant F. S., 1985: Mineral exploration aspects of gravity and aeromagnetic
surveys in the Sudbury-Cobalt area, Ontario. SEG; The Utility of Regional Gravity
and Magnetic Anomaly Maps, W. J. Hinze (Editor), 392–411.
Gupta V. K., Sutcliffe R. H., 1990: Mafic-ultramafic intrusives and their gravity field:
Lac des Iles area, northern Ontario. Geological Society of America Bulletin, 102,
1471–1483.
Granser H., Meurers B., Steinhauser P., 1989: Apparent density mapping et 3D gravity
inversion in the eastern Alps. Geophysical Prospecting, 37, 225–330.
Keating P., 1991: Density mapping from gravity data using Walsh transform. Geophysics,
57, 637–642.
Keating P., Pilkington M., 2004: Euler deconvolution of the analytic signal and its appli-
cation to magnetic interpretation. Geophysical Prospecting, 52, 165–182.
Last B. J., Kubik K., 1983: Compact gravity inversion. Geophysics, 48, 713–721.
Lehotsky I., 1979: Carte géologique au 1/50.000 d’Ebba Ksour. Feuille No 52. Office
National des Mines. Tunisie.

158
Contributions to Geophysics and Geodesy Vol. 48/2, 2018 (133–160)

Lehotsky I., Bujnowsky A., 1995: Carte géologique au 1/50.000 de Kalaat Es Senan.
Feuille No 59. Office National des Mines. Tunisie.
Mahjoub K., 1997: Carte géologique au 1/50.000 de Les Salines. Feuille No 45. Office
National des Mines. Tunisie.
Mahjoubi H., 1978: Un exemple de gisement ferrifère en un milieu recifal, Jerissa (Tunisie).
PhD. Thesis., Univesité de Tunis.
Mareschal J. C., 1985: Inversion of potential field data in Fourier transform domain.
Geophysics, 50, 685–691.
Maus S., Dimri V., 1996: Depth estimation from the scaling power spectrum of potential
fields? Geophysical Journal International, 124, 113–120.
Miller H. G., Singh V., 1994: Potential field tilt – A new concept for location of potential
field sources. Journal of Applied Geophysics, 32, 213–217.
Murthy I. V. R., Rao P. R., 1993: Inversion of gravity and magnetic anomalies of two-
dimensional polygonal cross sections. Computer Geosciences, 19, 1213–1228.
Muzaffer O. A., Ünal D., 2013: Edge detection of magnetic sources using enhanced total
horizontal derivative of the tilt angle. Bulletin of the Earth Sciences Application
and Research, Centre of Hacettepe University.
Negi J. G., Dimri V. P., Agrawal P. K., Petey O. P., 1986: A spectral analysis of the
profiles for thickness estimation of flood basalt of India. Exploration Geophysics,
17, 105–111.
Nettleton L. L., 1942: Gravity and magnetic calculation. Geophysics, 7, 293–310.
Ould Begga M. A., 2003: Evolution tectono-sédimentaire et analyse structural d’un seg-
ment des magrébides orientales: la région de Ghardimaou-Fernana (Tunisie Nord
occidentale). PhD. Thesis., Université El Manar, Faculté des sciences de Tunis (in
French).
Pawlowsky R. S., Hansen R. O., 1990: Gravity anomaly separation by Wiener filtering.
Geophysics, 55, 539–548.
Perthuisot V., 1978: Dynamique et pétrogenèse des extrusions triasiques en Tunisie
septentrionale. PhD. Thesis., Ecole Normale Supérieure Paris (in French).
Perthuisot V., Aoudjehane M., Bouzenoune A., Hatira N., Laatar E., Mansouri A., 1998:
Les corps triasiques des monts du Mellègue (confins algéro – tunisiens) sont-ils des
diapirs ou des “glaciers de sel”. Bulletin de la Société géologique de France, 169,
53–61 (in French).
Reid A. B., Allsop J. M., Granser H., Millett, A. J., Somerton I. W., 1990: Magnetic
interpretation in three dimensions using Euler deconvolution. Geophysics, 55, 80–
91.
Rouvier H., Henry B., Le Goff M., Hatira N., Laatar E., Mansouri A., 1998: Preuves
paléomagnétiques de la non-interstratification des évaporites du Trias dans l’Albien
du Maghreb oriental. Comptes Rendus de l’Académie des Sciences, 326, 363–368
(in French).
Sainfeld P., 1951 : Carte géologique au 1/50000 d’Ouargha. Feuille No 38. Direction des
travaux publics. Tunisie.

159
Arfaoui M. et al.: An optimal multiscale approach . . . (133–160)

Salem A., Ravat D., 2003: A combined analytic signal and Euler method (AN-EUL) for
automatic interpretation of magnetic data. Geophysics, 68, 1952–1961.
Simpson S. M., 1954: Least- squares polynomial fitting to gravitational data and density
plotting by digital computer. Geophysics, 19, 808–811.
Singh A. P., Michra D. C., Axman G., 2003: Apparent density mapping and 3D gravity
inversion of Dharwar crustal province. Journal of Indian Geophysical Union, 7, 1–9.
Smati A., 1986: Les Gisements de Pb-Ba et de Fe du Jebel Slata (Tunisie du centre-nord) :
minéralisations épigénétiques dans le crétacé néritique de la bordure d’un diapir de
Trias, gisements de Sidi Amor Ben Salem et de Slata-Fer. PhD. Thesis., Université
Pierre et Marie Curie Paris VI (in French).
Spector A., Grant F. S., 1970: Statistical models for interpreting aeromagnetic data.
Geophysics, 35, 293–302.
Talwani M., Worzel J. L., Letisman M., 1959: Rapid gravity computations for twodimen-
sional bodies with application to the Mendocino submarine fracture zone. Journal
of Geophysical Research, 64, 49–59.
Talwani M., Ewing M., 1960: Rapid computation of gravitational attraction of threedi-
mensional bodies of arbitrary shape. Geophysics, 25, 203–225.
Thompson D. T., 1982: EULDPH: A new technique for making computer assisted depth
estimates from magnetic data. Geophysics, 47, 31–37.
Thurston J. B., Smith R. S., 1997: Automatic conversion of magnetic data to depth, dip,
susceptibility contrast using the SPITM method. Geophysics, 62, 807–813.
Verduzco B., Fairhead J. D., Green C. M., Mackenzie C., 2004: New insights into magnetic
derivatives for structural mapping. The Leading Edge, 23, 116–119.
Vila J. M., Ben Youssef M., Charriere A., Chikhaoui M., Ghanmi M., Kamoun F., 1994:
Découverte en Tunisie, au SW du Kef, de matériel salifère triasique interstratifié
dans l’Albien: Extension du domaine à “glacier de sel” sous-marin des confins algéro-
tunisiens. Comptes Rendus de l’Académie des Sciences, 318, 1661–1667 (in French).
Vila J. M., Ben Youssef M., Chikhaoui M., Ghanmi M., 1996: Un grand “glacier de sel”
sous marin albien moyen du Nord-Ouest tunisien (250 Km2 ?): Le matériel salifčre
triasique du “diapir” de Ben Gasseur et de l’anticlinal d’El Kef. Comptes Rendus
de l’Académie des Sciences, 322, 221–227 (in French).
Werner S., 1953: Interpretation of Magnetic Anomalies at Sheet-like Bodies. Sveriges
Geologiska Undersokning, Arsbok, 43, 6.
Wijns C., Perez C., Kowalczyk P., 2005: Theta map: Edge detection in magnetic data.
Geophysics, 70, 39–43.
Zaier A., 1999a: Carte géologique au 1/50000 d’Ain Ksiba. Feuille No 60. Office National
des Mines. Tunisie.
Zaier A., 1999b: Evolution tectono-sédimentaire du basin phosphate du centre-ouest de
la Tunisie Minéralogie, pétrographie, géochimie et genèse des phosphorites. PhD.
Thesis., Université de Tunis II, Faculté des sciences de Tunis (in French).

160