Tectonophysics 599 (2013) 170–196

Contents lists available at SciVerse ScienceDirect

Tectonophysics
journal homepage: www.elsevier.com/locate/tecto

The crustal structure of the Central Mozambique continental margin — Wide-angle

seismic, gravity and magnetic study in the Mozambique Channel, Eastern Africa
Volker Thor Leinweber a,⁎, Frauke Klingelhoefer b, Sönke Neben a, 1, Christian Reichert c, Daniel Aslanian b,
Luis Matias d, Ingo Heyde c, Bernd Schreckenberger c, Wilfried Jokat a
a
Alfred-Wegener-Institut (AWI), Bremerhaven, Germany
b
Institut français de recherche pour l'exploitation de la mer (IFREMER), Brest, France
c
Bundesanstalt für Geowissenschaften und Rohstoffe (BGR), Hannover, Germany
d
Centro de Geofísica da Universidade de Lisboa (CGUL), Portugal

a r t i c l e i n f o a b s t r a c t

Article history: The continental margin of Mozambique formed during the initial dispersal of Gondwana about 180 Ma. Due
Received 24 May 2011 to the lack of deep seismic and dense potential field data, many details of the timing and geometry of the
Received in revised form 30 March 2013 early breakup in this region remained unknown to date. To close this gap, a research project (MoBaMaSis
Accepted 16 April 2013
(“Mozambique Basin Marine Seismic Survey”) with the French research vessel R/V Marion Dufresne II was
Available online 29 April 2013
conducted in 2007. This paper presents the results of P-wave, magnetic and 2D-gravity modelling along
Keywords:
two parallel seismic refraction profiles between 37° and 41° E, crossing the Mozambique rifted margin.
Crustal structure of Central Mozambique The crust shows the characteristics of normal to slightly thickened oceanic crust. A lower crustal high-
continental margin velocity-body with P-wave-velocities of 7.0–7.5 km/s is observed along both profiles. Its origin is discussed
Continent–ocean-boundary in the in the context of upper mantle convection and thermal properties. The existing magnetic anomaly identifica-
Mozambique Channel tions have been extended to older ages. We postulate that the oldest oceanic crust near the Central Mozam-
Africa–Antarctica Corridor bique continental margin has been formed around M41n (166 Ma). Closer to the coast a pronounced
Wide-angle seismic data negative magnetic anomaly exists that we interpret to coincide with the continent–ocean-transition. This im-
High-velocity-body
plies that the position of the continent–ocean-transition is located significantly closer to the shoreline than
Gondwana breakup
proposed before.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction Jurassic movements after Gondwana breakup are still controversial.

The assembly of the supercontinent Gondwana was completed
around 500 million years ago (Sahu, 2001). Around 300 Ma, it
merged with other continent masses, forming the supercontinent
Pangaea. In the Early Jurassic, the parts of Gondwana started to dis-
perse (for example Dalziel, 1992; Eagles and König, 2008; Grunow
et al., 1991; Jokat et al., 2003; Lawver et al., 1991). This process was
initiated by rifting between the western part of Gondwana (South
America and Africa) and its eastern part (Antarctica, India, Madagascar,
Sri Lanka, Australia and New Zealand) about 180 Ma.
Numerous publications discuss the general plate movements of the
southern continents since the breakup of Gondwana using various geo-
physical data sets (Bergh, 1977; Eagles and König, 2008; Ghidella et al.,
2007; Jokat et al., 2003; König and Jokat, 2010; Martin and Hartnady,
1986; Norton and Sclater, 1979; Roeser et al., 1996). While the Cenozoic
kinematic history between Africa and Antarctica is well constrained
(Bernard, 2005), the pre-breakup fit of the continents as well as the
⁎ Corresponding author.
E-mail address: Volker.Thor.Leinweber@awi.de (V.T. Leinweber).
1
Deceased.
This is mainly because of the lack of high-quality geophysical data,
which could reveal the location of the continent–ocean-transition. Clos-
er to the coast of Mozambique only commercial seismic reflection data
exist, which do not provide any constraints on the deep crustal fabric or
on the position and structure of the continent–ocean-transition. Mod-
ern deep seismic sounding data are absent on the conjugate East African
and Antarctic margins and potential field data characterising the oldest
oceanic crust are still sparse.
The modern day geographic setting of the study area and a simpli-
fied overview over the terranes of southern Africa is shown in Fig. 1.
The only stripe of seafloor providing direct evidence of the move-
ments between Africa and Antarctica since Mesozoic times, is the
Africa–Antarctica corridor (AAC). The Mozambique and Somali basins
are the oldest African oceanic basins formed during the initial break-
up of Gondwana. Both basins are separated by the NNW–SSE trending
Davie Fracture Zone, which is considered by many authors as the fos-
sil transform fracture zone, along which the southward drift of
Madagascar from a more northerly position during Gondwana break-
up occurred (Bunce and Molnar, 1977; Heirtzler and Burroughs,
1971; Rabinowitz et al., 1983; Scrutton et al., 1981; Segoufin and
Patriat, 1981). In the northwest of the Mozambique Channel, south

0040-1951/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.tecto.2013.04.015
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196
Fig. 1. Geographic and tectonic setting of the Mozambique Channel. Major terranes and
physiographic features are labelled. The black lines in the Mozambique Channel mark
the position of our model lines (see Fig. 2) and their magnetic profile prolongations to
the south. The inset in the upper left corner shows the research area at a larger scale as
well as the Africa–Antarctica Corridor. The area with question marks (?) indicates the
zone that is in question to belong to the Africa–Antarctica Corridor. Red colours mark the
Karoo continental flood basalts. Rivers and lakes are blue. The pronounced magnetic anom-
aly pattern in South Africa is drawn in black. Precambrian and Archean terranes are plotted
with different grey shading. The thin black lines in the Mozambique Channel mark the
MoBaMaSis seismic refraction profiles used in the study with its southward magnetic pro-
file prolongations. Abbreviations: ABFZ: Andrew Bain Fracture Zone; AFZ: Agulhas Fracture
Zone; AP: Agulhas Plateau; BA: Beattie Magnetic Anomaly; BdI: Bassas da India; BH: Beira
High; CFB: Cape Fold Belt; Com: Comores; DFZ: Davie Fracture Zone; GB: Gariep Belt; KB:
Karoo Basalts; KC: Kaapvaal Craton; KhB: Kheiss Belt; LB: Limpopo Belt; LM: Lebombo
Monocline; MaB: Magondi Belt; MAD: Madagascar; MB: Mozambique Basin; MC: Mozam-
bique Channel; MCP: Mozambique Coastal Plains; MFZ: Mozambique Fracture Zone;
MozB: Mozambique Belt; MR: Madagascar Ridge; MSM: Mateke-Sabi-Monocline; MZR:
Mozambique Ridge; NNB: Namaqua–Natal Belt; NNV: Northern Natal Valley; RT:
Rehoboth Triangle; Sa: Sambesi; SB: Somali Basin; SC: Sambesi Canyon; SNV: Southern
Natal Valley; SWIR: South-west Indian Ridge; TB: Transkei Basin; ZB: Zambezi Belt; ZC:
Zimbabwe Craton.
Bathymetry is taken from GEBCO_08 (2003). The terranes are digitised after Hansen et al.
(2009), Nguuri et al. (2001), Sahu (2001) and CGS (2000).
of the Sambesi estuary, a pronounced basement high, the “Beira
high”, is located (Coster et al., 1989). Oceanwards, the subsea Sambesi
canyon heads east–southeast, bending to the south at the Davie Frac-
ture Zone at 20°S.
Onshore, the geology of Mozambique (Fig. 1) can be divided into
Precambrian and Phanerozoic terrains, whereby the northern and
western central parts are predominantly constituted of Precambrian
rocks (Jamal, 2003). This region belongs to the Mozambique Belt,
which extends between the Sambesi River and northern Kenia as
part of the East African Orogen (Kröner, 1977). It is composed of
Meso- to Neo-proterozoic high-grade gneisses, granulites, quartzites,
migmatites and granitoides (Afonso, 1976; Jamal, 2003; Kröner,
1977) and has been reworked in Pan-African times. The coastal plains
covering the southern and eastern central parts of Mozambique rep-
resent — including the adjacent continental shelf regions — the larg-
est African sedimentary basin south of the equator (De Buyl and
Florès, 1986; Förster, 1975). The basin is limited to the north by the
Mozambique Belt, to the north-west by the Precambrian Zimbabwe
171
(Rhodesian) Craton and to the west by the Kaapval Craton and the
volcanic Lebombo Monocline (Coster et al., 1989; see Fig. 1). These
plains are blanketed by Phanerozoic sediments (Jamal, 2003).
The rifting in the study area (Fig. 1) during the Gondwana breakup
was accompanied by massive volcanism, which formed the continental
Karoo flood basalts in Africa and the Ferrar flood basalts in Antarctica
(Cox, 1992; Jourdan et al., 2005). The most prominent parts of the con-
tinental Karoo flood basalts along the eastern African margin are the
Lebombo and Mateke-Sabi monoclines forming the northern and west-
ern termination of the Mozambique Coastal Plains (Fig. 1). In a smaller
scale, rift related Mesozoic magmatism can also be found at the coast of
central-northern Mozambique (Tectonic Map of Mozambique, Scale
1:2.000.000, 2001). Jourdan et al. (2005) report that the majority of
Karoo flood basalts were emplaced between ca. 184 and 178 Ma. In
Antarctica, radiometrical dating of the conjugate Ferrar basaltic prov-
ince (183.6 ± 1.0 Ma, Encarnacion et al., 1996) resulted in dates,
which are similar to those of the Karoo basalts, though the Ferrar basalts
erupted over a shorter time interval.
Offshore, the first seismic reflection studies in the Mozambique
Channel were conducted in the early 70s (Beck and Lehner, 1974;
De Buyl and Florès, 1986; Fortes and Kihle, 1983; Heirtzler and
Burroughs, 1971; Lafourcade, 1984; Lort et al., 1979; Mougenot et
al., 1986; Virlogeux, 1987). The earliest identification of east–west-
trending magnetic spreading anomalies in the Mozambique Basin
was reported by Segoufin (1978) and Simpson et al. (1979) describ-
ing the seafloor spreading from chron M0r to M22. Based on a denser
and extended newer ship-towed magnetic data set (Jokat, 2006),
König and Jokat (2010) generally confirmed these chron identifica-
tions. They extended the chron identifications up to M26n.4n in the
Mozambique Basin, and identified M22r close to 22°S 37.4°E as oldest
anomaly in the area north of the island of Bassas da India (Fig. 1).
However, the latter authors could not map the magnetic field closer
to the coast of Central Mozambique, since their survey terminated
at 21°S. Thus, the nature of around 440 km of crust between their
oldest identifications and the shoreline of Mozambique remained
unknown.
This paper presents the first deep seismic sounding profiles using
ocean bottom receiver stations across the rifted margin of Central
Mozambique. Additionally, we present 2D-gravity models along these
profiles and a magnetic model based on new identifications of sea-
floor spreading anomalies in the northern part of the Mozambique
Channel.
2. Data acquisition and processing
From 15th September to 26th October 2007, as a collaborative effort
between BGR (Bundesanstalt für Geowissenschaften und Rohstoffe,
Germany), IFREMER (Institut Français de Recherche pour l'exploitation
de la mer, France), IPEV (Institut Paul Emile Victor, France) and AWI
(Alfred-Wegener-Institut, Germany), the MoBaMaSis survey (“Mozam-
bique Basin Marine Seismic Survey”) was conducted using the French
research Vessel R/V Marion Dufresne II. Two seismic refraction profiles
were acquired offshore Mozambique (Reichert et al., 2008). The profile
set-up is shown in Fig. 2. Both profiles were extended landwards onto
Mozambique using portable land stations by the MoBaMaSis Terra
team. An airgun array composed of eight G-Guns with a total volume
of 67.2 l (4100 in.3) was used as seismic source. In total, 2684 shots
were fired with an interval of 60 s and a mean distance of 152 m be-
tween the shot points along profile 20070201. Along profile 20070202,
1268 shots were fired with the same interval and a mean shot distance
of 171 m.
Three types of ocean bottom instruments and two types of land
stations were used. On the profile 20070201, 11 OBS (Ocean Bottom
Seismometer) and 24 MicrOBS (Micro Ocean Bottom Seismometer)
were deployed. On land, four 3-channel REFTEK land stations (each
with 3 strings with 6 geophones per channel, resulting in 54 geophones
172 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

using both a KSS31M sea gravimeter from the BGR and a Micro-g
LaCoste Air–Sea gravity meter, which is permanently installed on R/V
Marion Dufresne II (Reichert et al., 2008).

3. Modelling

3.1. Magnetic data

Trying to extend the anomaly identifications of König and Jokat

(2010), we compiled the data of the AISTEK II cruise (Jokat, 2006) to-
gether with the MoBaMaSis magnetic data set as well as old ship data
taken from the NGDC database and re-picked the anomalies. Table 1
provides information about all lines used in this study. Constant
values were added to the magnetic data of the profiles as listed, to
allow for a common interpretation. For some older profiles from the
NGDC database, this value was quite high due to a very low zero
level of the magnetic data. For unknown reason, the MoBaMaSis
data are generally characterised by a low zero level, too (Reichert et
al., 2008). We used the dates and notations of the magnetic reversal
timescale of Gradstein et al. (2004). The identification of spreading
anomalies in Jurassic times is generally hampered by the strong de-
crease of the amplitudes within the “Jurassic Quiet Zone” (McElhinny
and Larson, 2003; Tivey et al., 2006). Thus, picking the minima and
maxima of the magnetic anomalies and referring to the mean ages of
the younger and older anomaly ends in the timescale seemed more pre-
cise to us. Magnetic modelling was conducted using the Matlab based
programme ModMag (Mendel et al., 2004).

Fig. 2. Model geometry of seismic refraction profiles 20070201 (eastern profile) and
20070202 (western profile). Yellow circles mark Ocean Bottom Seismometer (OBS) Table 1
or Ocean Bottom Hydrophones (OBH), orange circles mark MicrOBS (MO). Yellow tri- Ship tracks used for magnetic anomaly identifications with line code, line name, oper-
angles mark REFTEK seismic land stations and orange triangles mark LEAS seismic land ating ship, year of the cruise and constant shift value applied to the data to adjust their
stations. different levels.
Contour lines are drawn after GEBCO_08 (2003).
Code Line name Ship Year of cruise Data shift [nT]

Aa MD163_20070201 Marion Dufresne II 2007 +25

Ab MD163_20070202 Marion Dufresne II 2007 +25
Ac MD163_20070203 Marion Dufresne II 2007 +30
per station) and four 3-component HATHOR-3 seismic station of the
Ad MD163_20070206 Marion Dufresne II 2007 +50
French manufacturer LEAS (one seismometer per station) were used. Ae MD163_20070208 Marion Dufresne II 2007 +75
Profile 20070202 consisted of 8 OBS, 3 ocean bottom hydrophones Af MD163_200702M1 Marion Dufresne II 2007 +25
(OBH), five REFTEK and four LEAS seismic land stations. The land sta- Ag MD163_200702M4 Marion Dufresne II 2007 +30
Ah MD163_200702M5 Marion Dufresne II 2007 +25
tions recorded the seismic data with a sampling rate of 100 Hz. The ma-
Bb SO283-P07 Sonne 2005 +40
rine seismic stations recorded with a sampling rate of 250 Hz. The Bc SO283-P08 Sonne 2005 +40
entire length (distance between the southernmost shot and the north- Bd SO283-P09a Sonne 2005 +40
ernmost land station) was 528 km for profile 20070201 and 365 km for Be SO283-P09b Sonne 2005 +40
profile 20070202 (Fig. 2). The receiver spacing along both profiles var- Bf SO283-P10 Sonne 2005 +40
Bg SO283-P11 Sonne 2005 +40
ied on- and offshore between 10 and 18 km.
Bh SO283-P12a Sonne 2005 +40
After standard data processing, the ocean bottom instruments Bi SO283-P12b Sonne 2005 +40
were relocated by adjusting their positions until the shapes of the di- Bj SO283-P13 Sonne 2005 +40
rect waves were symmetrical. The geophone channels of the REFTEK Bk SO283-P14 Sonne 2005 +40
Bl SO283-P15 Sonne 2005 +40
stations were filtered using a zero phase Butterworth filter and
Bm SO283-P16 Sonne 2005 +40
stacked. Then, the data of all receiver stations were uniformly filtered Bn SO283-P17 Sonne 2005 +40
with a bandpass filter (corner frequencies 3–5–24–36 Hz). For dis- Bo SO283-P18 Sonne 2005 +40
play, the seismic data were reduced using a velocity of 8 km/s. The Bp SO283-P19 Sonne 2005 +40
data quality of the offshore and onshore stations was in general Bq SO283-P20 Sonne 2005 +40
Ca LUSI7DAR ARGO 1963 +20
good to excellent. Onshore, the stacked REFTEK data provided a better
Cc V1911 Vema 1963 +75
signal to noise ratio than the LEAS data. Cd V3619 Vema 1980 +450
A 3000 m digital streamer with 240 channels was used to record Ce V3501 Vema 1978 +448
coincident seismic reflection data. The data were standard processed Cf TD267 Thomas B. Davie 1971 +300
Cg DSDP25GC Glomar Challenger 1972 +410
on the vessel (Reichert et al., 2008). Magnetic data were continuously
Ci MDU02 Marion Dufresne I 1973 +45
acquired with a SeaSpy™ Gradiometer system consisting of two Cj MDU02 Marion Dufresne I 1973 +50
Overhauser magnetometer sensors and a Magson fluxgate sensor in Ck MDU02 Marion Dufresne I 1973 +45
between. The SeaSpy system was towed up to 1000 m behind the Cl MDU07 Marion Dufresne I 1975 +75
ship with a distance of 150 m between the two Overhauser sensors Cm MDU07 Marion Dufresne I 1975 +45
Cn MDU07 Marion Dufresne I 1975 +45
(Reichert et al., 2008). Gravity data were continuously recorded
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 173

3.2. Wide-angle seismic data Table 3a

Assignment of picked phases (see Table 2) to model layers in profile 20070201.

The wide-angle seismic data were forward modelled using the Layer of velocity model Reflected Phase Refracted Phase
2D-raytracing software RAYINVR (Zelt and Smith, 1992). For the 20070201 wave on identification wave in identification
model geometry set-up, straight lines between the first and the last top of layer number layer number

shot points were taken and extrapolated onshore. Perpendicularly Layers 1 and 2: first Pw (on top 12 Pg1 (in 31
projected positions of the land stations and of the relocated OBS (unconstrained) of Layer 1) Layer 2)
sediment layer and
onto a straight model line were used as model station locations. On-
second sediment layer
shore, it was often not possible to find suitable locations in direct pro- Layer 3A: third sediment P2aP 32 Pg2a 41
longation of the shot lines due to the lack of major roads. This resulted layer A
in larger perpendicular distances between the model lines and the Layer 3B: third sediment P2bP 32 Pg2b 51
real positions of some receiver stations. The final model lines were layer B
Layer 4: fourth sediment P3Pa (through 42 Pg3 61
chosen to extend the northernmost station and the southernmost
layer layer Pg2a)
shot position equally by 11 km for profile 20070201 and by 17.5 km P3Pb (through 52
for profile 20070202, resulting in a model line length of 550 km for layer Pg2b)
profile 20070201 and of 400 km for profile 20070202 (Fig. 2). Layer 6: sixth sediment P5P 62 Pg5 71
layer
The software ZP of B.C. Zelt (http://www.soest.hawaii.edu/users/
Upper crust land station Pc1a 85
bzelt/zp/zp.html) was used for picking the P-wave travel time arrivals. Upper crust OBS Pc1bP 72 Pc1b 81
From the OBS and MicrOBS the hydrophone channel was chosen for Layer: lower crust Pc2P 82 Pc2 91
modelling, because of its superior quality. The vertical component was Layer: High velocity Pc3P Pc3 91
also taken into consideration, where it provided more details. Seismic body
Layer: upper mantle PmP 92 Pn 99
reflection data acquired in parallel (Reichert et al., 2008) were used to
provide additional indications for the depths of layer boundaries in
the P-wave models. Picking errors between 50 and 200 ms (see
Table 2 for mean values) were individually assigned to each phase of density differences between the lowermost sediment and the upper-
every station depending on the signal to noise ratio. The phases were most crustal layer resulted in a low impedance contrast at the bound-
assigned to model layers as listed in Tables 3a and 3b. ary. Here, the combined interpretation of the OBS data with the
Pw marks the reflection on the seabed. To model the first simultaneously acquired multichannel seismic data provided excel-
refracted phases, it was necessary to introduce an uppermost sedi- lent constraints for the raytracing.
ment layer, which is not directly constrained by identified reflected
or refracted phases. P2P to P5P mark reflections on deeper layers, 3.3. Gravity data
which are interpreted as sediment layers. Pg2–Pg5/Pg7 phases repre-
sent waves, which are refracted in sediment layers. The phase, which To simulate the observed free-air gravity from shipboard measure-
represents waves travelling through the upper crustal layer of the on- ments, we used the module GMSYS of the Geosoft Oasis montaj™ soft-
shore part of the profiles, is labelled Pc1a. Pc1bP and Pc2P are the reflec- ware. In parts of the models, where no ship data were available (close
tions on top of the upper and lower crust. Pc1b and Pc2 phases represent to the coast and onshore), gravity data were taken from the free-air
waves, which are refracted in the upper or lower crustal layer of oceanic anomaly map of Sandwell and Smith (2009), version 18. The results of
or transitional crust. PmP is the Moho-reflection and the phase, which is seismic velocity modelling were taken as boundary conditions for the
refracted in the uppermost mantle, is labelled Pn. 2D density models. The models base on distinct blocks with assigned
The identification of phases constraining the offshore upper crust- constant densities. Starting with layer boundaries according to those of
al layer was sometimes difficult, because the apparent velocity- and the P-wave-models and typical crustal and sedimentary densities
(Christensen and Mooney, 1995; Ludwig et al., 1970; Nafe and Drake,
1957), we subsequently modified the models, until reasonable fits of
Table 2
the measured and calculated free-air anomalies were obtained. The
Phase identification numbers, numbers of receiver stations, for which the according
phase was identified and mean values of assigned picking errors for every phase, calcu-
lated over all stations, where the phase occurred.
Table 3b
Profile 20070201 Profile 20070202 Assignment of picked phases (see Table 2) to model layers in profile 20070202.

Phase Number Mean value of Phase Number Mean value of Layer of velocity model Reflected Phase Refracted Phase
ID of stations assigned picking ID of stations assigned picking 20070202 wave on identification wave in identification
error (calculated error (calculated top of layer number layer number
over all stations) [s] over all stations) [s]
Layers 1 and 2: first Pw (on top 12 Pg1 (in 31
12 29 0.050 12 11 0.055 (unconstrained) sediment of Layer 1) Layer 2)
31 26 0.057 31 11 0.059 layer and second sediment
32 26 0.089 32 9 0.092 layer
41 12 0.050 41 10 0.060 Layer 3: third sediment P2P 32 Pg2 41
42 12 0.148 42 8 0.100 layer
51 17 0.059 51 7 0.061 Layer 4: fourth sediment P3P 42 Pg3 51
52 13 0.104 52 5 0.100 layer
61 29 0.072 61 11 0.071 Layer 5: fifth sediment layer P4P 52 Pg4 61
62 5 0.095 62 10 0.095 Layer 6: sixth sediment P5P 62 Pg5 75
71 28 0.097 71 8 0.100 layer
72 29 0.112 72 10 0.113 Layer 8: eighth sediment P7P 71 Pg7 71
81 27 0.087 81 11 0.091 layer
82 4 0.075 82 5 0.110 Upper crust land station Pc1a 85
85 7 0.139 85 9 0.061 Layer: upper crust OBS Pc1bP 72 Pc1b 81
91 36 0.100 91 19 0.107 Layer: lower crust Pc2P 82 Pc2 91
92 35 0.118 92 18 0.124 Layer: High velocity body Pc3P Pc3 91
99 11 0.111 99 13 0.133 Layer: upper mantle PmP 92 Pn 99
174 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 3. Station 24 (MicrOBS) of profile 20070201. The data were bandpass filtered and reduced using a velocity of 8 km/s. A gain according to offset has been applied.

software extends the models on both sides to avoid edge effects. There-
fore, the resulting model-anomalies had to be shifted by a constant value.
The maximum absolute error of the modelled free-air anomaly is
4.09 mGal (mean absolute error: 0.49 mGal, standard deviation:
1.53 mGal) for model 20070201 and 5.71 mGal (mean absolute error:
0.25 mGal; standard deviation: 1.16 mGal) for model 20070202.
4. Results
Data examples of both profiles are shown in Figs. 3–8. Identified
phases are marked and labelled (see Tables 3a and 3b for details).
Fig. 3 shows station 24 (MicrOBS) of profile 20070201 (see Fig. 2).
The section is characterised by an asymmetry between the apparent

Fig. 4. Detailed view of identified phases of station 19 (MicrOBS), profile 20070201.

 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 175

Fig. 5. Seismic land station 08 (LEAS) of profile 20070201. The data were bandpass filtered and reduced using a velocity of 8 km/s. A gain according to offset has been applied.

velocities in the northwestern and the southeastern travel time
branches caused by seafloor topography. The Pn phase (a refracted
wave, turning in the upper mantle) is identified only in the north-
western part. On both sides, a gap in the travel time curve of 0.7 s is
observed between Pg3 and Pg5 caused by a low velocity sedimentary
layer.
The next example is the southeastern part of station 19 on profile
20070201 (MicrOBS, Fig. 4). Several sedimentary phases are readily
observed resulting in a larger number of model layers in the sediment
column. Even though detailed modelling of the sediments is not cru-
cial for our crustal study, the sedimentary part of our models might
provide constraints for future stratigraphic studies.

Fig. 6. Station 08 (OBS) of profile 20070202. The data were bandpass filtered and reduced using a velocity of 8 km/s. A gain according to offset has been applied.
176 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 7. Detailed view of identified phases of station 06 (OBS), profile 20070202.

Fig. 5 shows the land station 08 (LEAS) on profile 20070201. In
this seismic section, two crustal refracted phases (Pc1a and Pc2) as
well as the Moho reflection PmP were identified. Fig. 6 shows OBS 8
of profile 20070202. Similar to OBS 24 of profile 20070201 (see
Fig. 3), several sedimentary refractions and reflections can be identi-
fied. Both sides show PmP phases and weak Pn mantle refractions.
Between Pg5 and Pg7, a gap in the travel time curve, which is slightly
smaller (0.4 s) than at station 24 of 20070201, is visible in the south-
eastern part of the seismic record.
Fig. 7 shows a zoomed portion of OBS 06 on profile 20070202.
Here again, many sediment phases are visible, but without indica-
tions for a low velocity layer in the sedimentary column. The seismic

Fig. 8. Seismic land station 14 (REFTEK) of profile 20070202. The data were bandpass filtered and reduced using a velocity of 8 km/s. A gain according to offset has been applied.
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 177

Fig. 9. Magnetic profiles in Northern Mozambique Channel with picks of magnetic anomalies. The lines are labelled with identification codes, which are listed in Table 1. The
coloured data are the profiles for which the anomalies were modelled in this study (identification codes Af/Aa and Ah/Ab). Positive anomalies are drawn in red, negative ones
in blue. For all other data, light grey marks positive and dark grey marks negative anomalies. The identified magnetic chrons are listed at the right side with different symbols
and colours. The solid red lines are fracture zones corresponding to our identification of the magnetic anomalies.
Flowlines after Eagles and König (2008) are drawn with dashed lines. The dotted line marks the COB location published by Raillard (1990). Contour lines are drawn after GEBCO_08
(2003).

land station 14 of profile 20070202 (Fig. 8, REFTEK) has an initial
offset of 120 km, but shows clearly the lower crustal phases. In addi-
tion to the two crustal refraction phases Pc1a and Pc2 and the PmP
phase, the mantle refraction Pn can easily be identified in the seismic
record.
Our magnetic anomaly picks are shown in Fig. 9. The MoBaMaSis
magnetic profile 200702M1 is located at the southern prolongation
of profile 20070201 as is the case for 200702M5 and 20070202
(Fig. 9, profiles printed in colour). The magnetic profiles together
with labelled identified spreading anomalies and magnetic models
are shown in Fig. 10.
Our identifications conform to those of König and Jokat (2010) in
the northern Mozambique Channel, but extend them to the north and
to the northeast. The picks were used by Leinweber and Jokat (2012)
for calculating the initial fit of both continents. We identified the
Mesozoic sequence from M11.An.1n (137.1 Ma) to M33n (159.1) in
profile 20070201 and from M22n.1n (149.5 Ma) to M33n (159.1 Ma)
in profile 20070202. From M33n to the coast, the magnetic field
shows similar characteristics in both profiles: it slowly decreases by
50 nT over a distance of around 100 km (Fig. 10, km 300–200 in
20070201, and km 320–220 in 20070201). Finally, the magnetic field
decreases rapidly and forms a pronounced negative anomaly on both
profiles, before it sharply increases again (Fig. 10, between km 180
and 140 in 20070201, and between km 200 and 175 in 20070202).
Both profiles terminate close to the coast of Mozambique. The younger
edge of the strong negative anomaly was picked as M41n (165.6 Ma),
although this identification is tentative. It relies on an extrapolation of
the spreading velocity at M33n of profile 20070202, to some extent.
178 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 10. Magnetic model of profiles 20070201&2M1 and 20070202&2M5. The distances along the x-axis are identical with those of the P-wave models. Proposed fracture zones are
marked with “FZ”.

The profiles presumably cross a fracture zone respectively. One frac-
ture zone crosses profile 20070201 between anomaly M15n and M18n
and offsets the northern part of the profile by around 40 km in northern
direction. Another fracture zone crosses profile 20070202 between
anomaly M22n.1n and anomaly M23n and produces a southerly offset
of 40 km (Fig. 10).
The fit between the model and the observed magnetic data is
good. However, the correlation of the identified chrons between the
two profiles is tentative and has to be confirmed by a denser magnetic
data set.
5. Interpretation
5.1. Magnetic data
M41n (~166 Ma; timescale of Gradstein et al., 2004) is interpreted
as oldest magnetic spreading anomaly in the Mozambique Channel.
The parameters of the seafloor spreading model (Fig. 10) are listed in
Table 4. The model reveals spreading velocities increasing from older
M22n.1n (149 Ma) to M12.r.1r (138 Ma) in the area south of Bassas
da India. Our findings were included in a general Gondwana breakup
model by Leinweber and Jokat (2012).
5.2. P-wave and gravity models
All picks of the land and OBS stations are shown with the
modelled travel time arrivals in Fig. 11a and b. The number of picks,
the root-mean-square error of the modelling and the percentage of
successfully traced picks are given in Tables 5a and 5b for both pro-
files. The normalised χ 2-value is based on the chosen error for each
pick (Tables 5a and 5b). About 93% of the picks of arrivals in model
20070201 and about 90% of the picks in model 20070202 were suc-
cessfully traced. The values of the RMS traveltime residual indicate a
good fit between the modelled and picked arrivals.
Table 4
Seafloor spreading episodes in our magnetic model and their spreading half-rates.

to younger ages. For profile 20070201, they rise from 15.0 km/Myr Timespan 20070201 20070202
at 167 Ma to 28.0 km/Myr at 149 Ma. Along profile 20070202 they [Ma]
Profile distance Spreading Profile distance Spreading
increase from 16.0 at 167 Ma to 23.5 km/Myr at 153.5 Ma. From [km] halfrate [km] halfrate
167 to 159.6 Ma we found a slightly slower spreading half rate for [km/Myr] [km/Myr]
20070201 than for 20070202 (15.0 km/Myr versus 16.0 km/Myr) 167.0–159.6 190–295 15.0 210–415 16.0
and a higher spreading rate between 159.6 and 153.5 Ma (19.5 ver- 159.6–153.5 295–416 19.5
sus 16.0 km/Myr). Our spreading rates are comparable to those of 153.5–149.0 416–519 23.5 415–480/519–544 23.5
König and Jokat (2010), who found half-rates of 23.5 km/Myr from 149.0–141.0 519–625 22.0
141.0–134.0 661–758 28.0
M26n.1n (155 Ma) to M22n.1n (149 Ma), and of 21 km/Myr from
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 179

Fig. 11. a: Profile 20070201, all stations with picked arrivals (shown as black error bars) and modelled arrivals (yellow). b: Profile 20070202, all stations with picked arrivals (shown
as black error bars) and modelled arrivals (yellow).
180 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 11 (continued).
V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 181

Fig. 11 (continued).
182 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Table 5b
Quality of raytracing and modelling errors, profile 20070202.

Phase (RAYINVR-code Picks Successfully Successfully RMS χ2

of phase) traced picks traced picks [%] [s]

Pw (1.2) 1387 1384 99.8 0.018 0.132

Pg1 (3.1) 319 303 95.0 0.042 0.606
P2P (3.2) 709 555 78.3 0.059 0.399
Pg2 (4.1) 510 510 100.0 0.036 0.422
P3P (4.2) 969 748 77.2 0.071 0.509
Pg3 (5.1) 243 191 78.6 0.037 0.515
P4P (5.2) 403 310 76.9 0.125 1.566
Pg4 (6.1) 839 801 95.5 0.041 0.399
P5P (6.2) 768 739 96.2 0.060 0.417
Pg5 (7.1) 1127 1054 93.5 0.052 0.420
Pg7 and P7P (9.1 and 8.2) 429 344 80.2 0.056 0.424
Fig. 11 (continued).
Pc1a (11.1) 1059 950 89.7 0.097 0.948
Pc1bP (9.2) 2091 1950 93.3 0.058 0.303
Pc1b (10.1) 1109 1010 91.1 0.066 1.039
Pc2P (10.2) 208 108 51.9 0.070 0.320
The ray coverage with traced refracted and reflected rays is good
Pc2 and Pc3 (12.1 and 13.1) 4029 3653 90.7 0.081 0.590
for all sedimentary and crustal layers in both profiles (Fig. 12). In Pc3P and PmP (12.2 and 13.2) 4425 4087 92.4 0.102 0.822
both models only the uppermost mantle is covered by rays. The con- Pn (14.1) 936 763 81.5 0.132 1.331
tinental portions of the models are less well constrained due to the Total 21,560 19,460 90.3 0.079 0.621
absence of reversed shots.
In order to test our models, synthetic seismograms were calculated
and compared to the data sections. The finite difference modelling code
from the Seismic Unix package (Cohen and Stockwell, 2003; Stockwell, Comparisons to other nodes of the same layer allowed us to deter-
1999) was used to calculate synthetic seismograms (Figs. 13b, 14b mine the precision of the complete model layer (Tables 6a and 6b).
and 15b). The software uses the explicit second-order differencing The resolution of the models (Zelt and Smith, 1992) is presented in
method for modelling the acoustic wave equation. The input veloc- Fig. 16.
ity model was calculated from sampling the forward velocity model
at a lateral 25 m interval and 10 m interval in depth. In order to
avoid grid dispersion, the peak frequency of the Ricker wavelet 5.3. Profile 20070201, northeastern profile
source signal was calculated to be equal to the lowest velocity of
the medium divided by the grid points per wavelength multiplied The final velocity model of profile 20070201 together with the ob-
by 10. In this case the source wavelet was centred at 8 Hz, similar served magnetic and gravity data is shown in Fig. 17. A detailed inter-
to the signal from the air gun array used during the cruise. The pretation of the sedimentary layers shall be part of a separate study.
boundary conditions were set to be absorbing at the sides and bot- Therefore, these layers are not discussed in detail here. The model
tom of the model and free at the surface. Figs. 13–15 show that layer velocities and maximum thicknesses as well as the assigned
the synthetic data are in good agreement with the observed seismic densities are summarised in Table 6a. It is important to note that,
data. based on travel time gaps in eight OBS record sections (OBS 18–25
We performed error estimations of the modelled layer velocities and and OBS 23A) a low velocity layer has been modelled in the sedimen-
the depth of the layer boundary nodes (Barton and White, 1997). tary column. Since a low velocity layer does not provide any refracted
A boundary or a velocity node was perturbed until the resulting fit arrival to calculate its interval velocity, we chose a constant velocity
between the observed and the calculated phase was no longer of 4.0 km/s, which is slightly lower than the lowermost velocity at
acceptable considering the picking error. The perturbation was the bottom of the layer above. The thickness of the low velocity
then taken as the model error for that node. This was performed layer ranges from 0.2 km to 1.0 km (Fig. 17, around km 230). Below
for some representative depth and velocity nodes of each layer. this layer, the last sedimentary unit above the top of the crystalline
crust was identified. Here, velocities range from 4.5 km/s to 5.0 km/s.
Table 5a
The shallow layer velocities of the onshore part of the line could
Quality of raytracing and modelling errors, profile 20070201. not be determined by the profile layout. The minimum offset between
the airgun shots and the receiver stations on land was 50 km. There
Phase (RAYINVR-code Picks Successfully Successfully RMS χ2
were no reverse shots from Mozambique. However, from onshore
of phase) traced picks traced picks [%] [s]
geological mapping no significant sediments except of a coastal stripe
Pw (1.2) 5550 5532 99.7 0.018 0.125
(De Buyl and Florès, 1986; Jamal, 2003; Tectonic Map of Mozambique,
Pg1 (3.1) 1364 1319 96.7 0.042 0.641
P2aP and P2bP (3.2) 3354 3059 91.2 0.101 1.081 2001) are reported. Mostly, Precambrian rocks were mapped at the
Pg2a (4.1) 1179 1154 97.9 0.060 1.441 surface in our research area. The coastal sediment units were not in-
P3Pa (4.2) 2470 2300 93.1 0.086 0.397 cluded in the P-wave models, because the region in question was
Pg2b (5.1) 1176 1053 89.5 0.055 0.961 not constrained by rays and no detailed information about the
P3Pb (5.2) 2787 2302 82.6 0.073 0.575
shape and thickness of the sediment basin were available for this
Pg3 (6.1) 3064 3004 98.0 0.046 0.395
P5P (6.2) 269 267 99.3 0.071 0.573 study. Yet for modelling the gravity field, the assumption of sediment
Pg5 (8.1) 2724 2162 79.4 0.050 0.415 units was plausible. The uppermost P-wave-model layer in the on-
Pc1bP (8.2) 5393 5356 99.3 0.059 0.295 shore part of the model has upper continental crust velocities ranging
Pc1b (9.1) 1231 1133 92.0 0.046 0.310
between 5.7 and 5.8 km/s and lower boundary velocities between
Pc2P (9.2) 143 60 42.0 0.038 0.144
Pc1a (10.1) 570 543 95.3 0.063 0.186 6.0 km/s and 6.3 km/s. The maximum thickness is 5.5 km. A model
Pc2 and Pc3 (11.1 and 12.1) 9919 9243 93.2 0.061 0.367 layer with a maximum thickness of 26 km represents the middle con-
Pc3P and PmP (11.2 and 12.2) 10,421 9390 90.1 0.057 0.247 tinental crust. It has upper velocities of about 6.2 km/s and lower ve-
Pn (13.1) 1009 878 87.0 0.086 0.587 locities around 6.7 km/s, decreasing to 6.3 km/s when approaching
Total 52,623 48,755 92.7 0.061 0.416
the edge of the layer at km 140.
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 183

Fig. 12. Ray coverage in the P-wave models.

Offshore, the uppermost crustal layer thickness ranges between lower crust shows upper boundary velocities varying laterally from
1.0 and 1.6 km. The layer has velocities between 5.6 and 6.0 km/s at 6.7 km/s beneath the continent to 6.4 km/s seawards. The lower
its upper and between 5.9 and 6.2 km/s at its lower boundary. The boundary velocity of the layer is uniformly 7.0 km/s. In a large part
184 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 13. a) Band-pass-filtered (3/5 Hz, 24/36 Hz) hydrophone channel from OBS 24 of profile 20070201. The data are displayed with a gain proportional to source–receiver offset and are
reduced at a velocity of 8 km/s. PmP and Pn phases are annotated; and b) synthetic seismogram calculated from the velocity model for the same station using the finite difference modelling
code from the Seismic Unix package (Cohen and Stockwell, 2003; Stockwell, 1999]. The synthetic seismogram was calculated every 100 m with a source frequency centred around 8 Hz.

of the model the lower crust is underlain by a high-velocity-body
(HVB), defined by an upper boundary velocity of 7.0 km/s. These high
velocities are constrained by first onsets of a continuous refracted
phase, which was subdivided in phase Pc2 (velocities up to 7.0 km/s)
and Pc3 (velocities higher than 7.0 km/s). The lower boundary velocity
of the HVB reaches its maximum value of 7.5 km/s around km 180.
However, the major portion of the HVB has velocities between 7.0 and
7.3 km/s. Having a maximum thickness of 8 km near the shelf edge at
km 150, the HVB thins to the south and pinches out at km 450. The
upper mantle has velocities of 8.0–8.1 km/s. The Moho shows signifi-
cant depth variations at km 300 and around km 440.
The results of the gravity modelling of profile 20070201 are shown
in the lower part of Fig. 17. Onshore, the upper crust was modelled
with a density of 2.69 g/cm3, the middle continental crust with a
density of 2.91 g/cm 3 and the lower continental crust with a density
of 3.05 g/cm3. These values are in good agreement with typical
density-values for continental crust (Christensen and Mooney, 1995).
A sedimentary basin with a maximum thickness of 4.5 km was intro-
duced to model the gravity low at km 100. The density of the sediments
is 2.00 g/cm 3. In the P-wave model, this sedimentary basin was not in-
cluded, because it is mainly situated in an area not covered by rays (see
P-wave model in Fig. 17) and due to the fact, that neither shots were
fired from land nor seismic reflection data were available to provide fur-
ther constraints for the detailed structure of the upper continental crust
and the sedimentary basin. Therefore, the structure and depth of this
basin is only constrained by the ambiguous gravity modelling approach.

Fig. 14. a and b) Band-pass-filtered (3/5 Hz, 24/36 Hz) hydrophone channel from OBS 06 of profile 20070202 with according synthetic seismogram, calculated as described in the
caption of Fig. 13.
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 185

Fig. 15. a and b) Band-pass-filtered (3/5 Hz, 24/36 Hz) hydrophone channel from OBS 08 of profile 20070202 with according synthetic seismogram, calculated as described in the
caption of Fig. 13.

Generally, the lack of constraints for the continental sedimentary
and crustal layers as well as the fact that the onland and offshore
data come from different data sets make the continental part of the
gravity model very ambiguous. These remarks are also valid for
model 20070202.
Offshore, the sedimentary densities range from 1.95 g/cm3 near the
seafloor increasing downwards to 2.60 g/cm3 above the top of the crys-
talline crust. The upper crust (offshore) was assigned with a density of
2.67 g/cm3. The lower crust was modelled with a density of 2.90 g/cm3
from km 140–460 and of 2.80 g/cm3 from km 460 to 550.
The HVB has been divided in several density blocks, ranging from
3.05 to 3.20 g/cm 3. These changes correlate roughly to the velocity
distribution inside the high-velocity-body (7.0–7.5 km/s). The mantle
was modelled with a density of 3.27 g/cm 3. The model densities are
summarised in Table 6a.
5.4. Profile 20070202, southwestern profile
This layer 7 has a maximum thickness of 0.9 km and velocities of
4.4, locally up to 4.7 km/s. Sediment layer 8 just below (directly
above the crystalline crust) has upper boundary velocities between
4.6 and 4.9 km/s and lower boundary velocities between 4.7 and
5.0 km/s, which are very similar to the upper crustal velocity. To dis-
tinguish both units, information from the multichannel seismic data
were taken into account.
For the continental part of the profile a simple velocity distribu-
tion was chosen again, because of the lack of reversed shots:
• upper continental crust with upper boundary velocities ranging
from 5.6 to 5.9 km/s, lower boundary velocities ranging from 6.0
to 6.3 km/s and a thickness varying between 1.8 and 4.7 km
• middle continental crust, represented by an up to 32 km thick body
(Fig. 18, km 0–175) with minimum velocities of 6.2–6.3 km/s and
lower velocities between 6.6 and 6.7 km/s
• lower continental crust with velocities of 6.4–6.8 km/s and lower
velocities of 6.9 km/s and a thickness up to 8 km.

The final velocity model of profile 20070202 together with the Offshore, the upper crustal velocities range between 5.7 and 6.0 km/s
observed magnetic and gravity data is shown in Fig. 18. As in model at the upper and between 5.9 and 6.1 km/s at the lower boundary. The
20070201, a low velocity layer in the sedimentary column is observed. thickness of this layer varies between 0.9 km and 1.5 km. The lower

Table 6a
Properties of model 20070201.

Layer Type Max. thickn. [km] P-wave-vel. [km/s] Main density [g/cm3] Upper bound. uncertainty [km] Velocity uncertainty [km/s]

Water layer Water 2.9 1.5 1.03 ±0.0 ±0.01

Sediment layer 1 Sed. 0.9 1.5–2.0 1.95 ±0.05 ±0.1
Sediment layer 2 Sed. 2.0 2.4–2.9 2.18 ±0.1 ±0.1
Sediment layer 3A Sed. 1.9 3.0–3.5 2.46 ±0.1 ±0.1
Sediment layer 3B Sed. 2.4 2.9–4.0 2.36–2.46 ±0.1 ±0.1
Sediment layer 4 Sed. 2.7 4.0–4.7 2.57 ±0.1 ±0.1
Sediment layer 5 Sed. 1.0 4.0 2.59 ±0.2 ±0.1
Sediment layer 6 Sed. 2.4 4.5–5.0 2.60 ±0.2 ±0.1
Continental sediments Sed. 4.5 – 2.00 – –
Upper crust onshore Cont. 5.5 5.7–6.3 2.69 ±0.1 ±0.1
Middle crust onshore Cont. 24.0 6.2–6.7 2.91 ±2.0 ±0.1
Upper crust offshore Trans./oc. 1.6 5.6–6.2 2.67 ±0.5 ±0.1
Lower crust Cont./trans./oc. 9.0 6.4–7.0 2.80–3.05 Cont.: ±1.0 ±0.1
Trans./oc.: ±0.5
High velocity body Und./serp. 8.0 7.0–7.5 3.05–3.20 ±0.5 ±0.1
Upper mantle Mantle – 8.0–8.1 3.27 ±1.0 ±0.1
186 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Table 6b
Properties of model 20070202.

Layer Type Max. thickn. [km] P-wave-vel. [km/s] Main density [g/cm3] Upper bound. uncertainty [km] Velocity uncertainty [km/s]

Water layer Water 2.6 1.5 1.03 ±0.0 ±0.01

Sediment layer 1 Sed. 0.9 1.5–2.3 1.95 ±0.05 ±0.1
Sediment layer 2 Sed. 0.8 2.3–2.6 2.18 ±0.1 ±0.1
Sediment layer 3 Sed. 1.8 2.6–3.2 2.39–2.46 ±0.1 ±0.1
Sediment layer 4 Sed. 1.8 3.1–3.7 2.40 ±0.1 ±0.1
Sediment layer 5 Sed. 2.4 3.4–4.5 2.51–2.58 ±0.1 ±0.1
Sediment layer 6 Sed. 1.3 4.5–4.9 2.62 ±0.2 ±0.1
Sediment layer 7 Sed. 0.9 4.4–4.7 2.60 ±0.2 ±0.1
Sediment layer 8 Sed. 2.6 4.6–5.0 2.64 ±0.2 ±0.1
Continental sediments Sed. 1.5 – 2.00 – –
Upper crust onshore Cont. 4.7 5.6–6.3 2.69 ±0.1 ±0.1
Middle crust onshore Cont. 32.0 6.2–6.7 2.91 ±2.0 ±0.1
Upper crust offshore Trans./oc. 1.5 5.7–6.1 2.65–2.76 ±0.5 ±0.1
Lower crust Cont./trans./oc. 8.0 6.4–7.0 2.90–3.05 Cont.: ±1.0 ±0.1
Trans./oc.: ±0.5
High velocity body Und./serp. 7.0 7.0–7.4 3.00–3.20 ±0.5 ±0.1
Upper mantle Mantle – 8.1 3.27 ±1.0 ±0.1

crust was modelled by a 1.5 to 3.0 km thick layer with upper boundary
velocities between 6.4 and 6.9 km/s and a lower boundary velocity of
7.0 km/s. Below this unit a HVB with a thickness of 3.5–7.0 km and
with velocities ranging from 7.0 to 7.4 km/s (Fig. 18, km 150–400) is ob-
served. Similar to profile 20070201, these high velocities are constrained
by refracted waves, which were subdivided in phase Pc2 (velocities up to
7.0 km/s) and Pc3 (velocities higher than 7.0 km/s).
In general, the total crustal thickness ranges from 6.5 to 8.5 km
between km 210 and 400. However, between km 320 and 350,
the oceanic crust thickens significantly to 9.5 km and thins again
eastwards of km 360. This variation of the Moho depth corresponds
to the Moho-step at km 320 in model 20070201.
The gravity model of profile 20070202 is shown in the lower part
of Fig. 18. The positive anomaly that can be seen landwards of both
profiles in the free-air gravity data of Sandwell and Smith (2009)
has a sharper southern flank in profile 20070202 than in 20070201.
The marine gravity measurements for profile 20070202 terminated
yet southwards of this flank.
Onshore, the upper crust was modelled with a density of 2.69 g/cm 3
and the middle continental crust with a density of 2.91 g/cm 3, as in

Fig. 16. Resolution of the P-wave models. The value of “one” indicates a very good resolution, “zero” means no ray coverage at all. The white bars represent the uncertainty of the
model layers depth nodes as listed in Tables 6a and 6b.
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 187

Fig. 17. P-wave and 2D-density model for profile 20070201. The magnetic anomaly field is shown in the upper panel with some identified chrons. Question marks are assigned to
the area, where no spreading anomalies could be identified. The panel below shows the P-wave model. A hatched spot marks the region, where indications for SDRS exist in the
seismic reflection data. A white dashed line marks the top of the crystalline crust. The observed (black) and modelled (red) free-air gravity is shown in the third panel together
with the magnetic data (hatched). The lowermost panel shows the 2D-density model. The densities are written in the blocks.

model 20070201. The density of the lower continental crust was introduced between km 70 and 150. We point out that the continen-
chosen to 3.05 g/cm 3. A sedimentary basin with a maximum thick- tal part of the model is very ambiguous (see remarks for model
ness of 1.5 km and a density of the sediments of 2.00 g/cm 3 was 20070201).
188 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 18. P-wave and 2D-density model for profile 20070202. The magnetic anomaly field is shown in the upper panel with some identified chrons. Question marks are assigned to
the area, where no spreading anomalies could be identified. The panel below shows the P-wave model. A hatched spot marks the region, where indications for SDRS exist in the
seismic reflection data. A white dashed line marks the top of the crystalline crust. The observed (black) and modelled (red) free-air gravity is shown in the third panel together
with the magnetic data (hatched). The lowermost panel shows the 2D-density model. The densities are written in the blocks.

Offshore, the sedimentary densities increase downwards from The upper crust was modelled with a density of 2.76 g/cm3 from km
1.95 g/cm3 near the seafloor to 2.64 g/cm3 above the oceanic basement. 175 to 318 and of 2.72 g/cm3 from km 355 to 400. In between, the upper
The resulting velocity–density distribution is in agreement with the layer of the locally thickened crust shows a density of 2.65 g/cm3, which
values for marine sediments given by Nafe and Drake (1957). is comparable in density with the lowermost sediment layer. Apart from
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 189

that crustal block, the offshore upper crustal densities are somewhat
higher in model 20070202 than in model 20070201. The lower crust
in model 20070202 was modelled with a density of 2.96 g/cm3 from
km 160–230 and with 2.90 g/cm3 from km 230 to 400.
As in model 20070201, the HVB was divided in several density
blocks and densities ranging from 3.00 up to 3.20 g/cm 3 were
modelled. These changes roughly reflect the internal velocity distri-
bution of the high-velocity-body (7.0–7.4 km/s). A summary of the
modelled velocities and densities is given in Table 6b.
To compare our modelling results with the coincident multichannel
seismic (MCS) profiles, the final raytracing models were converted into
two-way-traveltime and plotted in colour onto the time-migrated seis-
mic reflection sections of profiles 20070201 and 20070202 (Fig. 19).
The shape and layering of the lower sediments of the raytracing models
as well as the top of the crystalline crust are in good agreement with the
MCS data. Differences exist in the upper sedimentary column, where
prominent reflectors of the MCS data do not coincide with P-wave
model layer boundaries. Most likely, here, these P-wave model layer
boundaries mark stratigraphic boundaries rather than changes in seis-
mic impedance. A more detailed interpretation and discussion of the
MCS data is envisaged as part of a separate study.
6. Discussion
6.1. Nature of the crust
Our interpretation of the oceanic crustal layers is based on the clas-
sification given by White et al. (1992) with oceanic layer 1 consisting of

Fig. 19. a: P-wave model of profile 20070201, converted to TWT, overlain by MCS data. A black frame marks the region, where indications for SDRS exist in the seismic reflection
data. b: P-wave model of profile 20070202, converted to TWT, overlain by MCS data. A black frame marks the region, where indications for SDRS exist in the seismic reflection data.
190 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196
sediments and oceanic layer 2 consisting of extrusive basaltic lavas
and dykes. Oceanic layer 3 consists of gabbroic material with veloci-
ties and is usually more than twice as thick as layer 2. The global
mean thickness of normal oceanic crust is 7.1 ± 0.8 km (White et
al., 1992).
In order to compare the final velocity models with the typical
structure of oceanic crust, velocity–depth profiles were extracted
every 100 km along the offshore part of model 20070201 and every
75 km along the offshore part of model 20070202. They are plotted
in colour in Fig. 20 together with values of typical oceanic crust
(White et al., 1992).
Both models are characterised by a thin upper crustal layer (thick-
nesses around 1.5 km) with relatively high P-wave velocities around
5.6 km/s and a gradient, which is lower than that of normal oceanic
crust (Fig. 20). This might be explained by the age of the crust and
the large thickness of the sediments. The crustal thickness of model
20070201 fits into the normal range of oceanic crust for all velocity–
depth-profiles up to km 400. Further south, the crust has a thickness
of only 5.1 km, about 2 km thinner than normal oceanic crust. The
crustal thickness is generally higher along line 20070202 than along
line 20070201. The velocity–depth-functions located at km 210 and
285 of profile 20070202 show a total crustal thickness of 8.4 km,
about 1.3 km thicker than that of normal oceanic crust. Towards the
southeastern end of the model, at km 360, the crustal thickness de-
creases to 6.5 km (Fig. 20).
Our structural interpretation of the models is shown in Fig. 21. The
continental crust is thinned seawards over a distance of around
130 km by 50%, from 40 to around 20 km thickness. The identification
of seafloor spreading anomalies up to M33n leads us to interpret the
crust from km 275 to 550 along line 20070201 and from km 300 to
400 along line 20070202 as being oceanic in origin. The nature of the
crust between km 145 and 275 of profile 20070201 (Fig. 17) and be-
tween km 175 and 300 of profile 20070202 (Fig. 18) couldn't be un-
equivocally determined. Does the negative magnetic anomaly (Fig. 17:
km 140–200; Fig. 18: km 180–210) mark the initiation of oceanic
spreading or is the crust older than M33n of transitional nature and
true oceanic crust was not formed before M33n?
To constrain possible interpretations, the major structural features
of our models are briefly described:
– The velocity depth functions southeast of km 200 are similar to
those of standard oceanic crust (Fig. 20).
– Weak indications for Seaward Dipping Reflector Sequences (SDRS;
Hinz and Krause, 1982) are observed between km 160 and 190 in
profile 20070201 and between km 175 and 190 in profile 20070202
(Reichert et al., 2008). In profile 20070201, their location coincides
with the maximum thickness of the HVB at km 160.
– Between km 140 and 200 in profile 20070201 and between km 180
and 210 in profile 20070202, a pronounced negative magnetic anom-
aly is observed, which can be correlated with the position of the weak
SDRS and a region of significant crustal thinning. This strong neg-
ative anomaly might be caused not only by seafloor spreading
processes. A combination of different effects might add to the
remanent magnetization of oceanic crustal layer 2. SDRS, when
erupted during a reversed magnetic chron, can carry a strong
remanent negative magnetization. The same is true if highly
serpentinized peridotites are present. They might have enough
natural remanent magnetization and susceptibility to contribute
significantly to the magnetic anomaly field (Oufi et al., 2002).
The vicinity of a lower magnetized continental crustal block and
highly magnetized oceanic crust can produce a strong edge-
effect anomaly, and finally, intrusive complexes can also bear
strong magnetizations.
– The free-air gravity map after Sandwell and Smith (2009) shows
a positive free-air anomaly towards the coast, which has partially
been mapped during the MoBaMaSis cruise along line 20070201
(Fig. 17). The anomaly might be caused by intrusions into
stretched continental crust (see the brownish block in Fig. 21a).
The proximity of the both profiles makes the existence of similar
intrusions at the according position in profile 20070202 likely.
Yet, the marine measurements for profile 20070202 terminated
too far away from the coast to cover the positive gravity anoma-
ly, which is why the area of possible intrusions is marked in
brown and with a question mark in Fig. 21b.
– Close to the continent the crust is mainly composed of a high-
velocity-body, while at the end of the lines thin oceanic crust is
present. This HVB is defined by anomalous high P-wave veloci-
ties of more than 7.0 km/s. Maximum velocities of 7.5 km/s along
line 20070201 and 7.4 km/s along line 20070202 have locally been
found in the HVB. However, for the major part of the HVB the veloc-
ities range from 7.0 to 7.3 km/s.
Taking these observations into account, we favour an oceanic inter-
pretation of the crustal area in question (Fig. 17: km 145–275; Fig. 18:
km 160–300). The position of the (weak) SDRS (normally correspond-
ing to the onset of oceanic crust, see for example Ajay et al. (2010) or
Mjelde et al. (2007)) and the strong variations in the magnetic field
are indications for the continent–ocean-transition (COT) terminating
close to the coast. The similarity of the upper crustal velocities to
those found for oceanic layer 2 at the southeastern end of the profiles
and the agreement of the modelled velocity–depth-relations with
those of oceanic crust corroborate an oceanic origin.
This interpretation narrows the size of stretched continental or
transitional crust to 120–150 km (Figs. 17 and 18; km 50–200 in pro-
file 20070201, km 90–210 in profile 20070202). Then, the high ampli-
tude negative magnetic anomaly formed shortly before anomaly
M41n, when oceanic spreading started. A denser grid of magnetic
data would be essential to confirm this interpretation. In any case,
the oceanic crust extends farther to the coast than previously as-
sumed (see Fig. 9 for the position of the continent–ocean-boundary
after Raillard, 1990).
6.2. Origin of the high-velocity-body
Two classes of rifted continental margin have been established in lit-
erature: volcanic and nonvolcanic margins (Mutter et al., 1988; White
et al., 1992; Whitmarsh et al., 2001). The characteristics of volcanic pas-
sive margins are massive extrusive lava flows and strong seaward-
dipping reflector sequences (SDRS) as well as igneous intrusions and
crustal underplating in the area of the continent–ocean-transition
(COT). Thickened oceanic crust and the link to flood basalt provinces
are taken as indications for volcanic margins, too. Good examples
for this type of continental margins can be found in the North Atlantic
e.g. off Norway and East Greenland (Eldholm and Grue, 1994; Mjelde
et al., 1997; Voss and Jokat, 2007).
Nonvolcanic passive margins on the other hand show heavily
stretched continental crust with tilted blocks, zones of serpentinized,
sometimes exhumed mantle and thin transitional/early oceanic crust.
The margin of West Iberia can be regarded as a classic example of this
margin type (Whitmarsh et al., 1996). Lower crustal layers with
anomalous high P-wave-velocities ≥ 7 km/s have been found on
both types of margins (Chian et al., 1999; Voss and Jokat, 2007). De-
spite the abundance of these high-velocity-bodies, their lithology and
formation mechanisms are still not understood (Chian and Louden,
1994). Some authors interpret them as being created by magmatic
underplating; others suggest partial serpentinization of mantle perido-
tite being the reason for the high velocities.
Serpentinization of upper mantle peridotites by hydration of olivine
to serpentine is a low temperature metamorphic process (O'Hanley,
1996), generally assumed to be related to seawater penetrating into
the crust along faults and cracks (Mjelde et al., 2002). Probably, the
amount of water reaching crustal levels of at maximum 5 km is not
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196 191

Fig. 20. a–d: Velocity–depth profiles of the offshore part of the models (every 100 km for profile 20070201, every 75 km for profile 20070202), compared with crustal velocity of
the Atlantic Ocean (a and c) and of the Pacific Ocean (b and d) after White et al. (1992).

sufficient enough to lower the mantle velocities to around 7.2 km/s
(Minshull et al., 1998). The serpentinization process is gradual, and
may lead to a complete absence of Moho reflections as observed on
the West Iberia margin by Chian et al. (1999).
Magmatic underplating means the attachment of high density
mantle material to existing lower crust, often together with sill-like
intrusions. It is assumed, that this happens due to a convectional trans-
port of upper mantle material having an elevated potential temperature
above the “normal” ~1280 °C (Hawkesworth et al., 1999; Storey, 1995;
White and McKenzie, 1989; White et al., 1987) into temperature and
pressure regions, where partial melting occurs. The upper mantle con-
vection may be a consequence of spreading (passive convection, White
192 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 21. a: Interpretation of the crustal part of model 20070201 with the magnetic anomaly field in the upper panel. The dark spot marks the position of the SDRS indications in the
according seismic reflection profile. b: Interpretation of the crustal part of model 20070202 with the magnetic anomaly field in the upper panel. The dark spot marks the position of
the SDRS indications in the according seismic reflection profile.

and McKenzie, 1989) or may additionally be driven by temperature gra- ridge forming the crust) and the mantle potential temperature are the
dients in the upper mantle (active convection/upwelling, Mutter et al., driving factors defining the amount of melt production, and as a conse-
1988). The active upwelling ratio (defined as the relation between the quence the thickness and the velocity structure of the lower crust. As re-
upwelling rate of mantle material and the spreading half-rate of the sult of underplating, a significant impedance contrast is often observed
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196
at the border between the lower crust and the underplating body
causing reflections from the top of the HVB in seismic reflection pro-
files (Funck et al., 2007; Klingelhoefer et al., 2005; Voss and Jokat,
2007).
In our case, according to its seismic velocities, the HVB might
consist either of underplating, heavily intruded lower crust or of
serpentinized mantle. Intra-crustal reflections from the top of the
underplating body are not present in our data. Moreover, the smooth
velocity transition from the lower crust to the HVB argues against
magmatic underplating. The Moho is, however, well constrained by
reflections throughout both profiles (Fig. 12), which in turn makes
serpentinization of upper mantle peridotites unlikely.
The lack of reflections on top of the HVB might be explained by
crustal material that was heavily but gradually modified by mafic intru-
sions, leading to a smooth transition between the lower crust and the
HVB in the seismic resolution scale. These intrusions could have been
formed during magmatic underplating, but could also result from intru-
sions of mantle material before the onset of serpentinization.
Voss and Jokat (2007) gave an overview about dimensions of prom-
inent high-velocity-bodies found at different margins. The maximum
thickness of these bodies ranges from 9 km (Southeast Greenland,
Hopper et al., 2003) to 18 km (Namibia, Bauer et al., 2000; Gladczenko
et al., 1998). The maximum horizontal extent ranges from 90 km
(Hatton Bank, Morgan et al., 1989) to 225 km (East Greenland, Voss
and Jokat, 2007).
A comparison of the HVB of our profiles with those dimensions re-
veals, that its maximum width is rather big (350 and 250 km, proba-
bly extending over the southern end of profile 20070202, see Figs. 17
and 18), but its maximum thickness, however, is not that extreme (11
and 6 km, see Figs. 17 and 18).
A similar situation of a HVB occurring in combination with thin oce-
anic crust was found by Collier et al. (2009) at the northern Seychelles
continental margin. There, the HVB was interpreted as magmatic
underplating, which happened prior to breakup. Yet, in contrast to our
profiles, the velocities of the HVB are significantly higher (7.5–7.8 km/s)
(Collier et al., 2009).
Fig. 22. Crustal thickness against profile distance for ages older than M25n. Starting points of the curves are the southernmost edges of continental crust in both profiles. We chose
the position of M33n as reference point in both profiles and shifted profile 20070202 by 24 km for plotting purposes. The solid lines show the crustal thicknesses, the dotted lines
show the mean crustal P-wave velocities. Black lines are used for profile 20070201, red ones are used for profile 20070202. The green lines are the mean values between both
profiles.
193
To test the possibility of magmatic underplating, we calculated the
crustal thicknesses and mean seismic velocities along both profiles
and then used the relations of Korenaga et al. (2002) to draw conclu-
sions on the mantle potential temperature and the ratio of possible ac-
tive mantle upwelling, which would have been necessary to produce
the observed crustal architecture. The existence of a lithospheric lid
(Korenaga et al., 2002) was not assumed for our calculations, because
we deal with young transitional or oceanic crust at the time of the
HVB's formation.
Figs. 22–24 show the results of these calculations. Comparing the
differences in thickness and velocity between both profiles and the
error bars (Fig. 23), we decided to average the values, keeping in
mind the small lateral distance of 110 km between both profiles.
The crustal thickness decreases over a distance of around 50 km
from 14.3 to 7.7 km, then slightly increases again to 8.0 km and de-
creases finally to 6.2 km at M25n. The mean crustal velocity rises
first from 7.03 km/s to 7.10 km/s. Apart from this first increase, the
velocity and the thickness curves correlate in general. The velocity
decreases to 6.98 km/s at M25n, having in between a small maximum
after M33n, which coincides with a maximum in the crustal thickness
of profile 20070202 (Fig. 12).
Fig. 23 shows the crustal velocity plotted against thickness with
marked ages and error bars. The curved lines in the back represent
the relations of Korenaga et al. (2002), calculated for constant mantle
potential temperatures as well as for constant upwelling ratios (χ). In
Fig. 24 the coordinate system is changed to potential temperature
against upwelling ratio with crustal thickness and mean velocity
shown as curved lines. Again referring to the averaged curve, a man-
tle dynamic and temperature development as follows can explain the
lower crustal fabric of our profiles: The crustal region near the shore-
line and older than M33n was formed over a highly active convective
mantle with potential temperatures around 30 °C higher than the
1280 °C usually taken as “normal”. The mantle was strongly upwell-
ing at these times (upwelling ratio around 6). The intrusions in the
stretched continental crust (Fig. 21) might be connected to a pulse
of initial heating, which heavily modified the already weakened
194 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196

Fig. 23. Mean crustal P-wave velocity against crustal thickness. Diamonds represent profile 20070201, squares profile 20070202. Circles show the fitted mean values (Fig. 22) of
both profiles. Contour lines are drawn for constant mantle potential temperatures and for constant upwelling ratios.

crust and facilitated the emplacement of weak SDRS. After a time of
slightly increasing mantle potential temperature up to around 1330 °C
and mantle convection slowing down to χ = 2, the mantle thermal
anomaly seems to have been exhausted and the trend changed. After-
wards, the mantle started to cool down rapidly till it reached “normal”
potential temperatures around 1280 °C at M25n. In parallel, the mantle
convection slightly increased again to χ = 2.5 (Fig. 24). For the times
younger than M25n, we have only information from one profile, in
which the HVB pinches out at M24n.
This temperature and convection regime of the mantle is in agree-
ment with other margins (e.g. in the North Atlantic: Holbrook et al.,
2001). A margin formation, more controlled by dynamic rift processes
rather than by large-scale thermal mantle anomalies, was predicted
for the Atlantic margin of North America by Holbrook and Kelemen
(1993), too, considering the rapid decrease of crustal thickness
and lower crustal velocities perpendicular to that margin. Armitage
et al. (2010) indicated that another important factor controlling the
amount of generated melt, is the specific rift history of a margin, partic-
ularly the relation in time between crustal thinning and the emplace-
ment of a mantle thermal anomaly. More detailed knowledge about
the rift history in our research area would be needed to take this aspect
into account and to come to a reliable model of the mantle–lithosphere
interactions at the conjugate margins in the Africa–Antarctica-Corridor
during the Gondwana breakup.
As the rifting between West- and East-Gondwana was accompa-
nied by massive volcanism at the conjugate margins of Antarctica
and Africa and rift related magmatism also took place landward of
our profiles (Fig. 1), we propose that the Central Mozambique conti-
nental margin belongs to the class of volcanic margins.
7. Conclusions
Two wide-angle seismic models were established along parallel
profiles running from the Mozambique coast into the Mozambique

Fig. 24. Mantle potential temperature against active upwelling ratio χ. Diamonds represent profile 20070201, squares profile 20070202. Circles show the fitted mean values (Fig. 22
of both profiles. Contour lines are drawn for constant crustal thicknesses and for constant seismic velocities.
 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196
Channel between longitude 37°30′ E and 41° E. Both models show up
to 40 km thick continental crust at the Mozambique continental mar-
gin, which thins to 20 km over a distance of around 130 km. A lower
crustal HVB (7.0–7.4/7.5 km/s) is observed along both profiles and
can be explained by a combination of active upper mantle convection
and increased upper mantle potential temperatures. Profile 20070201
covers the Mesozoic anomaly sequence up to M11An.1n as youngest
identification. In the shorter profile 20070202, M22n.1n was identi-
fied as youngest anomaly. The magnetic anomalies were correlated
between both profiles and are in agreement with existing identifica-
tions in the Mozambique Basin/Mozambique Channel. From the
shoreline towards chron M33n, a 130 to 155 km wide crustal domain
was found, where magnetic spreading anomalies could not be identi-
fied anymore. Nevertheless, our data support the existence of oceanic
crust in this area. Extrapolating the modelled spreading velocities to-
wards the coast we postulate chron M41n (166 Ma) being the oldest
magnetic spreading anomaly near the Central Mozambique continen-
tal margin. This means that the continent–ocean-transition in the
Mozambique Channel is located closer to the coast than previously
assumed.
Acknowledgements
We thank the BGR for funding the scientific projects and the
French polar institute (IPEV) for the logistic and financial support.
The MoBaMaSis land acquisition was possible thanks to the support
provided by the Direcção Nacional de Geologia de Moçambique
(Eng. Elias Daudi, National Director), the Geology Department of the
Eduardo Mondlane University (Maputo), and also to the financial
contributions from IFREMER, GRICES, IDL, AWI and CGE. Finally, we
thank the captain and crew of R/V Marion Dufresne II (IPEV) for
their excellent work during the expedition.
References
Afonso, R.S., 1976. A geologia de Moçambique. Noticia explicativa da carta de Moçambique,
Direcçao dos serviços de geologia e minas.Impressa national de Moçambique,
Maputo.
Ajay, K.K., Chaubey, A.K., Krishna, K.S., Gopala Rao, D., Sar, D., 2010. Seaward dipping
reflectors along the SW continental margin of India: evidence for volcanic passive
margin. Journal of Earth System Science 119 (6), 803–813.
Armitage, J.J., Collier, J.S., Minshull, T.A., 2010. The importance of rift history for volca-
nic margin formation. Nature 465, 913–917.
Barton, A.J., White, R.S., 1997. Crustal structure of Edoras Bank continental margin and
mantle thermal anomalies beneath the North Atlantic. Journal of Geophysical
Research 102 (B2), 3109–3129.
Bauer, K., Neben, S., Schreckenberger, B., Emmermann, R., Hinz, K., Fechner, N., Gohl, K.,
Schulze, A., Trumbull, R.B., Weber, K., 2000. Deep structure of the Namibia conti-
nental margin as derived from integrated geophysical studies. Journal of Geophys-
ical Research 105 (B11), 25829–25853.
Beck, R.H., Lehner, P., 1974. Oceans, new frontier in exploration. American Association
of Petroleum Geologists Bulletin 58 (3), 376–395.
Bergh, H.W., 1977. Mesozoic seafloor off Dronning Maud Land, Antarctica. Nature 269,
686–687.
Bernard, A., 2005. Refined spreading history at the Southwest Indian Ridge for the last
96 Ma, with the aid of satellite gravity data. Geophysical Journal International 162,
765–778.
Bunce, E., Molnar, P., 1977. Seismic reflection profiling and basement topography in the
Somali basin: possible fracture zones between Madagascar and Africa. Journal of
Geophysical Research 82 (33), 5305–5311.
CGS, 2000. Regional Aeromagnetic data, 1:1,000,000 Total Magnetic Intensity Map of
the Republic of South Africa. Compiled by: Stettler, E.H., Fourie, C.J.S., Cole, P. Council
for Geoscience (CGS), Pretoria, South Africa.
Chian, D., Louden, K.E., 1994. The continent–ocean crustal transition across the south-
west Greenland margin. Journal of Geophysical Research 99 (B5), 9117–9135.
Chian, D., Louden, K.E., Minshull, T.A., Whitmarsh, R.B., 1999. Deep structure of the
ocean–continent transition in the southern Iberia Abyssal Plain from seismic re-
fraction profiles: Ocean Drilling Program (Legs 149 and 173) transect. Journal of
Geophysical Research 104 (B4), 7443–7462.
Christensen, N.I., Mooney, W.D., 1995. Seismic velocity structure and composition of
the continental crust: a global view. Journal of Geophysical Research 100 (B7),
9761–9788.
Cohen, J.K., Stockwell, J.W., 2003. Seismic Unix Release 37: A Free Package for Seismic
Research and Processing. Center for Wave Phenomena, Colorado School of Mines.
195
Collier, J.S., Minshull, T.A., Hammond, J.O.S., Whitmarsh, R.B., Kendall, J.-M., Sansom, V.,
Lane, C.I., Rumpker, G., 2009. Factors influencing magmatism during continental
breakup: new insights from a wide-angle seismic experiment across the conjugate
Seychelles–Indian margins. Journal of Geophysical Research 114, 1–25.
Coster, P.W., Lawrence, S.R., Fortes, G., 1989. Mozambique: a new geological frame-
work for hydrocarbons exploration. Journal of Petroleum Geology 12 (2), 205–230.
Cox, K.G., 1992. Karoo igneous activity, and the early stages of the break-up of Gondwa-
naland. In: Storey, B.C., Alabaster, T., Pankhurst, R.J. (Eds.), Magmatism and the
Causes of Continental Break-up: Geol. Soc. Spec. Publ., 68, pp. 137–148 (London).
Dalziel, I.W.D., 1992. Antarctica; a tale of two supercontinents? Annual Review of Earth
and Planetary Sciences 20, 501–526.
De Buyl, M., Florès, G., 1986. The southern Mozambique Basin: the most promising
hydrocarbon province offshore East Africa. In: Halbouty, M.T. (Ed.), Future Petroleum
Provinces of the World: AAPG Mem., 40, pp. 399–425.
Eagles, G., König, M., 2008. A model of plate kinematics in Gondwana breakup.
Geophysical Journal International 173, 703–717.
Eldholm, O., Grue, K., 1994. North Atlantic volcanic margins: dimensions and produc-
tion rates. Journal of Geophysical Research 99 (B2), 2955–2968.
Encarnacion, J., Fleming, T.H., Elliot, D.H., Eales, H.V., 1996. Synchronous emplacement of
Ferrar and Karoo dolerites and the early breakup of Gondwana. Geology 24, 535–538.
Förster, R., 1975. The geological history of the sedimentary basin of Southern Mozam-
bique, and some aspects of the origin of the Mozambique Channel. Palaeogeography,
Palaeoclimatology, Palaeoecology 17, 267–287.
Fortes, G., Kihle, R., 1983. Offshore Mozambique, recent surveys outline new potential.
Noroil 27–32.
Funck, T., Jackson, H.R., Louden, K.E., Klingelhoefer, F., 2007. Seismic study of the transform-
rifted margin in Davis Strait between Baffin Island (Canada) and Greenland: what hap-
pens when a plume meets a transform. Journal of Geophysical Research 112, 1–22.
General Bathymetric Chart of the Oceans 08 (GEBCO_08), 2003. International Hydro-
graphic Organization (IHO) and the Intergovernmental Oceanographic Commission
(IOC) of UNESCO. http://www.gebco.net/.
Ghidella, M.E., Lawver, L.A., Gahagan, L.M., 2007. Break-up of Gondwana and opening of
the South Atlantic: review of existing plate tectonic models. U.S. Geological Survey
and The National Academies; USGS OF-2007-1047, Short Research Paper 055, pp. 1–5.
Gladczenko, T.P., Skogseid, J., Eldholm, O., 1998. Namibia volcanic margin. Marine
Geophysical Researches 20, 313–341.
Gradstein, F.M., Ogg, J.G., Smith, A.G. (Eds.), 2004. A Geologic Time Scale 2004. Cambridge
University Press, Cambridge (589 pp.).
Grunow, A.M., Kent, D.V., Dalziel, I.W.D., 1991. New palaeomagnetic data from Thurston
Island: implications for the tectonics of West Antarctica and Weddell Sea opening.
Journal of Geophysical Research 96 (B11), 17,935–17,954.
Hansen, S.E., Nyblade, A.A., Julià, J., 2009. Estimates of crustal and lithospheric thickness
in sub-saharan Africa from S-wave receiver functions. South African Journal of
Geology 112, 229–240.
Hawkesworth, C., Kelley, S., Turner, S., Le Roex, A., Storey, B., 1999. Mantle processes
during Gondwana break-up and dispersal. Journal of African Earth Sciences 28,
239–261.
Heirtzler, J.R., Burroughs, R.H., 1971. Madagascar's paleoposition: new data from the
Mozambique Channel. Science 174, 488–490.
Hinz, K., Krause, W., 1982. The continental margin of Queen Maud Land/Antarctica:
seismic sequences, structural elements and geological development. Geologisches
Jahrbuch E23, 17–41.
Holbrook, W.S., Kelemen, P.B., 1993. Large igneous province on the US Atlantic margin and
implications for magmatism during continental breakup. Nature 364, 433–436.
Holbrook, W.S., Larsen, H.C., Korenaga, J., Dahl-Jensen, T., Reid, I.D., Kelemen, P.B.,
Hopper, J.R., Kent, G.M., Lizarralde, D., Bernstein, S., Detrick, R.S., 2001. Mantle ther-
mal structure and active upwelling during continental breakup in the North Atlantic.
Earth and Planetary Science Letters 190, 251–266.
Hopper, J.R., Dahl-Jensen, T., Holbrook, W.S., Larsen, H.C., Lizarralde, D., Korenaga, J.,
Kent, G.M., Kelemen, P.B., 2003. Structure of the SE Greenland margin from seismic
reflection and refraction data: implications for nascent spreading center subsi-
dence and asymmetric crustal accretion during North Atlantic opening. Journal of
Geophysical Research 108 (B5), 2269.
Jamal, D.L., 2003. Geology and Geochronology of northern Mozambique, with implications
for Gondwana reconstructions. (Ph.D. thesis, unpublished) Univ. of Capetown.
Jokat, W., 2006. Southeastern Atlantic and southwestern Indian Ocean: reconstruction
of the sedimentary and tectonic development since the Cretaceous, AISTEK-II:
Mozambique Ridge and Mozambique Basin, report of the RV “Sonne” cruise SO-183,
project AISTEK-II, 20 May to 7 July 2005. Berichte zur Polarforschung 521, 1–71.
Jokat, W., Boebel, T., König, M., Meyer, U., 2003. Timing and geometry of the early
Gondwana breakup. Journal of Geophysical Research 108 (B9), 1–15.
Jourdan, F., Féraud, G., Bertrand, H., Kampunzu, A.B., Tshoso, G., Watkeys, M.K., Le Gall,
B., 2005. Karoo large igneous province: brevity, origin, and relation to mass extinc-
tion questioned by new 40Ar/39Ar age data. Geology 33, 745–748.
Klingelhoefer, F., Edwards, R.A., Hobbs, R.W., England, 2005. Crustal structure of the NE
Rockall Trough from wide-angle seismic data modelling. Journal of Geophysical
Research 110, 1–25.
König, M., Jokat, W., 2010. Advanced insights into magmatism and volcanism of the
Mozambique Ridge and Mozambique Basin in the view of new potential field
data. Geophysical Journal International 180, 158–180.
Korenaga, J., Kelemen, P.B., Holbrook, W.S., 2002. Methods for resolving the origin of
large igneous provinces from crustal seismology. Journal of Geophysical Research
107 (B9), 1–27.
Kröner, A., 1977. Precambrian mobile belts of southern and eastern Africa: ancient su-
tures or sites of eusialic mobility? A case for crustal evolution towards plate tecton-
ics. Tectonophysics 40, 101–135.
196 V.T. Leinweber et al. / Tectonophysics 599 (2013) 170–196
Lafourcade, 1984. Etude géologique et géophysique de la marge continentale au Sud de
Mozambique (17°S à 28°S). (Thèse 3ème cycle) Université P. et M. Curie, Paris.
Lawver, L.A., Royer, J.Y., Sandwell, D.T., Scotese, C.R., 1991. Evolution of the Antarctic con-
tinental margins. In: Thomsons, M.R.A., Crame, J.A., Thompson, J.W. (Eds.), Geological
Evolution of Antarctica. Cambridge University Press, Cambridge, pp. 533–540.
Leinweber, V., Jokat, W., 2012. The Jurassic history of the Africa–Antarctica Corridor —
new constraints from magnetic data on the conjugate continental margins.
Tectonophysics 530–531, 87–101. http://dx.doi.org/10.1016/j.tecto.2011.11.008.
Lort, J.M., Limond, W.Q., Segoufin, J., Patriat, P., Delteil, J.R., Damotte, B., 1979. New seis-
mic data in the Mozambique Channel. Marine Geophysical Researches 4, 71–89.
Ludwig, W.J., Nafe, J.E., Drake, C.L., 1970. Seismic refraction. In: Maxwell, A.E. (Ed.), The
Sea, 4. Wiley-Interscience, pp. 53–84.
Martin, A.K., Hartnady, C.J.H., 1986. Plate tectonic development of the South West
Indian Ocean: a revised reconstruction of East Antarctica and Africa. Journal of
Geophysical Research 91 (B5), 4767–4786.
McElhinny, M.W., Larson, R.L., 2003. Jurassic dipole low defined from land and sea. EOS,
Transactions American Geophysical Union 84, 362–366.
Mendel, V., Munschy, M., Sauter, D., 2004. MODMAG user's manual. Computers and
Geosciences 1–12.
Minshull, T.A., Muller, M.R., Robinson, C.J., White, R.S., Bickle, M.J., 1998. Is the oceanic
Moho a serpentinization front? In: Mills, R.A., Harrison, K. (Eds.), Modern Ocean
Floor Processes and the Geological Record, Geological Society of London, London,
United Kingdom, Geol. Soc. Spec. Publ., 148, pp. 71–80.
Mjelde, R., Kodaira, S., Digranes, P., Shimamura, H., Kananzawa, T., Shiobara, H., Berg,
E.W., Riise, O., 1997. Comparison between a regional and a semi-regional crustal
OBS model in the Vøring Basin, Mid-Norway Margin. Pure and Applied Geophysics
149, 641–665.
Mjelde, R., Kasahara, J., Shimamura, H., Kamimura, A., Kanazawa, T., Kodaira, S., Raum, T.,
Shiobara, H., 2002. Lower crustal seismic velocity-anomalies; magmatic underplating
or serpentinized peridotite? Evidence from the Voering Margin, NE Atlantic. Marine
Geophysical Researches 23, 169–183.
Mjelde, R., Raum, T., Murai, Y., Takanami, T., 2007. Continent–ocean-transitions: review,
and a new tectono-magmatic model of the Vøring Plateau, NE Atlantic. Journal of
Geodynamics 43, 374–392.
Morgan, J.V., Barton, P.J., White, R.S., 1989. The Hatton Bank continental margin—III. Struc-
ture from wide-angle OBS and multichannel seismic refraction profiles. Geophysical
Journal International 98, 367–384.
Mougenot, D., Virlogeux, P., Vanney, J.R., Malod, J.A., 1986. La marge continentale au
Nord du Mozambique: résultats préliminaires de la campagne MD40/MACAMO.
Bulletin de la Societe Geologique de France 2 (3), 419–422.
Mutter, J.C., Buck, W.R., Zehnder, C.M., 1988. Convective partial melting: 1. A model for
the formation of thick basaltic sequences during the initiation of spreading. Journal
of Geophysical Research 93 (B2), 1031–1048.
Nafe, J.E., Drake, C.L., 1957. Variations with depth in shallow and deep water marine sed-
iments of porosity, density and the velocity of compressional and shear waves. Geo-
physics 22 (3), 523–552.
Nguuri, T., Gore, J., James, D., Webb, S., Wright, C., Zengeni, T., Gwavava, O., Snoke, J.,
Kaapvaal Seismic Group, 2001. Crustal structure beneath southern Africa and its
implications for the formation and evolution of the Kaapvaal and Zimbabwe cra-
tons. Geophysical Research Letters 28, 2501–2504.
Norton, I.O., Sclater, J.G., 1979. A model for the evolution of the Indian Ocean and the
breakup of Gondwanaland. Journal of Geophysical Research 84 (B12), 6803–6830.
O'Hanley, D.S., 1996. Serpentinites: Records of Tectonic and Petrological History.Oxford
University Press, New York.
Oufi, O., Cannat, M., Horen, H., 2002. Magnetic properties of variably serpentinized
abyssal peridotites. Journal of Geophysical Research 107 (B5), 3-1–3-19.
Rabinowitz, P.D., Coffin, M.F., Falvey, D., 1983. The separation of Madagascar and Africa.
Science 220, 67–69.
Raillard, S., 1990. Les marges de l'Afrique de l'Est et les zones de fracture associées:
Chaine Davie et Ride du Mozambique, Ph.D. thesis, Laboratoire de Géodynamique
sous-marine Villefranche-sur-mer et Département de Géologie Océanique Paris,
Université Pierre et Marie Curie.
Reichert, C., Neben, S., Adam, J., Aslanian, D., Roest, W., Mahanjane, S., James, E.S., Bargeloh,
H.-O., Behrens, T., Block, M., Ehrhardt, A., Heyde, I., Jokat, W., Kallaus, G., Klingelhoefer,
F., Moulin, M., Kewitsch, P., Schrader, U., Schreckenberger, B., Sievers, J., 2008. Cruise
Report — BGR Cruise BGR07 — R/V Marion Dufresne Cruises MD163 & MD164 —
Project: MoBaMaSis 2008.Bundesanstalt für Geowissenschaften und Rohstoffe,
Hannover.
Roeser, H.A., Fritsch, J., Hinz, K., 1996. The development of the crust off Dronning Maud
Land, East Antarctica. In: Storey, B.C., King, E.C., Livermore, R.A. (Eds.), Weddell Sea
Tectonics and Gondwana Break-up, Geological Society of London, London, United
Kingdom, Geol. Soc. Spec. Publ., 108, pp. 243–264.
Sahu, Bijay K., 2001. Aeromagnetics of selected continental areas flanking the Indian
Ocean; with implications for geological correlation and reassembly of central
Gondwana. (PhD thesis) University of Capetown, South Africa.
Sandwell, D.T., Smith, W.H.F., 2009. Global marine gravity from retracked Geosat and
ERS-1 altimetry: ridge segmentation versus spreading rate. Journal of Geophysical
Research 114, 1–18.
Scrutton, R.A., Heptonstall, W.B., Peacock, J.H., 1981. Constraints on the motion of
Madagascar with respect to Africa. Marine Geology 43, 1–20.
Segoufin, J., 1978. Anomalies magnétiques mésozoïque dans le bassin de Mozambique.
Comptes Rendus de l'Académie des Sciences Paris, Série D t. 287, 109–112.
Segoufin, J., Patriat, P., 1981. Reconstruction de l'Océan Indien occidental pour les
époques des anomalies M21, M2 et 34. Paléoposition de Madagascar. Bulletin de
la Société Géologique de France 23, 603–607.
Simpson, E.S.W., Sclater, J.G., Parsons, B., Nortion, I.O., Meinke, L., 1979. Mesozoic mag-
netic lineations in the Mozambique basin. Earth and Planetary Science Letters 43,
260–264.
Stockwell, J.W., 1999. The CWP/SU: Seismic Unix package. Computers and Geosciences
25 (4), 415419.
Storey, B.C., 1995. The role of mantle plumes in continental breakup: case histories
from Gondwanaland. Nature 377, 302–308.
Tectonic Map of Mozambique, 2001, Republic of Mozambique, Ministry of Mineral
Resources and Energy, National Directorate of Geology, Scale 1:2,000,000.
Tivey, M.A., Sager, W.W., Lee, S.-M., Tominaga, M., 2006. Origin of the Pacific Jurassic
quiet zone. Geology 34, 789–792.
Virlogeux, P., 1987. Géologie de la marge nord-mozambique et de la chaîne Davie (9°S
à 21°S). (Campagne MD40-MACAMO, Thèse d'Université, dactyl) Univ. P. et M.
Curie, Paris.
Voss, M., Jokat, W., 2007. Continent–ocean transition and volominous magmatic
underplating derived from P-wave velocity modelling of the East Greenland conti-
nental margin. Geophysical Journal International 170, 580–604.
White, R., McKenzie, D., 1989. Magmatism at rift zones: the generation of volcanic conti-
nental margins and flood basalts. Journal of Geophysical Research 94 (B6), 7685–7729.
White, R.S., Spence, G.D., Fowler, S.R., McKenzie, D.P., Westbrook, G.H., Bowen, A.N.,
1987. Magmatism at rifted continental margins. Nature 330, 439–444.
White, R.S., McKenzie, D., O'Nions, R.K., 1992. Oceanic crustal thickness from seismic
measurements and rare earth element inversions. Journal of Geophysical Research
97 (B13), 19,683–19,715.
Whitmarsh, R.B., White, R.S., Horsefield, S.J., Sibuet, J.C., Recq, M., Louvel, V., 1996. The
ocean–continent boundary off the western continental margin of Iberia; crustal
structure of Galicia Bank. Journal of Geophysical Research 101, 28291–28314.
Whitmarsh, R.B., Manatschal, G., Minshull, T.A., 2001. Evolution of magma-poor conti-
nental margins from rifting to seafloor spreading. Nature 413, 150–154.
Zelt, C.A., Smith, R.B., 1992. Seismic traveltime inversion for 2-D crustal velocity struc-
ture. Geophysical Journal International 108, 16–34.