Surv Geophys

https://doi.org/10.1007/s10712-018-9463-5

Geostatistical Interplay Between Geophysical

and Geochemical Data: Mapping Litho‑Structural
Assemblages of Mesozoic Igneous Activities
in the Parnaíba Basin (NE Brazil)

David L. de Castro1 · Diógenes C. Oliveira1 · Maria Helena B. M. Hollanda2

Received: 1 June 2017 / Accepted: 30 January 2018

© Springer Science+Business Media B.V., part of Springer Nature 2018

Abstract Two widespread magmatic events are recorded in the Parnaíba basin (NE Bra-
zil) during the Jurassic/Cretaceous opening of the Central and South Atlantic Oceans. The
Early Jurassic (~ 200 Ma) lava flows of the Mosquito Formation occur essentially in the
western and southern basin segments, representing one of the largest expressions of the
Central Atlantic Magmatic Province in the South American Plate. In contrast, sill com-
plexes and dike swarms of the Early Cretaceous (129–124 Ma) Sardinha Formation occur
in the eastern part of the basin and are chrono-correlated to the large Paraná–Etendeka
igneous province and to the Rio Ceará–Mirim Dike Swarm. We gathered geophysical,
well logging, outcrop analogs and geochemical data to recognize geometrical shapes and
areal distribution patterns of igneous-related constructions. Seismic and well data reveal
hundreds of km wide multilayered sill complexes and dikes, which are widespread across
vast regions of the basin without evident structural control from either the Precambrian
basement grain or the basin internal architecture. Anomaly enhancement techniques and
self-organizing maps (SOM) procedure were applied on airborne magnetic data, unraveling
near-surface magmatic features in four distinct magnetic domains. Using SOM analysis,
the basaltic rocks were divided into six groups based on magnetic susceptibility and major
elements composition. These results suggest common origin for both magmatic episodes,
probably a combination of effects of edge-driven convection and large-scale mantle warm-
ing under the westward moving West Gondwana during the Central and South Atlantic
opening, which caused a shifted emplacement to the east of the igneous rocks in the Par-
naíba basin.

Keywords Airborne magnetic · Reflection seismic · Major elements · Magmatic events ·

Parnaíba basin · NE Brazil

* David L. de Castro
[email protected]
1
Programa de Pós‑graduação em Geodinâmica e Geofísica, Universidade Federal do Rio Grande
do Norte, Campus Universitário, Natal 59078‑970, Brazil
2
Departamento de Mineralogia e Geotectônica, Instituto de Geociências, Universidade de São
Paulo, Rua do Lago, 562, São Paulo 05508‑080, Brazil

13
 Surv Geophys

1 Introduction

One of the most fascinating events in plate tectonics is the widespread continental breakup
and the creation of new basaltic oceanic crust. The rupture of the continental lithosphere is
often preceded by rifting and large amounts of magmatism, which intrude and extrude over
vast areas of the crust concentrating within old or newly formed sedimentary basins (e.g.,
Turcotte and Schubert 1982; Condie 1997). The identification of magmatic processes and
deposits and the evaluation of their impact on the continental rupture history are, thus, two
important aspects of driving mechanism of plate motions. A spectacular example of this
global-scale tectonic mechanism producing extraordinary volumes of igneous rocks is the
onset of the Atlantic Ocean when the Pangea supercontinent split in the Mesozoic.
The Pangea supercontinent started to break apart in the Late Triassic driven by global
diverging forces, which culminated in the opening of the Proto-Caribbean/Central Atlan-
tic/Alpine-Tethys oceanic system at the Triassic–Jurassic boundary (Stampfli and Hochard
2009; Stampfli et al. 2013). At that time, a 7500-km-wide continental flood basalt (CFB),
the Central Atlantic Magmatic Province (CAMP), took place in parts of North America,
Europe, Africa and South America, leading to the initial opening of the Central Atlantic
Ocean (Bertrand et al. 1982; Marzoli et al. 1999; Nomade et al. 2007; Merle et al. 2011). In
South America, CAMP is characterized by extensive tholeiitic flows, sills and dike swarms
emplaced into Archean–Early Proterozoic cratonic areas, Late Proterozoic to Paleozoic
Amazonas and Parnaíba basins, and a late Permian–Jurassic rift system in the sub-Andean
region of Bolivia (e.g., Marzoli et al. 1999; De Min et al. 2003; Merle et al. 2011; Bertrand
et al. 2014). The main period of CAMP magmatism ranged from 202 to 190 Ma with a
peak volume around 199 Ma (Nomade et al. 2007).
The continued destruction of Pangea was followed by another large magmatic event,
accompanying the South Atlantic opening (Cordani and Vandoros 1967; Peate 1997). It
was represented by extraordinarily large volcanic (flood basalts) accumulations and dike
swarms forming large igneous provinces (LIPs). The most representative LIP is the Par-
aná–Etendeka igneous province (PEIP), covering a surface area of ~ 1.2 × 106 km2 over the
southern South America and Africa (Namibia, Angola), with a preserved volume in excess
of one million cubic kilometers in the Paleozoic cratonic Paraná Basin and offshore basins
along the Brazilian and African coasts (e.g., Cordani and Vandoros 1967; Fodor and Vetter
1984; Chang et al. 1992; Peate 1997). The widespread Paraná–Etendeka magmatic events
seem to have also affected the Neoproterozoic Borborema Province and Parnaíba basin
2000 km to the north (Cordani 1970; Bellieni et al. 1990). However, considerable debate is
ongoing about the real nature of the northernmost magmatism and its possible relationship
with the Paraná–Etendeka LIP is not yet well established.
In the South American Platform, the emplacement of the large igneous provinces (LIPs)
is concentrated mainly into the widespread Paleozoic cratonic basins (Amazonas, Paraná,
Parecis and Parnaíba) (e.g., Bellieni et al. 1990; De Min et al. 2003). Besides the large
volume of intruded and extruded magma, cratonic basins are regions of long-lived subsid-
ence history with major regional and sometimes global geodynamic significance (Saunders
et al. 2007; Allen and Armitage 2012). Lack of major normal faulted boundaries implies
that the long-lived subsidence periods argued for those basins should have been preferen-
tially driven by lithosphere/asthenosphere dynamics rather than by regional-scale exten-
sional tectonics. Association with continental stretching followed by flexural compensation
of the thermal contraction load plays an important role in many tectonic models (Allen and
Armitage 2012). New thermal models ascribe crustal flexure in cratonic basins to push-up

13
Surv Geophys

of their margins by lateral temperature variations in the sub-asthenospheric mantle (Mor-

gan 2015). Thus, the understanding of the spatial distribution of the magmatic structures
in them is a way to shed light on the tectonic processes of LIP formation in a continental
breakup setting.
The Parnaíba basin presents an interesting geodynamic scenario since it concentrates
in NE Brazil (Fig. 1) extensive Lower-Middle Jurassic tholeiitic flood basalts and Lower
Cretaceous tholeiitic sills and dikes associated with the two-stage opening of the Atlantic
Ocean, of which CAMP and PEIP are the most representative magmatic provinces. This
interesting geological setting has meant that this large Paleozoic basin has become the
target of numerous petrological, geochronological and geotectonic studies. Furthermore,
recent production of natural gas boosted prominent exploratory efforts from Brazilian Oil
Agency (ANP) and many international oil companies to investigate different aspects of the
basin formation and evolution. Despite all previous interpretation and new geological and
geophysical data, the actual extent and origin of the Mesozoic magmatic episodes within

Fig. 1  Simplified geological map of the Parnaíba basin, NE Brazil. 1—Rift zones (WRZ—Western Rift
Zone; TRZ—Transbrasiliano Rift Zone); sedimentary sequences (2—Silurian; 3—Devonian–Carbon-
iferous; 4—Permian–Triassic; 5—Jurassic; 6—Cretaceous); 7—basaltic rocks (Mosquito (dark green)
and Sardinha (dark blue) Formations); 8—Neocretaceous San Franciscana Basin; 9—Cenozoic to recent
sedimentary cover. Brasiliano structures: AR—Araguaia Suture Zone; SP—Senador Pompeu Shear Zone;
TB—Transbrasiliano Lineament

13
 Surv Geophys

the Parnaíba basin and their relationship with the two large magmatic provinces and the
role in the multiphasic opening of Equatorial and South Atlantic Ocean still remain poorly
understood.
Mapping magmatic occurrence is a difficult task since the smallest part of its volume
actually crops out, which often provides restricted evidence of the real extent and complex-
ity of the magmatism and prevents a greater understanding of the regional tectonic and
thermal histories. However, contrasts of their physical properties (e.g., density, magnetic
susceptibility, seismic velocity, radioelement content) allow applying various geophysical
methods to map their behavior in different depth levels and investigation scales. In sedi-
mentary basins, short-wavelength and high-amplitude gravity and magnetic anomalies are
generally related to basaltic structures, whose high density and magnetic content contrast
against less dense and non-magnetic sedimentary country rocks (e.g., Grant 1985; Blaikie
et al. 2014; Drenth et al. 2014). In the Parnaíba basin, dikes, sill complexes and lava flow
fields tend to have a high magnetic signature since they are essentially composed of mafic
(diabase and basalt) rocks. Filtering and anomaly enhancement techniques can highlight
this magnetic pattern and facilitate the mapping of the magmatic structures. Moreover,
introducing a sophisticated automatic procedure to quantitatively measure geophysical fea-
tures or patterns of interest, interpreters can reduce the subjectivity of the magnetic anom-
alies interpretation. Given this, we applied an unsupervised mapping method to identify
zones of magmatic rock occurrence in the Parnaíba basin based on stochastic calculations.
The reflection seismic method is by far the most important geophysical technique in
sedimentary basin exploration due to its high accuracy, high resolution and great penetra-
tion (Telford et al. 1998). Seismic data are especially important to provide insights into
intrusion architecture, emplacement mechanisms and magma flows in sedimentary basins
(e.g., Smallwood and Maresh 2002; Planke et al. 2005; Schofield et al. 2017). Magmatic
structures interbedded in basin infill are a singular target in seismic exploration since the
contrast of acoustic impedance between basaltic and sedimentary rocks is usually very pro-
nounced. High-amplitude reflections of various geometries (layer parallel, saucer shape,
climbing, eye shape, dome shape, etc.) unravel sills, dikes, buildups and other igneous
features (Planke et al. 2005). However, sedimentary features, such as salt layers or gas-
saturated layers, can also produce a seismic pattern (high reflectivity) similar to magmatic
structures (de Figueiredo et al. 1982). In addition, processes of thermal metamorphism and
interdigitations of the sedimentary layers may hinder the detection of basaltic rocks. There-
fore, the seismic interpretation is constrained by geological and geophysical data from
boreholes, which provide an accurate location of the intrusions in depth (Smallwood and
Maresh 2002; Schofield et al. 2017).
In order to map the occurrence of litho-structural assemblages of Mesozoic igneous
activities in surface and various depth levels across the Parnaíba basin we adopted a mul-
tidisciplinary approach integrating: (a) compilation of previous geological maps; (b) mag-
netic anomaly enhancement techniques; (c) semiautomatic mapping; (d) seismic and well
data interpretation, supported by seismic attributes; and (e) geochemical clustering, using
major elements. Based on typical magnetic patterns the areal distribution of near-surface
basaltic bodies could be determined and afterward confirmed in field surveys. Seismic and
well data reveal the behavior and extension of dike and sill complexes within the basin
infill with high accuracy, while geochemical data of outcropping rock samples allow us to
address the distribution of the two magmatic events identified in the Parnaíba basin, the
Mosquito and Sardinha formations.
The main aims of the present study are: (a) mapping magmatic rocks in the Parnaíba
basin using magnetic, seismic data and well data; (b) discriminating the superficial

13
Surv Geophys

occurrence of the two main magmatic events, using magnetic susceptibility and geochem-
istry data; and c) discussing the distribution and emplacement of both magmatic groups
in terms of the Pangea breakup. In addition, the technique of self-organizing maps (SOM)
was applied to improve the process of mapping magmatic rocks based on magnetic and
geochemistry data.

2 Parnaíba Basin

The cratonic Parnaíba basin is one of the largest Paleozoic basins in the South American
Platform. It is located in NE Brazil (Fig. 1) and covers a Precambrian basement composed
of Archean–Early Proterozoic cratonic blocks, Late Proterozoic Brasiliano/Pan-African
fold belts and basement inliers (Cordani et al. 1984, 2009; de Castro et al. 2014). The tec-
tono-stratigraphic evolution of the Parnaíba basin involved a Cambrian–Ordovician buried
continental rifting followed by a long period of sag sedimentation (Oliveira and Mohriak
2003; de Castro et al. 2016). Extensive Brasiliano shear zones partially controlled the up
to 4500-m-deep rift structures, especially in the central-western part and at the eastern
edge of the basin. Considerable brittle reactivations of these crustal structures occurred
sometimes along the basin history, perhaps related to the first stage of mid-Atlantic rifting
during the Mesozoic (Arora et al. 1999). The post-rift phase of infilling comprises four
Silurian to Late Jurassic tectono-sedimentary sequences and two main Mesozoic magmatic
pulses (Góes and Feijó 1994).
The Brasiliano/Pan-African tectonic events marked the final orogenic collage of West
Gondwana with overall amalgamation of cratonic blocks and accretion of extensive fold
belts between the late Neoproterozoic and early Cambrian (e.g., Almeida et al. 2000; Fuck
et al. 2008; Brito Neves and Fuck 2014). Thereafter, widespread shift in the tectonic regime
took place, causing crustal extension and rifting in West Gondwana, from the Amazonian
toward the West African cratons and from the São Francisco toward the Congo cratons,
without continental breakup. In NE Brazil, a complex Cambrian–Ordovician rift system
was formed primarily along brittle-reactivated Brasiliano shear zones. A few NE- to E–W-
trending exposures of graben-like structures are concentrated at the eastern and southern
boundaries of the Parnaíba basin (Cordani et al. 1984; Oliveira and Mohriak 2003). How-
ever, these small troughs are just a minor part of large aborted rifts, concealed beneath the
Parnaíba basin, as seismic data unraveled (Góes et al. 1990; Ferreira 2013). Combining
gravity and magnetic anomalous patterns and seismic data, de Castro et al. (2016) depicted
two main rift zones buried by post-rift sequences. In the eastern portion of the basin, the
Transbrasiliano Rift Zone (TRZ in Fig. 1), an elongated rift zone up to 150 km wide, was
formed by brittle reactivation of old shear zones in an oblique-slip crustal extension set-
ting with a 30-km-wide symmetric central graben anchored by the Transbrasiliano shear
zone. To the west, the up to 150-km-wide Western Rift Zone (WRZ in Fig. 1) exhibits a
NW–SE-oriented main axis, orthogonal to ductile shear zones and regional metamorphic
foliations within the Parnaíba block. The central graben is up to 70 km wide and 4.5 km
deep and exhibits a flat bottom.
Cambrian–Ordovician sedimentation (Sequence I—Jaibaras Formation) filled the
rift zones at a subsidence rate of 47 m­ My−1 (de Castro et al. 2016). This ~ 3.0-km-thick
basal sequence is constituted of immature siliciclastic sediments deposited in alluvial fans
and fluvial to lacustrine systems (Oliveira 2000). After the rift process ceased, a long-
term thermal and episodic subsidence established a vast epirogenetic depression, where
continental to shallow marine sag deposition occurred in five main episodes between

13
 Surv Geophys

the Silurian and Late Cretaceous. The Silurian–Early Devonian sandstones and shales
(Sequence II—Serra Grande Group) partially crop out at the eastern border of the basin
and become deeper westward (Figs. 1, 2), indicating that the basin depocenter migrated
from east to west during the post-rift phase. The Middle Devonian–Early Carboniferous
(Sequence III—Canindé Group) and Late Carboniferous–Early Triassic (Sequence IV—
Balsas Group) sequences represent two transgressive–regressive cycles separated by global
orogenic events in Gondwanaland (Góes and Feijó 1994). Throughout this long time, the
low to moderate thermal subsidence caused these tectono-sedimentary units (conglomer-
ates and sandstones) to become increasingly thicker in the central portion of the basin,
which assumed a saucer shape typical of cratonic sag basins (Fig. 2). The slight thickening
of the Sequence II (Canindé Group) can be mostly accounted for larger Devonian subsid-
ence rates (Sousa 1996; Zalán 2004; de Castro et al. 2016). The incipient Mesozoic sedi-
mentation is a direct consequence of extensional events of the Equatorial Atlantic opening
and their counterpart magmatic events (Góes et al. 1990; Vaz et al. 2007). The Jurassic
Sequence V (Pastos Bons Formation), in turn, is constituted of thin siliciclastic layers dis-
cordantly overlying the Sequences III and IV. During Aptian times, extensional tectonics
caused incipient mechanical subsidence in the northernmost part of the basin, moving
the depocenters northward close to the newly formed continental margin. Siliciclastic and
carbonatic sediments (Cretaceous Sequence VI—Codó, Corda, Grajaú and Itapecuru For-
mations) were deposited in arid, fluvial, lacustrine and proximal platform environments
(Vaz et al. 2007). Finishing the basin infill in Cenozoic times, alluvial, aeolian and coastal
deposits overlay large areas of the Parnaíba basin.
The oldest Mesozoic magmatic event in the Parnaíba basin comprises extensive tholei-
itic flood basalts cropping out in the western and southern basin sectors (Fig. 1) (e.g.,
Cordani 1970; Góes and Feijó 1994). The Early Jurassic Mosquito Formation is closely
related to the Central Atlantic Magmatic Province (CAMP), which is the precursor of
the Pangea fragmentation and Central Atlantic Ocean opening (e.g., Bellieni et al. 1990;
Marzoli et al. 1999). Merle et al. (2011) identified three chemical groups of the Mosquito
Formation (low-Ti, high-Ti and evolved high-Ti tholeiites) with mean 40Ar/39Ar ages of
199 ± 2.4 Ma. Combined effects of edge-driven convection and large-scale mantle warm-
ing under the Pangea are suggested by the authors as a viable genetic mechanism for the
CAMP magmatism in the Parnaíba basin. The second Mesozoic magmatic episode affected
the eastern portion of the Parnaíba basin in the Early Cretaceous (Bellieni et al. 1990;

Fig. 2  Schematic geological transect crossing the Parnaíba basin (after Góes et al., 1992). 1—Precambrian
basement; 2—syn-rift sequence; 3—Silurian; 4—Devonian–Carboniferous; 5—Permian–Triassic; 6—mag-
matic rocks; 7—Cenozoic to recent sedimentary cover

13
Surv Geophys

Fodor et al. 1990). The so-called Sardinha Formation, in turn, comprises high-Ti tholei-
itic sills and dikes with 40Ar/39Ar ages of 129–124 Ma. High-Ti magmas may relate to
Early Cretaceous decompressional melting associated with the rifting process of the South
Atlantic Ocean opening. Igneous rocks from the Sardinha Formation correspond quite well
in age, geochemistry and isotope characteristics to those from PEIP (Bellieni et al. 1990)
and from Rio Ceará–Mirim Dike Swarm (Bellieni et al. 1992).
In the subsurface, seismic sections and exploratory wells unraveled extensive tholeiitic
sills and dike swarms widespread within the Parnaíba basin. Several layers of magmatic
rocks (mainly diabase) were drilled in a considerable number of exploratory wells (Góes
et al. 1990; Góes and Feijó 1994). Some sills extend for more than 200 km along the seis-
mic sections, showing high-amplitude reflectors (Ferreira 2013; de Castro et al. 2016).
However, the actual areal distribution of the magmatic events within the basin remains
unknown. Recently, an interesting and useful workflow using magnetic signature of the
igneous events points out considerable zones of concentration of magmatic rocks in the
northeastern, western and southern portions of the basin (Mocitaiba et al. 2017). Com-
bination of those magnetic data with published geological and geophysical information
allows us to verify that up to three layers of sills were emplaced in the basin infill, mainly
affecting the Sequence III (Canindé Group), secondarily the Sequence II (Serra Grande
Group) and rarely the Sequence IV (Balsas Group) (Góes et al. 1990; Góes and Feijó 1994;
Ferreira 2013) (Fig. 2). Taking into account intrusive features observed within the buried
grabens, such as sills intercalating Cambrian–Ordovician rift sedimentary sequences that
belong to the Jaibaras Group, a new interpretative framework could be achieved (de Castro
et al. 2016 and this study). Intrusive structures are clearly interpreted from available seis-
mic sections, for instance dikes and sills, interbedded with siliciclastic rocks in the Jaibaras
rift at the NE boundary of the Parnaíba basin, which could be interpreted as a new igneous
event in the basin history probably Ordovician in age. These results are improved in the
present paper by integrating magnetic maps with seismic data, well logs and geochemistry
of exposed rocks.

3 Geophysical and Geological Datasets

3.1 Magnetic Data

An airborne magnetic dataset of the Parnaíba basin was compiled from eight surveys car-
ried out by the Brazilian Geological Survey (CPRM) and the Brazilian Oil Agency (ANP)
(Fig. 3). The total magnetic intensity (TMI) data were corrected for the diurnal variations
and International Geomagnetic Reference Field (IGRF) and afterward leveled or micro-
leveled. Table 1 shows the data acquisition characteristics for each survey. The datasets
were collected from 1975 to 2006. Each survey was interpolated onto a 250-m cell size
grid, using the bidirectional method. A decorrugation filter combined with a directional
cosine filter was applied to some grids to eliminate high-frequency noise along the flight
direction (Table 1).
Each magnetic grid was processed separately to enhance the geophysical pattern of the
magmatic structures. The processing routine includes reduction to the magnetic north pole
(RTP), matched-filter analysis (MFA) and analytical signal amplitude (ASA). The RTP
transformation centers magnetic anomalies over their causative sources, making the inter-
pretation easier and more reliable (Blakely 1996). A pseudo-inclination factor of 60° to 85°

13
 Surv Geophys

Fig. 3  Schematic map of the Parnaíba basin with location of the airborne magnetic surveys (colored num-
bers), seismic profiles (gray lines), and exploratory wells (yellow triangles). Labeled seismic profiles (black
lines) and wells (orange triangles) are described in the text. Colored areas represent exposure of the Mos-
quito (green) and Sardinha (blue) Formations. Brasiliano structures: AR—Araguaia Suture Zone; TB—
Transbrasiliano Lineament

Table 1  Airborne geophysical survey specifications

Project 0050 1022 1027 1044 4031 4032 4047 4050

Year 2004/2006 1975 1976 1979 1981 1983 1988 1989

NFH (m) 100 150 150 150 300 700 500 800
FLD N–S N–S N–S N30°W N–S N45°E N–S N–S
FLS 500 1000 2000 1000 2500 2000 3000 3000
TLD E–W E–W E–W N60°E E–W N45°W E–W E–W
TLS 4000 20,000 20,000 20,000 8000 12,000 18,000 18,000
MFS (nT) 24,919 27,220 25,670 26,095 26,378 26,845 26,808 25,852
MFI − 11.36° 1.99° − 5.85° − 6.13° − 1.15° 3.29° − 0.83° − 8.23°
MFD − 20.65° − 19.70° − 18.78° − 20.42° − 19.43° − 18.64° − 20.43° − 20.86°

FLD flight-line direction, FLS flight-line spacing, TLD tie-line direction, TLS tie-line spacing, MFS mag-
netic field strength, MFI magnetic field inclination, MFD magnetic field declination, NFH nominal flight
height

13
Surv Geophys

was used to suppress the amplitude and power in and near the declination direction at low
magnetic latitudes (± 20°). The magnetic inclinations and declinations, which were calcu-
lated in the center point of each survey using the IGRF, are described in Table 1. Figure 4a
shows the RTP magnetic anomalies of the Parnaíba basin.
The matched-filter analysis is an inversion technique that iteratively fits multiple straight
lines to radially averaged power spectra of the magnetic or gravity anomalies (Phillips
2001), thereby providing estimates of source depth for an equivalent source layer. With
these depth estimates, amplitude and Wiener filters can be designed to emphasize anoma-
lies originating from various layers. Each depth layer corresponds to the maximum aver-
age depth of a statistical population of true sources (Curto et al. 2014). The matched-filter
analysis allowed us to identify four primary magnetic zones in the Parnaíba basin (Fig. 5):
(1) zone of deep magnetic sources between 12 and 26 km depth related to structures in
the lower crust; (2) intermediate I zone between 3.1 and 6.3 km depth related to basement
features; (3) intermediate II zone between 1.4 and 0.8 km related to sills and dikes within
the basin; (4) shallow zone between 0.2 and 0.3 km depth related to near-surface magmatic
rocks. In addition, the analytic signal amplitude (ASA) of the intermediate II magnetic
anomalies was calculated to further enhance the magnetic response of the magmatic bodies
(Figs. 4b, 5). The amplitude of the analytical signal was derived from the three orthogonal
gradients of the total magnetic field in order to enhance magnetic lineaments and magnetic
source boundaries (Roest et al. 1992).

3.2 Seismic Data and Well Logs

The seismic and well datasets comprise 41 time-migrated 2-D seismic lines and 24 explor-
atory wells obtained from the Brazilian Oil Agency (ANP) (Fig. 3). The seismic data were
acquired in the 1970s and 1980s by the Brazilian Oil Company (Petrobras) and since 2002

Fig. 4  Total magnetic intensity (TMI) (a) and analytical signal amplitude (ASA) (b) maps of the Parnaíba
basin, showing the limits of the airborne surveys and magnetic domains related to magmatic structures
(NMD—North Mosquito Domain; SMD—South Mosquito Domain; SD—Sardinha Domain; TD—Transi-
tional Domain). Brasiliano structures: AR—Araguaia Suture Zone; TB—Transbrasiliano Lineament

13
 Surv Geophys

Fig. 5  Magnetic processing workflow applied to enhance the magnetic patterns of the magmatic occur-
rence in the Parnaíba basin. The location of the area selected to demonstrate the processing steps is in
Fig. 4. White line and green areas represent exposure of the Mosquito Formation

by ANP. The previously processed post-stack seismic sections have imaging depths that
vary from 2.5 to 6.6 s two-way time (TWT), or approximately 20 km, which allow us to
investigate the whole basin infill and deep in the upper crust. Post-processing procedure
consisted of frequency filtering and application of seismic attribute. A running average 2-D
filter was applied to suppress high-frequency trace-dependent noise and emphasize hori-
zontally coherent energy of the reflectors (Hall 2007). In addition, we combined the Hilbert
Trace attribute with the Energy attribute to generate a pseudo-relief aspect in the seismic
section and thus highlight high-amplitude reflectors related to magmatic features (Fig. 6).
The Hilbert Trace is the imaginary part of the complex seismic trace that can be computed
from the real trace via the Hilbert transform (Chopra and Marfurt 2007). For the Energy
attribute, the ratio of the sum of squares of the amplitudes to the number of samples within
a specified time window of 4 ms (in this study) is calculated, and this ratio should be close
to the length of the central frequency period of the signal.
Since 1953 wells have been drilled in the Parnaíba basin, covering large areas of the
basin (Fig. 3). Public well data consist of gamma-ray (GR), self-potential (SP), electrical

13
Surv Geophys

Fig. 6  Seismic profile L01 with Hilbert Trace/Energy attribute (a) and seismostratigraphic interpretation
(b), highlighting six layers of extensive, planar and discontinuous sills with saucer-shaped terminations
(green lines). Well log W01 showing the typical gamma-ray (GR), resistivity (IDL) and density (RHO)
signatures of basaltic rocks interlayered in the basin infill between 1050 and 1200 m depth (top of the
Sequence III) (c)

resistivity (IDL) and density (RHO) logs, as well description of drill cuttings, stacked
stratigraphic package along the well (1D), outcrop analogs and, sometimes, time–depth
curves. We used the well logs to identify the main sedimentary sequences and the mag-
matic rocks in different stratigraphic levels within the basin infill (Fig. 6). The time–depth
curves allowed to tie the limits of the main sedimentary and magmatic features (in depth)
to corresponding reflectors (in time). We interpreted the seismic sections based on well ties
and typical geophysical and lithological associations in the Parnaíba basin as established
by Góes and Feijó (1994) and Vaz et al. (2007).
The seismic sections unravel the behavior of the magmatic structures within the Par-
naíba basin. Extensive sills and volcanic buildups occur overall in the basin, interlayered
within both rift and post-rift sedimentary sequences. For example, Fig. 6 illustrates the sag
depositional sequences (II–VI) directly overlying the Precambrian basement and intruded
by many layers of up to 100-km-long sills in the northern part of the basin (L01 in Fig. 3).
Basaltic igneous rocks were drilled in the well W01 between 1080 and 2023 m, which
emplaced into the Sequence III (Canindé Group). Their thicknesses vary from 26 to 95 m,
totalizing more than 290 m in the whole well W01. The typical geophysical responses of
the basaltic rocks in the well logs are box-shaped gamma-ray curves with low and constant

13
 Surv Geophys

counts (< 30 cps), high-resistivity zones (> 150 Ohm.m) and inverted box-shaped density
curves (> 2.89 g/cm3) (Fig. 6c). The box shape of the gamma-ray and density curves indi-
cates abrupt contacts between magmatic and sedimentary rocks. In turn, the sills mostly
intruded parallel to bedding or forming typical saucer-shaped structures and locally faulted
blocks. They can clearly be tracked in the seismic sections by discontinuous, high-ampli-
tude reflectors, highlighted by Hilbert Trace/Energy attribute (Fig. 6a). Climbing and sau-
cer-shaped reflectors help us to differentiate sills from sedimentary sequences with high
impedance contrast, as salt rocks that occur in the Sequence IV (Balsas Group) in the well
W01. Even so, identifying sills in some areas is not an easy task during seismic interpreta-
tion in the Parnaíba basin, since the sheet intrusions (sills and dikes) are not always well
imaged on the seismic profiles. So, we produced a conservative estimate of the sill distribu-
tion from the seismic interpretation.

3.3 Rock Samples

We measured magnetic susceptibilities (MS) of magmatic and sedimentary rocks from the
Parnaíba basin in order to establish the relationship between these rock types and their
magnetic response. Previously mapped exposures of basaltic rocks and areas showing typi-
cal magnetic pattern were visited as measurement points, covering almost the entire basin
(Fig. 7a). We used a portable magnetic susceptibility meter KT-10 v2 (Terraplus Inc.),
which is able to provide the MS with a high sensitivity of 1.0 × 10−6 SI units.
The MS of basaltic rocks are much higher (41 × 10−3–70 × 10−3 SI) than the host sedi-
mentary bedrocks (0.021 × 10−3–1.0 × 10−3 SI) (Table 2) mainly due to the higher iron
content of the magmatic rocks. This expressive MS contrast is the main cause of the high
magnetic anomalies in the Parnaíba basin (Fig. 4). Four groups of MS anomalies could be
distinguished, corresponding to the two main magmatic events (Mosquito and Sardinha

Fig. 7  Distribution of magnetic susceptibility (MS) of basaltic (triangles) and sedimentary (circles) rocks
(a) and geochemistry data of basaltic rocks (b), separated in six clusters derived from self-organizing maps
(SOM) analysis. Magnetic domains: NMD—North Mosquito; SMD—South Mosquito; SD—Sardinha;
TD—Transitional. Colored areas represent the exposure of the Mosquito (green) and Sardinha (blue) For-
mations. Brasiliano structures: AR—Araguaia Suture Zone; TB—Transbrasiliano Lineament

13
Surv Geophys

Table 2  Summary of Parnaíba basin magnetic susceptibility (MS in × 10−3 SI) data and major elements
analyses, comprising 41 unpublished, 23 published by Merle et al. (2011), and 40 new samples, collected in
this study

Mosquito formation (45 measurements) Sardinha formation (156 measure-

ments)
Min Max Average Min Max Average

MS 8.60 21.01 14.74 3.73 68.10 29.08

Reddish sandstones (108 measurements) White sandstones (42 measurements)
Min Max Average Min Max Average

MS 0.11 0.92 0.59 0.02 0.08 0.05

Mosquito formation (34 samples) Sardinha formation (70 samples)
Min Max Average Min Max Average

SiO2 47.20 53.90 50.61 46.00 60.90 50.95

Al2O3 12.29 14.70 13.80 10.40 16.60 13.94
Fe2O3 8.95 16.00 11.20 7.84 16.90 12.50
MnO 0.13 0.27 0.18 0.10 0.23 0.18
MgO 2.52 7.68 6.32 0.86 7.37 4.72
CaO 0.98 10.95 9.29 3.01 10.10 7.57
Na2O 1.38 4.52 2.22 1.92 7.97 3.49
K2O 0.25 1.49 0.86 0.36 4.47 1.81
TiO2 0.99 3.51 1.49 0.97 3.90 2.47
P2O5 0.10 0.82 0.21 0.08 1.68 0.72
LOI 0.44 6.39 2.93 0.23 4.54 1.01
Na2O + K2O 1.84 5.47 3.08 2.64 9.30 5.30

LOI is the loss on ignition

Formations), reddish sandstones rich in iron oxide and white sandstones. The Mosquito
Formation rocks show MS ranging from 8.61 × 10−3 to 21 × 10−3 SI with an average value
of 14 × 10−3 SI. In contrast, the Sardinha Formation rocks have higher MS (29 × 10−3 SI),
despite a greater dispersion of their values (3.71 × 10−3 to 68 × 10−3 SI). The highest MS
are concentrated in the eastern portion of the Parnaíba basin (Fig. 7a). High MS basically
means high magnetic mineral content, comprising the titanomagnetite series (­Fe3-xTixO4;
0 < x < 1), also known as the spinel group (Dentith and Mudge 2014). In fact, Sardinha
Formation rocks exhibit average iron and titanium contents higher than Mosquito Forma-
tion basalts (Table 2). Although both formations are composed of similar litho-types, mag-
netic properties are strongly affected by subtle changes in geochemistry and crystallization
or recrystallization processes (Dunlop and Özdemir 1997; Dentith and Mudge 2014). In
this sense, the significant variability in MS response for Sardinha Formation rocks seems
to reflect the larger ranges in iron and titanium contents in comparison with the Mosquito
Formation rocks (Table 2).
Sandstones with iron oxide concretions are responsible for averaged MS of ca.
0.75 × 10−3 SI. They are probably the causative source of magnetic anomalies with inter-
mediate to high amplitudes. Iron oxide-bearing sandstones belong to the Cretaceous
sequence (Corda Formation) (Vaz et al. 2007) that crops out in vast areas northward

13
 Surv Geophys

(Fig. 1). However, we found out higher MS iron oxide concretions in the central and west-
ern sectors of the basin, which are derived from chemical weathering of iron-rich minerals
in local rocks and not necessarily correlated to the Corda Formation.
In order to correlate mineral chemistry contents with magnetic anomalies and magnetic
susceptibility, we assembled 104 samples of magmatic rocks of the Mosquito and Sardinha
Formations. Identifying physical and chemical characteristics that help to distinguish these
two magmatic events and correlating with Atlantic magmatic provinces is an important
goal of this study. Thus, major elements of 41 unpublished and 23 published by Merle et al.
(2011) rock samples are taken into account in this study. Forty new samples were collected
in field trips throughout the Parnaíba basin (Fig. 7b). Major element analyses were per-
formed by using X-ray fluorescence technique at the Laboratório de Caracterização Tec-
nológica of the Universidade de São Paulo (LCT Poli/University of São Paulo, Brazil) and
are listed in Table 3. Analytical uncertainty ranges generally from 1 to 3%. Each accuracy
and precision is, respectively, provided through analyses of the certified basalt (GB-07105)
standard and duplicate analyses of three unknown samples (see data table) carried out dur-
ing the XRF run. In order to be coherent with other references used as a database, we
applied the same criterion of Merle et al. (2011) and discarded all analyses displaying loss
on ignition (LOI) values higher than 4% wt. High LOI contents are commonly found in the
Mosquito lavas due the presence of hydrated (chlorite, iddingsite and clay) minerals, which
are alteration products of olivine and pyroxene under weathering conditions. Zeolites/
carbonate-filled amygdales, as well as the presence of Fe-oxyhydroxides in the devitrified
groundmass, also contribute to the high LOI results.
Chemical compositions listed in Tables 2 and 3 allow separating the Mosquito and
Sardinha basalts by complex relations between elements, specially ­TiO2, ­Fe2O3 and MgO,
as depicted by Fodor et al. (1990). Indeed, Merle et al. (2011) documented three chemi-
cal groups in the Mosquito Formation basalts, while da Silva et al. (2017) identified six
tholeiitic suites from borehole samples located in the eastern portion of the Parnaíba basin,
where Sardinha Formation rocks crop out. In order to group the basalts in a more effective
manner we used an automatic mapping procedure, which works in a hyperspace of vari-
ables. This approach and its results are described in the next section.

4 Self‑organizing Maps (SOM)

Self-organizing maps (SOM) is an unsupervised analysis technique for understanding com-

plex multivariate datasets with subtle relationships (Kohonen 2001). The SOM approach
is a nonlinear projection method, which can automatically convert a dataset of multiple
variables into self-similar nodes or clusters using principles of vectors quantization and
measures of vector similarity (Kohonen 2001; Fraser and Dickson 2007). SOM analysis
has been widely used in fields such as exploratory data analysis or data mining, remote
sensing imagery, identification and monitoring of complex process states, and pattern clas-
sification (Bierlein et al. 2008). This approach has gained widespread acceptance for the
analysis and integration of airborne geophysical data, calibration and interpretation of well
log and seismic data, and correlation of geochemical data (Strecker and Uden 2002; Fraser
and Dickson 2007; Carneiro et al. 2012).
The SOM approach treats each sample as a vector in an n-dimensional data space deter-
mined by its variables. Measures of vector similarity are then used to order and segment
the input data into meaningful natural patterns (Fraser and Dickson 2007). In an automatic

13
 Table 3  New major elements analyses for the Mosquito (M) and Sardinha (S) Formations
Sample Longitude Latitude SiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LOI Na2O + K2O

M2103a − 46.9033 − 6.0122 50.70 14.20 11.20 0.20 6.80 9.70 2.30 1.40 1.20 0.20 2.50 3.65
Surv Geophys

M2503 − 45.9542 − 9.8873 47.80 13.60 16.00 0.22 6.37 9.75 2.14 0.25 1.99 0.16 0.44 2.39
M2601 − 46.9394 − 10.1520 48.20 14.70 15.00 0.21 5.84 10.20 2.12 0.28 1.88 0.15 0.75 2.40
M2703 − 48.0338 − 7.3555 47.20 14.20 15.50 0.24 6.92 9.53 2.06 0.27 1.91 0.14 1.89 2.33
M2801 − 47.4423 − 7.0414 51.60 13.90 10.90 0.17 6.76 9.53 1.69 1.18 1.05 0.12 3.77 2.87
M2802 − 47.4244 − 6.6355 51.00 14.10 10.60 0.17 6.93 9.98 1.93 1.49 1.02 0.10 2.65 3.42
M2803 − 47.3864 − 6.3977 50.00 13.60 10.40 0.17 6.98 10.50 1.38 0.46 1.06 0.14 6.04 1.84
M2804 − 47.2760 − 6.2901 51.70 13.90 10.60 0.17 6.45 9.58 1.86 0.96 1.01 0.11 3.43 2.82
M2805 − 46.1412 − 5.8239 48.00 13.20 10.30 0.14 7.34 8.67 4.52 0.95 1.05 0.10 6.39 5.47
M2902 − 46.1559 − 6.7150 51.90 13.80 10.80 0.17 7.17 9.89 2.42 0.73 1.00 0.10 2.52 3.15
M2903 − 46.1512 − 6.7507 52.40 14.30 11.00 0.18 6.60 9.72 2.29 1.00 1.08 0.13 0.98 3.29
S2301 − 41.5545 − 7.0531 52.80 12.90 14.30 0.21 3.37 6.65 3.24 2.12 3.18 0.80 0.88 5.36
S2302b − 41.7945 − 7.2465 49.00 13.20 15.10 0.21 4.80 8.31 2.66 1.43 3.52 0.47 0.85 4.09
S2302a − 41.7945 − 7.2465 50.90 11.70 14.10 0.20 5.39 8.32 2.33 1.72 3.03 0.63 1.10 4.05
S2303 − 41.8929 − 7.7326 51.20 12.70 13.90 0.17 4.02 6.89 2.70 1.98 3.19 0.69 1.91 4.68
S2304 − 41.8787 − 7.6786 49.60 11.10 14.80 0.21 7.09 9.29 2.01 1.28 2.92 0.46 0.69 3.29
S2305 − 41.8803 − 7.6380 49.90 10.40 14.90 0.23 7.34 9.52 1.92 1.36 2.81 0.48 0.82 3.28
S2306 − 41.5600 − 7.6956 53.80 13.80 12.20 0.16 3.30 5.64 2.98 2.49 2.62 0.82 1.70 5.47
S2401 − 43.0471 − 7.5737 52.00 13.60 12.50 0.19 6.35 10.10 2.05 0.59 1.38 0.14 0.58 2.64
S2402 − 43.0513 − 7.5813 52.00 13.60 12.50 0.19 6.41 10.10 2.09 0.55 1.37 0.14 0.78 2.64
S2904 − 44.2397 − 6.5168 49.60 12.70 14.70 0.21 4.46 7.90 2.50 1.66 3.68 0.57 1.16 4.16
S2905 − 43.7116 − 6.5041 52.50 13.00 13.10 0.19 3.16 6.32 2.91 2.23 2.84 0.81 2.24 5.14
S3001 − 42.9655 − 6.8063 52.80 12.50 14.10 0.21 3.30 6.31 2.96 2.23 3.00 0.89 0.97 5.19
S3002 − 42.8422 − 6.2484 50.60 12.10 14.70 0.21 5.15 8.00 2.37 1.74 3.08 0.61 1.11 4.11
S3003 − 42.1717 − 6.6350 50.50 13.70 10.40 0.17 6.76 7.42 5.27 0.49 1.00 0.08 4.54 5.76

13
 Table 3  (continued)
Sample Longitude Latitude SiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LOI Na2O + K2O

13
S3004 − 42.1450 − 6.2285 49.60 12.90 14.50 0.20 4.99 7.88 2.69 1.69 3.36 0.55 0.94 4.38
S3005 − 42.5915 − 5.6657 50.60 12.40 14.30 0.21 4.55 7.85 2.46 1.72 3.16 0.65 1.38 4.18
S1601 − 41.8592 − 3.4940 52.20 14.40 10.80 0.20 7.40 11.40 2.20 0.60 1.00 0.10 0.60 2.74
S1602 − 41.8972 − 3.5133 53.20 13.40 13.80 0.20 5.60 9.70 2.50 0.80 1.60 0.20 0.30 3.35
S1603 − 41.8500 − 3.3295 49.70 14.10 14.80 0.20 6.20 10.70 2.50 0.40 2.00 0.20 0.40 2.92
S1605 − 42.7209 − 3.6579 54.10 11.80 16.90 0.20 2.70 7.20 3.00 1.30 2.50 0.30 0.80 4.29
S1801a − 42.5483 − 3.6739 53.30 12.90 13.60 0.20 3.10 7.00 2.90 2.40 2.90 1.10 1.00 5.29
S1801c − 42.5483 − 3.6739 52.30 13.90 12.00 0.20 7.00 10.90 2.20 0.60 1.20 0.10 0.50 2.84
S1802 − 42.2887 − 3.7690 52.30 14.50 11.10 0.20 7.50 11.40 2.10 0.70 1.10 0.10 0.20 2.78
S1803a − 42.1334 − 3.9683 52.80 13.30 13.90 0.20 5.50 9.80 2.40 0.90 1.60 0.20 0.50 3.27
S1803b − 42.1334 − 3.9683 52.80 13.30 14.00 0.20 5.50 9.80 2.40 0.80 1.60 0.20 0.40 3.24
S1804 − 41.4732 − 4.4308 49.00 12.00 19.30 0.20 4.60 8.90 2.40 0.90 2.70 0.20 0.30 3.29
S1806 − 41.9008 − 4.7402 51.20 14.60 11.30 0.20 6.20 10.20 2.40 0.90 1.20 0.10 0.60 3.30
S1807 − 42.6156 − 4.8718 52.70 13.70 12.50 0.20 6.20 10.50 2.30 0.90 1.40 0.20 1.80 3.19
S1901 − 42.8278 − 6.2464 51.20 12.20 14.70 0.20 4.90 8.30 2.50 1.80 3.20 0.60 0.60 4.31
M2802/R 51.40 14.10 10.70 0.16 7.05 10.20 2.05 1.49 1.04 0.10 2.27
S2301/R 53.00 12.90 14.20 0.20 3.43 6.76 3.44 2.10 3.21 0.77 0.51
S2904/R 49.80 12.90 14.60 0.19 4.61 8.03 2.68 1.64 3.76 0.57 1.08
GB-07105 44.4 13.75 13.40 0.17 7.75 8.83 3.34 2.29 2.34 0.95
SD (n = 4) 0.09 0.04 0.09 – 0.03 0.02 0.01 0.009 0.01 0.009

LOI is the loss on ignition. The certified basalt (GB-07105) was chosen as the standard samples
SD: standard deviation
Surv Geophys
Surv Geophys

process, seed vectors are modified to represent the distribution of the input data in the
data space and, once trained, become best-matching units. The SOM output consists of an
ordered array of nodes or clusters arranged in a regular, two-dimensional distribution. We
used the SOM implementation developed by the Commonwealth Scientific and Industrial
Research Organization (CSIRO—Australia) (CSOM). The reader is referred to Kohonen
(2001) for more details.
We applied the SOM analysis procedure to two different scenarios of this study: (1)
to assist in the interpretation of magmatic rocks from a suite of magnetic anomaly maps
and (2) to investigate genetic relationships of outcropping basaltic rocks from major ele-
ment analysis of geochemical data. The magnetic data analyzed using CSOM consisted
of five magnetic input variables, derived from matched filter and analytical signal (Fig. 5)
capable of mapping a range of causative sources from different depths and nature. Each
airborne survey was analyzed separately using SOM due to very distinct acquisition set-
ups (Table 1). It means higher-resolution data yield more SOM solutions, since the
expected magnetic responses of magmatic rocks are high-amplitude and short-wavelength
anomalies.
The SOM setup and processing parameters are as follows: (a) the data space that was
randomly initiated and seeded; (b) the map size of 88 rows × 80 columns chosen after
many experimental runs; and (c) the number of clusters, obtained after running the iterative
procedure innumerous times until a minimal number of clusters was achieved. The smallest
number of clusters ranged from 5 to 8. Then, we chose the results of SOM analysis with 8
clusters for each airborne survey. Solely the cluster more representative of basaltic rocks is
depicted in Fig. 8. Shallow rocks with high magnetic content and structural features are the
causative sources of high-amplitude and short-wavelength magnetic anomalies. Certainly,
more rock types were assigned to this cluster, as is evident by clusters in areas beyond the
basin boundaries. A high concentration of SOM results occurs at the western edge of the
Parnaíba basin (Fig. 8), related to orthogneisses, metasedimentary sequences, and gran-
ites of the Precambrian Tocantins Province (Alvarenga et al. 2000). By the way, NE–SW-
trending SOM solutions reflect the Neoproterozoic basement fabric, controlled by exten-
sive Brasiliano shear zones in the eastern border of the basin, where the sediment infill is
a few hundreds of meters thick; intra-basement magnetic sources can provide SOM results
not related to the Mesozoic basaltic rocks. For this reason, we use geological maps, analog
field data and magnetic susceptibility measures to check the consistency of the cluster-
ing yielded by the SOM technique. Spatially, the SOM analysis enlarged the extensions of
known basalt exposures many times throughout the basin.
We also used SOM analysis for a dataset comprising 104 geochemical samples each
with 11 major element variables for samples of basaltic rocks from the Parnaíba basin
(Fig. 7b). In addition, we included the total alkali content ­(Na2O + K2O) as an extra vari-
able in the SOM analysis in the expectation of increasing the chemical clustering of the
basaltic rocks. After a few hundred iterations, a 12 row × 7 column SOM was obtained
with six clusters. Each cluster, randomly numbered from 1 to 6, represents an optimized
classification of each analyzed sample into a chemical group with regard to similarities
between chemical contents of the Mesozoic magmatic rocks in the basin. Although the
rock samples belong to Mosquito or Sardinha Formations, the SOM-derived chemical
groups show common contents of S ­ iO2, ­Al2O3, ­Fe2O3 and MnO and differ in the other
major elements (Fig. 9). According to their chemical characteristics, the Mosquito rocks
are assigned to clusters 1, 2, 4 and mainly 6, restricted to the western and southern por-
tions of the basin (Fig. 7b). In turn, Sardinha basalts are equally distributed in clusters
3, 4 and 5, with slightly more occurrences in cluster 2 (Fig. 9). Spatially, they occupy a

13
 Surv Geophys

Fig. 8  Distribution of SOM-derived clusters related to magmatic occurrence in the Parnaíba basin
based on magnetic patterns. Magnetic domains: NMD—North Mosquito; SMD—South Mosquito; SD—
Sardinha; TD—Transitional. Colored areas represent the exposure of the Mosquito (green) and Sardinha
(blue) Formations. Brasiliano structures: AR—Araguaia Suture Zone; TB—Transbrasiliano Lineament

Fig. 9  Major element patterns of the basaltic rocks classified into six SOM-derived clusters

13
Surv Geophys

large N–S-oriented zone at the eastern part of the basin. Cluster 2 is most prominent in the
northernmost basalt exposures, whereas clusters 3, 4 and 5 are more common southward.
According to Merle et al. (2011), ­TiO2 concentrations are used to define types of tholei-
ites in continental flood basalt provinces. These authors identified three chemical groups in
the Mosquito Formation: low-Ti (­ TiO2 < 1.3 wt%), high-Ti (­ TiO2 ~ 2.0 wt%) and evolved
high-Ti ­(TiO2 > 3.0 wt%) basalts. The SOM analysis was able to properly separate these
groups in the clusters 6, 2 and 4, respectively (Fig. 9). Also based on T­ iO2 concentrations,
the SOM-derived results suggest a similar chemical zonation of the Sardinha magmatism,
where low-Ti basalts (cluster 2) are closer to the Equatorial Atlantic and high-Ti (cluster
5) and evolved high-Ti (cluster 3 and 4) rocks occur around 300–500 km inland (Fig. 7b).
In addition, we note a tendency that low-Ti basalts (clusters 1, 2 and 6) show lower MS
(5–25 × 10−3 SI) and high-Ti and evolved high-Ti basalts (clusters 2, 4 and 5) higher MS
(> 25 × 10−3 SI). On the contrary, ­Fe2O3 concentrations exhibit no clear pattern that could
address the differences in MS.

5 Mapping Magmatic Events in the Parnaíba Basin

5.1 Magnetic Anomalies

Significant contrasts of magnetic susceptibility between basaltic (3.7 × 10−3–68 × 10−3

SI) and sedimentary (0.02 × 10−3–1.0 × 10−3 SI) rocks (Table 2) generate high-amplitude
magnetic anomalies, easily mapped in the Parnaíba basin that normally present the total
magnetic intensity (TMI) with low amplitudes (Fig. 4). Igneous structures exhibit short-
wavelength magnetic anomalies since they are shallow magnetic sources (up to a few kilo-
meters). However, when this magnetic pattern is merged with the magnetic contribution
of deeper basement sources emplaced within the upper crust, a more reliable mapping of
magmatic bodies is a complex task, especially for deeper and smaller bodies.
Using the previously described anomaly enhancement techniques (Fig. 5), the mag-
netic anomalies of the shallow sources were highlighted, allowing the delineation of four
magnetic domains within the basin (Fig. 4). The South and North Mosquito Domains and
Sardinha Domain encompass areas where igneous rocks have been mapped in previous
studies (Vasconcelos et al. 2004; CPRM 2015). However, these magnetic domains extend
in a region much larger than the outcropping rocks, suggesting vast areas at the western,
southern and NE edges of the basin affected by the Mesozoic magmatism in depth. Fur-
thermore, a series of high-amplitude ASA anomalies, named as Transitional Domain (TD
in Fig. 4), occurs in the central sector of the Parnaíba basin, with areal distribution between
the exposures of Mosquito Formation to the west and Sardinha Formation to the east. We
were not able to find outcrops of basic igneous rocks within these magnetic anomalous
areas. Instead, iron oxide concretions occur embedded in sandstones with medium mag-
netic susceptibility (~ 0.75 × 10−3 SI—Fig. 7a). The iron oxide, which comprises these
concretions, can be derived from chemical weathering of iron-rich minerals in buried
basaltic rocks or probably outcropping close to this region. Seismic data reveal buried
magmatic features within this magnetic domain (see Sect. 5.2).
Although high-amplitude ASA anomalies are present in other areas of the basin besides
the above-mentioned magnetic domains (Fig. 4b), they seem to be more easily correlated to
either the Precambrian basement grain or unknown isolated magnetic sources than basal-
tic rocks due to distinct magnetic pattern from those relative to magmatic features. For

13
 Surv Geophys

this reason, we investigated the SOM solutions carefully, comparing them with geologi-
cal maps and performing field surveys. Despite the uncertainties in the SOM results, we
can extract important information about the occurrence of magmatic events in the Parnaíba
basin (Fig. 8). The clusters related to basaltic rocks form large highly concentrated clouds
without apparent orientation, especially in the Mosquito Domains (NMD and SMD in
Fig. 8). In the Sardinha Domain (SD), the cluster clouds are sparser, but also widely dis-
tributed in an area at the NE portion of the basin. On the other hand, SOM clusters trend
N–S and NE–SW at the western and eastern borders of the basin, following the Precam-
brian structural settings of the Tocantins and Borborema Provinces, respectively.
According to SOM results, the northernmost occurrence of the Mosquito Formation
(NMD) is two times larger (~ 350 km) than the mapped exposures on surface (green areas
in Fig. 8) northward. In this area, the lava flows are partially covered by sandstones of the
Sequences IV and V (Balsas Group and Jurassic formations—Fig. 1) and show a slightly
ENE–WSW orientation. The same characteristics are observed in the southern area (SMD),
where the basalt flows occur at the bottom of cliffs formed of Neocretaceous sequences of
the San Franciscana Basin (8 in Fig. 1). Their distribution beneath the basin infill is up to
60 km to the west, with ENE–WSW-trending SOM clusters oriented sub-parallel to the
Transbrasiliano Lineament (inset in Fig. 8). It suggests that the emplacement of the Mos-
quito Formation was controlled by crustal weakness zones of the Precambrian basement
grain. In fact, the lava flows of the NMD seem to be limited by the N–S-oriented Araguaia
Suture Zone at the western basin boundary.

5.2 Seismic Mapping

The seismic data and well logs reveal an even greater range of magmatic events within the
Parnaíba basin. Up to seven layers of sills and dikes occur interlayered in the sedimentary
sequences (Figs. 6, 10 and 11). A considerable number of intrusions exist emplaced in
the basin infill, but they cannot be imaged on the seismic data due to their thickness of
a few meters being less than the seismic vertical resolution. Planke et al. (2005) studied
a magmatic complex emplaced in the Voring and More Basins (NE Atlantic Ocean) and
pointed out that > 18-m-thick sills can be detected under perfect seismic imaging condi-
tions. According to them, quality of the seismic data and acquisition parameters, contrast
of the acoustic impedance between intrusive body and sedimentary bedrock or eventually
metamorphic aureole and structural variations in the overburden make it difficult to obtain
this resolution criterion. Indeed, > 10-m-thick diabase layers, which are drilled by several
wells, are poorly or not imaged on the seismic profiles. So, we hypothesized that our seis-
mic interpretation provides a conservative estimation of the sill distribution in the Parnaíba
basin.
Cambrian–Ordovician rift zones, as described by de Castro et al. (2016) (TRZ and WRZ
in Fig. 1), have been intensively affected by magmatic activities. Their respective central
grabens concentrate the largest amount of magmatic structures with varied intrusion geom-
etries interbedded in the sedimentary infill (Sequence I—Jaibaras Formation), particularly
in the Western Rift Zone (Fig. 11). In the central graben, the sill complex was emplaced
at progressively deeper levels toward the east, at depths ranging from 1.25 to 2.5 s (~ 1.6
to 4.8 km depth). It suggests that rift architecture influenced the emplacement of basalt
sills and dikes, which were channeled, in part, along deep-seated rift faults anchored in
preexisting deep Precambrian basement anisotropies. In fact, the basal post-rift Silurian
Sequence II (Serra Grande Group) shows just one layer of extensive sills close to both

13
Surv Geophys

Fig. 10  Deep (DMA) and intermediate II (IMA2) magnetic anomalies along the seismic profiles L02 (a)
and L03 (b) (location in Fig. 3). Sill complexes are highlighted by green lines. The thick green band indi-
cates the exposure of basaltic rocks on surface. Sedimentary sequences: I—Cambrian–Ordovician, II—
Silurian–E. Devonian, III—M. Devonian–E. Carboniferous, and IV—L. Carboniferous–E. Triassic

main grabens. In both wells W03 and W04 a ~ 70-m-thick diabase layer occurs at 1870 and
1990 m depth, respectively.
Sill complexes are most abundant within the Middle Devonian–Early Carboniferous
Sequence III (Canindé Group). Multiple segmented sills occur as layer-parallel or trans-
gressive planar saucer-shaped and cone-shaped intrusions with few tens to hundreds of
kilometers length (Figs. 6, 10 and 11). Sill terminations are abrupt, sometimes controlled
by normal faults, which played as preferential pathways to the rising magma. Three to six
layers of sill are located in the up to 1500-m-thick Sequence III at depths ranging from 590
to 2630 m. The sills were emplaced at progressively deeper levels toward the north, follow-
ing the deepening of this sequence in the northern and northwestern portions of the basin.
In this region, called Grajaú Basin (Góes 1995), the Canindé Group is overlapped by the

13
 Surv Geophys

Fig. 11  Deep (DMA) and intermediate II (IMA2) magnetic anomalies along the seismic profiles L04 (a)
and L05 (b) (location in Fig. 3). Sill complexes are highlighted by green lines. The thick green band indi-
cates the exposure of basaltic rocks on surface. The deep intra-basement reflections are discussed in the
text. Sedimentary sequences: I—Cambrian–Ordovician, II—Silurian–E. Devonian, III—M. Devonian–E.
Carboniferous, and IV—L. Carboniferous–E. Triassic

1300-m-thick Permian–Triassic and Mesozoic basin infill (Sequences IV–VI), which was
rarely intruded by more than one basaltic sill. In the uppermost sequences, the magmatic
intrusions have a typical thickness of 3–20 m, sometimes reaching 250 m based on well
data. Mostly, they are too thin to be detected by seismic data.

13
Surv Geophys

In order to better recognize the distribution of the magmatic features within the basin
we interpreted the anomalies of the deep (DMA) and intermediate II (IMA2) magnetic
zones along the seismic profiles (Figs. 10, 11). The DMA have higher amplitudes and
wavelengths, varying from − 170 to 145 nT and from 20 to 200 km, respectively. Mag-
netic highs frequently occur on the rift zones, where sill complexes intrude both syn- and
post-rift sequences. DMA also reveal deep intra-basement features in upper to middle
crust, unraveled by new 6.6 s TWT seismic lines acquired by the Brazilian Oil Agency
(ANP) in the Western Rift Zone (Fig. 3). The most prominent of the crustal structures
shows high-amplitude, 45-km-wide reflections dipping northeastward from 9.0 to 12.5 km
(Fig. 11). Based on a deep seismic reflection profile across the Parnaíba basin, Daly et al.
(2014) depicted high-amplitude, sub-horizontal, curved seismic events 20 km deep in the
crust and related them to a large mid-crustal laccolith in the northern portion of the basin.
Tozer et al. (2017) conducted a gravity modeling along this seismic transect, revealing a
relatively dense body in the lower crust. They suggested a magmatic origin for it. Like-
wise, deep reflections in the ANP seismic profiles can also indicate intra-basement intru-
sive features and probably feeder zones for intra-basinal and surface magmatic rocks. The
long-wavelength DMA support this interpretation, suggesting magnetic sources deep in the
crust.
On the other hand, the magnetic pattern of the intermediate and shallow causative
sources is dominated by low-amplitude anomalies (IMA2) with 2–10 km wavelength
(Figs. 10, 11). High-frequency magnetic signals, therefore, occur essentially where basaltic
rocks are partially exposed (profiles L02 and L04). The anomaly amplitudes range from
less than a dozen to 80 nT. It suggests that the contribution of the buried sills to the mag-
netic anomaly field seems to decrease rapidly with depth. De Castro et al. (2016) calculated
the magnetic effect of basaltic sills mapped in a seismic profile crossing the Transbrasi-
liano Lineament. They pointed out that magnetic data acquired in a 1100-m-height air-
borne survey by ANP in 2006 contain no contribution of buried magmatic structures due to
natural decay of the magnetic field with distance of the source. This is the reason why we
used in the present study airborne surveys carried out at 150–800 m height, which provide
better conditions to detect shallower basaltic sills.

5.3 Distribution of Magmatic Structures in Depth

In this study, we integrated geophysical data and analog outcrop information to obtain a
detailed distribution of the magmatic structures within the Parnaíba basin (Fig. 12). The
resulting multidisciplinary mapping provides a conservative estimate of the magmatic dis-
tribution because: (1) boreholes and seismic lines are concentrated in areas of more explor-
atory interest, (2) boreholes did not always drill the whole basin infill, (3) seismic interpre-
tation of sill intrusions is sometimes not reliable due to the limited quality of the seismic
data and dimension and depth of the intrusions, (4) magnetic anomalies show complex pat-
terns, dominated by shallower causative sources, and (5) basaltic exposures are not mapped
in detail throughout the basin. Keeping these limitations in mind, we can state that the
largest volumes of magmatism are found in the central and northern portions of the basin,
where from 4 to 7 layers of sills reach more than 500 m of total thickness. In the northwest-
ern and southern parts of the basin, sills were not penetrated in the wells. This indicates
that the outcropping basaltic rocks of the South Mosquito Domain (SMD in Fig. 12) are
an isolated magmatic event in the southernmost boundary of the Parnaíba basin. At the
eastern boundary of the basin, there are no wells or seismic profiles that could reveal the
presence of magmatic rocks at depth.

13
 Surv Geophys

Fig. 12  Total thickness of basaltic rocks derived from well data (yellow triangles). The numbers indicate
the quantity of drilled sills in each well. Magnetic domains: NMD—North Mosquito; SMD—South Mos-
quito; SD—Sardinha; TD—Transitional

The magnetic Sardinha Domain (SD in Fig. 12) encompasses the largest volume of
igneous rocks in both depth and surface. To the west, the magnetic anomalies and SOM
solutions are sparse (Figs. 4, 8). In this region, the basaltic sill intruded the Canindé Group,
which is overlapped by the 1300-m-thick Permian–Triassic and Mesozoic basin infill
(Sequences IV–VI—Fig. 6). These uppermost sequences attenuate the magnetic effect
of the underlying sills. In contrast, to the north the magnetic Northern Mosquito Domain
(NMD in Fig. 12) shows a thinner set of sills, varying from a few meters to the south to
350 m at its NE edge. We expected a larger number of sills at depth, since this region rep-
resents the largest basaltic rock exposure area in the entire basin. Interestingly, there is no
evidence of any assemblage of igneous rocks in wells located to the south of the basalt out-
crops. In the magnetic Transitional Domain (TD in Fig. 12), the occurrence of magmatic
features is similar to the northern part of the basin, where no outcropping basaltic rocks
were found, but an up to 250-km-wide set of sills is emplaced between 600 and 1750 m of
depth (Fig. 11b). In summary, the magmatic distribution revealed by geophysical and field
data suggests that the Jurassic Mosquito event mainly yielded surface lava flows, whereas
the Cretaceous Sardinha episode caused deep intrusive structures, especially sill and dike

13
Surv Geophys

complexes with extensions of hundreds of kilometers. In fact, different depths of emplace-

ment fit with more evolved compositions of Sardinha Formation basalts compared with
Mosquito Formation rocks, as the geochemistry indicates (Tables 2, 3).

6 Discussion

6.1 Sill Distribution and Emplacement Pattern in the Parnaíba Basin

The seismic and well data reveal several systematic variations in shape and geometry of
the sill complexes with widespread distribution over a region several times larger than the
today exposed basaltic rocks in the western and northeastern basin segments, without clear
structural control by basement grain or basin architecture. The results in this area provide
some integrated interpretation on the igneous activities as follows: (1) Various extensive
layers of sometimes segmented layer-parallel sills are a fundamental form and are com-
mon in the entire basin. Saucer-shaped, climbing and transgressive sills are also mapped
in many basin segments; (2) the sill complexes are preferentially emplaced in the Middle
Devonian–Early Carboniferous Sequence III (Canindé Group), although they also intruded
the lower and upper sequences; (3) sill intrusions are deeper in the NW basin segment,
where Permian–Triassic and Mesozoic basin infill (Sequences IV–VI) overlapped the Can-
indé Group. In the southernmost areas, sill layers become scarce and deeper and are not
penetrated by wells; and (4) very deep reflections might be intra-basement laccoliths and
indicate preferential zones of magma ascent in the crust.
By integrating the magnetic attributes, such as SOM analysis and magnetic data pro-
cessing, it is evident that shallow and superficial occurrence of igneous rocks is more
restricted to the western and northeastern basin segments. Among several characteristics,
the following are highlighted: (1) Four magnetic domains are characterized by high-ampli-
tude and low-wavelength intermediate anomalies and analytical signal, caused by expres-
sive contrast of magnetic susceptibilities between sedimentary and magmatic rocks; (2)
long-wavelength magnetic anomalies are related to deep crustal structures, unraveled in
seismic profiles; (3) sill intrusions covered by a few kilometers thick blanket of sedimen-
tary infill show no remarkable magnetic anomalies due to the natural decay of the magnetic
field with the increase in the causative source depth; and (4) seismic, well and magnetic
data indicate that the Mosquito Formation occurs as lava flows at the western and very
southern basin segments, whereas the Sardinha Formation was emplaced as sill complexes
widespread in vast areas within the basin, rising to the near surface in the central and NE
parts of the basin, where the total thickness of the drilled basalt layers reaches its maximum
values. However, only a systematic geochronology study of the drilled sills can determine
which magmatic events (Jurassic—Mosquito, or Cretaceous—Sardinha) the sill intrusions
belong to.

6.2 Jurassic–Cretaceous Magmatic Events

On the basis of analog outcrops, geochemistry and dating information, Mesozoic magmatic
events in the Parnaíba basin have been clearly differentiated in two events (Bellieni et al.
1990; Fodor et al. 1990; Baski and Archiblad 1997; Merle et al. 2011). While Early Juras-
sic (~ 200 Ma) tholeiitic basalts (Mosquito Formation) were emplaced as lavas flows in the
western and southern basin segments, Early Cretaceous (129–124 Ma) tholeiitic sills and

13
 Surv Geophys

dikes (Sardinha Formation) crop out in the northeastern part of the basin. The geophysi-
cal dataset assembled in this study demonstrates that the Mosquito basalt flows are more
restricted to shallow occurrences in two major regions on the SW side of the basin. This
magmatic activity should be one of the southernmost expressions of the Central Atlantic
Magmatic Province (CAMP) associated with the initial stage of the Equatorial Atlantic
opening (Bellieni et al. 1990; Marzoli et al. 1999; Bertrand et al. 2014). In contrast, exten-
sive Sardinha sill complexes spread through the central, northern and eastern portions of
the basin in an area at least 2.5 times larger than the outcrop region (see Sects. 5, 6.1).
The basaltic rocks of the Sardinha Formation have close chrono-stratigraphic correlation
with the Early Cretaceous Paraná–Etendeka igneous province (PEIP), one of the largest
continental flood basalt provinces in the world (White and McKenzie 1989; Baski and
Archibald 1997), as well with the Rio Ceará–Mirim Dike Swarm in the Neoproterozoic
Borborema Province (Bellieni et al. 1992; Ngonge et al. 2016).
We also correlated magnetic susceptibility distribution with previous and new major
elements analyses of basalt samples to delineate the boundaries of each Mesozoic mag-
matic event. Magnetic susceptibilities for the two pulses differ considerably. Mosquito
basalts systematically show lower magnetic susceptibility than Sardinha magmatic rocks
(Table 2). The magnetic susceptibility of basalts and diabases varies over a wide range
from 0 to 140 × 10−3 SI (Dunlop and Özdemir 1997; Xiong et al. 2016) and depends on
many factors such as the content of ferromagnetic minerals, amount of titanium substi-
tution for iron, oxidation history of the magnetic minerals (Dunlop and Özdemir 1997;
Helm-Clark et al. 2004). In general, the average susceptibility values of volcanic rocks are
lower than hypabyssal and plutonic mafic–ultramafic rocks mainly due to their higher mag-
netic mineral content (Dentith and Mudge 2014). In the Parnaíba basin, we can observe the
same characteristic, confirming field and geophysical data that Mosquito Formation was
preferentially emplaced as lava flows and Sardinha Formation in hypabyssal conditions.
Major elements clustering using the self-organizing maps (SOM) method grouped
the Mosquito and Sardinha basaltic rocks in six clusters distributed in a complex fashion
within the Parnaíba basin (Figs. 7, 9). Since ­TiO2 concentration has been used to define
different types of basalts in continental flood basalt provinces, especially in the CAMP
­ iO2 content to
(e.g., Bellieni et al. 1990; De Min et al. 2003), we paid special attention to T
shed light on the magma sources. In the Parnaíba basin, Merle et al. (2011) identified three
groups of basalts, corresponding to the low-Ti ­(TiO2 < 1.3 wt%), high-Ti ­(TiO2 ~ 2.0 wt%)
and evolved high-Ti (­TiO2 > 3 wt%) basalts in the Mosquito Formation. Da Silva et al.
(2017) documented six basalt suites based on T ­ iO2 concentration in the eastern portion of
the basin, where Sardinha Formation rocks crop out.
The SOM clusters 1 and 6 are related to low-Ti basalts in the North Mosquito magnetic
domain, but two samples of Sardinha have similar geochemical signature. The clusters 2
and 5 comprise the high-Ti group in the North and South Mosquito magnetic domains and
Sardinha rocks at the very NE segment of the basin. The presence of igneous rocks with
similar major elements to the Mosquito Formation on the NE edge of the basin may sug-
gest that the Early Jurassic event occurred also in this region. New geochronological data
should contribute to the determination of the real extension of the Mosquito event within
the basin. Finally, the evolved high-Ti group is represented by clusters 3 and 4, which are
located in the central part of the Sardinha Magnetic Domain and exceptionally on the SW
boundary of the Parnaíba basin. Merle et al. (2011) point out that combined effects of edge-
driven convection and large-scale mantle warming, triggered by mantle plumes, caused the
CAMP magmatism (Mosquito Formation) in the Parnaíba basin. Modern plate tectonic
reconstruction of the South Atlantic rift takes into account the presence of at least two

13
Surv Geophys

mantle plumes beneath the northern South American Platform during the Atlantic opening
(e.g., Heine et al. 2013). Considering separation velocities between South America and
Africa around 2 mm/y in the equatorial continental margin during the South Atlantic ini-
tial rift phase (140–125 Ma), the South American Plate dislocated roughly 150–200 km
westward over a mantle plume in a time window of 75 My, which represents the time inter-
val between the two magmatic events in the Parnaíba basin. This might be an explanation
for the eastward shift in the Sardinha Formation emplacement in relation to the Mosquito
Formation.

7 Conclusions

Regional geophysical and geochemical mapping in the Paleozoic cratonic Parnaíba basin
(NE Brazil) has documented two extensive Mesozoic magmatic events with an extent of
at least 10 times larger than the exposed basaltic rocks. The former event, the Early Juras-
sic Mosquito Formation associated with the Central Atlantic Magmatic Province (CAMP),
apparently occurs restricted to shallow flows in the central-western and southern parts
of the basin, as revealed by anomaly enhancement techniques and self-organizing maps
(SOM) applied on airborne magnetic data. In contrast, the Early Cretaceous magmatic
pulse, the Sardinha Formation correlated to Paraná–Etendeka igneous province (PEIP),
comprises hundreds of kilometers wide multilayered sill complexes with layer-parallel,
saucer-shaped, climbing and transgressive geometries in the central, northern and eastern
basin segments. Intra-basement reflections suggest magmatic features deep in the crust in
the central-eastern sector of the basin, where the largest volume of sills is mapped and the
basaltic rocks crop out more intensively.
Six magnetic domains were delineated in the western, southern and northeastern
regions within the basin based on high-amplitude and short-wavelength magnetic pattern.
Near the surface, magnetic susceptibility data and major element analysis of rock samples
collected in these magnetic domains allowed us to discriminate the two major Mesozoic
magmatic episodes in six clusters derived from the SOM approach. Early Jurassic Mos-
quito igneous rocks show lower magnetic susceptibility than Early Cretaceous Sardinha
Formation, probably related to the volcanic nature of the former magmatism. Consider-
ing the ­TiO2 content, the Mosquito Formation rocks have predominantly low-Ti composi-
tion, although we documented the occurrence of high-Ti and evolved high-Ti basalts. The
Sardinha rocks, on the other hand, belong mainly to the groups of high-Ti and evolved
high-Ti basalts. Regardless of the magmatic pulse, the coexistence of the three chemical
groups suggests combined effects of edge-driven convection and large-scale mantle warm-
ing under the West Gondwana during the Central and South Atlantic opening. The con-
tinued displacement of the supercontinent westward over the deep thermal source caused
a shifted emplacement of the Mosquito and Sardinha magmatic activities in the Parnaíba
basin.

Acknowledgements This research was supported by Conselho Nacional de Desenvolvimento Científico

e Tecnológico (CNPq) (Grant No. 471064/2013-0) and Instituto Nacional de Ciência e Tecnologia para
Estudos Tectônicos (INCT-ET). We acknowledge the Brazilian Agency of Petroleum and Natural Gas
(Agência Nacional de Petróleo, Gás Natural e Biocombustíveis - ANP) and the Brazilian Geological Survey
(CPRM), which provided the seismic, borehole and magnetic data used in our study. We also thank dGB
Earth Sciences and K2 Sistemas for educational licenses of Opendtect and ­Trace®, respectively. We are
grateful to two anonymous reviewers and Surveys in Geophysics Editor Michael J. Rycroft for providing
constructive criticism and suggestions that improved the final version of the manuscript. Many students of

13
 Surv Geophys

the Universidade Federal do Rio Grande do Norte (UFRN) were helpful in data collecting during field trips.
DLC and MHBMH thank CNPq for their research grants.

References
Allen PA, Armitage JJ (2012) Chapter 30: cratonic basins. In: Busby C, Azor A (eds) Tectonics of sedimen-
tary basins: recent advances, 1st edn. Blackwell Publishing Ltd, Hoboken, pp 602–620
Almeida FFM, Brito Neves BB, Carneiro CDR (2000) The origin and evolution of the South American
Platform. Earth Sci Rev 50:77–111
Alvarenga CJS, Moura CAV, Gorayeb PSS, Abreu FAM (2000) Paraguay and Araguaia belts. In: Cord-
ani UG, Milani EJ, Thomaz Filho A, Campos DA (eds) Tectonic evolution of South America. Rio de
Janeiro. 31st International geological congress. pp 183–193
Arora BR, Padilha AL, Vitorello I, Trivedi NB, Fontes SL, Rigoti A, Chamalaun FH (1999) Geoelectrical
model for the Parnaíba Basin conductivity anomaly of Northeast Brazil and tectonic implications. Tec-
tonophysics 302:57–69
Baski AK, Archibald DA (1997) Mesozoic igneous activity in the Maranhão province, northern Brazil:
40
Ar/39Ar evidence for separate episodes of basaltic magmatism. Earth Planet Sci Lett 151:139–153
Bellieni G, Picirillo EM, Cavazzini G, Petrini R, Comin-Chiaramonti P, Nardy AJR, Civetta L, Melfi AJ,
Zantedeschi P (1990) Low- and high TiO2, Mesozoic tholeiitic magmatism of the Maranhão basin
(NE-Brazil): K-Ar age, geochemistry, petrology, isotope characteristics and relationships with Meso-
zoic low- and high TiO2 flood basalts of the Paraná Basin (SE-Brazil). Neues Jahrbuch Mineralogis-
cher Abhandlungen 162:1–33
Bellieni G, Macedo MHF, Petrini R, Piccirillo EM, Cavazzini G, Comim-Chiaramonti P, Ernesto M, Mac-
edo JWP, Martins G, Melfi AJ, Pacca IG, de Min A (1992) Evidence of magmatic activity related
to Middle Jurassic and lower Cretaceous rifting from northeastern Brazil (Ceará-Mirim): K/Ar age,
paleomagnetism, petrology and Sr–Nd isotope characteristics. Chem Geol 97:9–32
Bertrand H, Dostal J, Dupuy C (1982) Geochemistry of early Mesozoic tholeiites from Morocco. Earth
Planet Sci Lett 58:225–239
Bertrand H, Fornari M, Marzoli A, García-Duarte R, Sempere T (2014) The Central Atlantic magmatic
province extends into Bolivia. Lithos 188:33–43
Bierlein FP, Fraser SJ, Brown WM, Lees T (2008) Advanced methodologies for the analysis of databases of
mineral deposits and major faults. Aust J Earth Sci 55(1):79–99
Blaikie TN, Ailleres L, Betts PG, Cas RAF (2014) A geophysical comparison of the diatremes of simple
and complex maar volcanoes, Newer Volcanics Province, south-eastern Australia. J Volcanol Geo-
therm Res 276:64–81
Blakely RJ (1996) Potential theory in gravity and magnetic applications. Cambridge University Press,
Cambridge
Brito Neves BB, Fuck RA (2014) The basement of the South American platform: half Laurentian
(N-NW) + half Gondwanan (E-SE) domains. Precambrian Res 244:75–86
Carneiro CC, Fraser S, Crósta AP, Silva AM, Barros CEM (2012) Semiautomated geologic mapping using
self-organizing maps and airborne geophysics in the Brazilian Amazon. Geophysics 77:17–24
Chang HK, Kowsmann RO, Figueiredo AMF, Bender AA (1992) Tectonics and stratigraphy of the east Bra-
zil rift system: an overview. Tectonophysics 213:97–138
Chopra S, Marfurt KJ (2007) Seismic attributes for prospect identification and reservoir characterization,
vol 11. SEG geophysical developments series. Society of Exploration Geophysicists, Tulsa
Condie KC (1997) Plate tectonics and crustal evolution, 4th edn. Butterworth-Heinemann, Oxford
Cordani UG (1970) Idade do vulcanismo no Oceano Atlântico Sul. Boletim de Instituto de Geociências e
Astronomia (Universidade de São Paulo) 1:9–75
Cordani UG, Vandoros P (1967) Basaltic rocks of the Paraná basin. In: Bigarella JJ, Becker RD, Pinto JD
(eds) Problems in Brazilian Gondwana geology. UFPR, Curitiba, pp 207–231
Cordani UG, Brito Neves BB, Fuck RA, Porto R, Thomaz-Filho A, Cunha FMB (1984) Estudo preliminar
de integração do Pré-Cambriano com os eventos tectônicos das bacias sedimentares brasileiras. Ciên-
cia Técnica Petróleo, Seção Exploração Petróleo 15:1–70
Cordani UG, Brito Neves BB, Thomaz Filho A (2009) Estudo preliminar de integração do Pré-Cambriano
com os eventos tectônicos das bacias sedimentares brasileiras (Atualização). Boletim de Geociências
Petrobras 17:205–219
CPRM - Geobank (2015) Cartas Geológicas do Brasil ao Milionésimo. http://geoba​nk.cprm. gov.br. Last
access in 11/10/2016

13
Surv Geophys

Curto JB, Vidotti RM, Fuck RA, Blakely RJ, Alvarenga CJS, Dantas EL (2014) The tectonic evolution
of the Transbrasiliano Lineament in northern Paraná Basin, Brazil, as inferred from aeromagnetic
data. J Geophys Res Solid Earth. https​://doi.org/10.1002/2013J​B0105​93
da Silva AG, Almeida CN, Valente SC, Almeida Leonardo F B (2017) The petrogenesis of tholeiitic
diabases in eastern Parnaíba Basin: evidence for geochemical heterogeneities in the subcontinental
lithospheric mantle in NE Brazil. Braz J Geol 47(1):109–126
Daly MC, Andrade V, Barousse CA, Costa R, McDowell K, Piggott N, Poole AJ (2014) Brasiliano crus-
tal structure and the tectonic setting of the Parnaíba basin of NE Brazil: Results of a deep seismic
reflection profile. Tectonics 33:1–19
de Castro DL, Fuck RA, Phillips JD, Vidotti RM, Bezerra FHR, Dantas EL (2014) Crustal structure
beneath the Paleozoic Parnaiba basin revealed by airborne gravity and magnetic data, Brazil. Tec-
tonophysics 614:128–145
de Castro DL, Bezerra FHR, Fuck RA, Vidotti RM (2016) Geophysical evidence of pre-sag rifting and
post-rifting fault reactivation in the Parnaíba basin, Brazil. Solid Earth 7:529–548
de Figueiredo AMF, Teixeira L, Amorim J, Carminatti M (1982) Projeto Barreirinhas, reavaliação da
Bacia Cretácea. Área terrestre e marítima. Internal Report. Petrobras, Rio de Janeiro, 58p
De Min A, Piccirillo EM, Marzoli A, Bellieni G, Renne PR, Ernesto M, Marques L (2003) The Central
Atlantic Magmatic Province (CAMP) in Brazil: petrology, geochemistry, 40Ar/39Ar ages, paleo-
magnetism and geodynamic implications. In: Hames WE, McHone JG, Renne PR, Ruppel C (eds)
The Central Atlantic Magmatic Province: insights from fragments of Pangea. AGU Geophysical
Monographs, vol 136. pp 209–226
Dentith M, Mudge ST (2014) Geophysics for the mineral exploration geoscientist. Cambridge University
Press, 454 pp
Drenth BJ, Anderson RR, Schulz KJ, Feinberg JM, Chandler VW, Cannon William F (2014) What lies
beneath: geophysical mapping of a concealed Precambrian intrusive complex along the Iowa–Min-
nesota border. Can J Earth Sci 52:279–293
Dunlop DJ, Özdemir Ö (1997) Rock magnetism. Fundamentals and frontiers. Cambridge studies in mag-
netism series. Cambridge University Press, Cambridge, p 573
Ferreira MA (2013) Bacia do Parnaíba. 12a Rodada—Licitações de Petróleo e Gás. Agência Nacional
do Petróleo, Gás Natural e Biocombustíveis, Rio de Janeiro. http://www.brasi​l-round​s.gov.br/round​
_12/portu​gues_R12/semin​arios​.asp. Last access: 18 June 2014
Fodor RV, Vetter SK (1984) Rift-zone magmatism: petrology of basaltic rocks transitional from CFB to
MORB, southeastern Brazil margin. Contrib Miner Pet 88:307–321
Fodor RV, Sial AN, Mukasa SB, Mckee EH (1990) Petrology, isotope characteristics and K-Ar ages of
the Maranhão, northern Brazil, Mesozoic basalt province. Contrib Miner Pet 104:555–567
Fraser SJ, Dickson BL (2007) A new method for data integration and integrated data interpretation:
self-organising maps. In: 5th decennial international conference on mineral exploration. Expanded
abstracts. pp 907–910
Fuck RA, Brito Neves BB, Schobbenhaus C (2008) Rodinia descendants in South America. Precambrian
Res 160:108–126
Góes AMO (1995) A Formação Poti (Carbonífero Inferior) da Bacia do Parnaíba, PhD Dissertation,
Instituto de Geociências, Universidade de São Paulo
Góes AMO, Feijó FJ (1994) Bacia do Parnaíba. Boletim de Geociências da Petrobras 8:57–67
Góes AMO, Souza JMP, Teixeira UB (1990) Estágio Exploratório e Perspectivas Petrolíferas da Bacia
do Parnaíba. Boletim de Geociências da Petrobras 4(1):55–64
Grant FS (1985) Aeromagnetics, geology and ore environments, I. Magnetite in igneous, sedimentary
and metamorphic rocks: an overview. Geoexploration 23(3):303–333
Hall M (2007) Smooth operator: smoothing seismic interpretations and attributes. Lead Edge 26:16–20
Heine C, Zoethout J, Muller RD (2013) Kinematics of the South Atlantic rift. Solid Earth 4:215–253
Helm-Clark CM, Rodgers DW, Smith RP (2004) Borehole geophysical techniques to define stratigraphy,
alteration and aquifers in basalt. J Appl Geophys 55:3–38
Kohonen T (2001) Self-organizing maps, 3rd extended edition, vol 30. Springer series in information
sciences. Springer, Berlin, Heidelberg
Marzoli A, Renne PR, Picirillo EM, Ernesto M, de Min A (1999) Extensive 200-million-year-old conti-
nental flood basalts of the Central Atlantic Magmatic Province. Science 284:616–618
Merle R, Marzoli A, Bertrand H, Reisberg L, Verati C, Zimmermann C, Chiaradia M, Bellieni G,
Ernesto M (2011) 40Ar/39Ar ages and Sr–Nd–Pb–Os geochemistry of CAMP tholeiites from West-
ern Maranhão basin (NE Brazil). Lithos 122:137–151
Mocitaiba LSR, de Castro DL, Oliveira DC (2017) Cartografia geofísica regional do magmatismo meso-
zoico na Bacia do Parnaíba. Geologia USP. Série Científica 17:169–192

13
 Surv Geophys

Morgan JP (2015) On the origin of cratonic sag basins: did they sag? Geophysical Research Abstracts,
17, EGU2015-14246
Ngonge ED, Hollanda MHBM, Archanjo CJ, Oliveira DC, Vasconcelos PMP, Muñoz PRM (2016)
Petrology of continental tholeiitic magmas forming a 350-km-long Mesozoic dyke swarm in NE
Brazil: constraints of geochemical and isotopic data. Lithos 258–259:228–252
Nomade S, Knight KB, Beutel E, Renne PR, Verati C, Féraud G, Marzoli A, Youbi N, Bertrand H
(2007) Chronology of the Central Atlantic Magmatic Province: implications for the Central Atlan-
tic rifting processes and the Triassic–Jurassic biotic crisis. Palaeogeogr Palaeoclimatol Palaeoecol
244:326–344
Oliveira DC (2000) Stratigraphic interplays between igneous and sedimentary events in the early Paleo-
zoic Jaibaras trough (Northeast Brazil). Revista Brasileira de Geociências 30(3):427–431
Oliveira DC, Mohriak WU (2003) Jaibaras trough: an important element in the early tectonic evolution
of the Parnaíba interior sag basin, Northern Brazil. Mar Pet Geol 20:351–383
Peate DW (1997) The Paraná-Etendeka province. In: Mahoney JJ, Coffin MF (eds) Large igneous prov-
inces: continental, oceanic and planetary flood volcanism. American Geophysical Union, Washing-
ton, pp 217–245
Phillips JD (2001) Designing matched band-pass and azimuthal filters for the separation of potential-
field anomalies by source region and source type. In: 15th geophysical conference and exhibition,
Australian society of exploration geophysicists, extended abstract. pp 1–4. https​://doi.org/10.1071/
aseg2​001ab​110
Planke S, Rasmussen T, Rey SS, Myklebust R (2005) Seismic characteristics and distribution of vol-
canic intrusions and hydrothermal vent complexes in the Vøring and Møre basins. In: Doré AG,
Vining BA (eds) Petroleum geology: north-west Europe and global perspectives. Proceedings of the
6th petroleum geology conference. Petroleum Geology Conferences Ltd. Published by the Geologi-
cal Society, London. pp 833–844
Roest WR, Verhoef J, Pilkington M (1992) Magnetic interpretation using 3-D analytic signal. Geophys-
ics 57:116–125
Saunders AD, Jones SM, Morgan LA, Pierce KL, Widdowson M, Xu YG (2007) Regional uplift asso-
ciated with continental large igneous provinces: the roles of mantle plumes and the lithosphere.
Chem Geol 241:282–318
Schofield N, Holford S, Millett J, Brown D, Jolley D, Passey SR, Muirhead D, Grove C, Magee Cr, Mur-
ray J, Hole M, Jackson CAL, Stevenson C (2017) Regional magma plumbing and emplacement
mechanisms of the Faroe–Shetland Sill Complex: implications for magma transport and petroleum
systems within sedimentary basins. Basin Res 29:41–63
Smallwood JR, Maresh J (2002) The properties, morphology and distribution of igneous sills: model-
ling, borehole data and 3D seismic data from the Faeroe–Shetland area. In: Jolley DW, Bell BR
(eds) The North Atlantic igneous province: stratigraphy, tectonic, volcanic and magmatic pro-
cesses, vol 197. Geological Society of London Special Publication, London, pp 271–306
Sousa MA (1996) Regional gravity modelling and geohistory of the Parnaíba Basin (NE Brazil), PhD
dissertation, Department of Physics, University of Newcastle (UK)
Stampfli GM, Hochard C (2009) Plate tectonics of the Alpine realm. In: Murphy JB, Hynes AJ, Keppie
JD (eds) Ancient orogens and modern analogues, vol 327. Geological Society London Special Pub-
lications, London, pp 89–111
Stampfli GM, Hochard C, Vérard C, Wilhem C, von Raumer J (2013) The formation of Pangea. Tectono-
physics 593:1–19
Strecker U, Uden R (2002) Data mining of poststack seismic attribute volumes using Kohonen self-
organizing maps. Lead Edge 21:1032–1036
Telford WM, Geldart LP, Sheriff RE, Keys DA (1998) Applied geophysics, 5th edn. Cambridge Univer-
sity Press, Cambridge
Tozer B, Watts AB, Daly MC (2017) Crustal structure, gravity anomalies, and subsidence history of the
Parnaíba cratonic basin, Northeast Brazil. J Geophys Res Solid Earth 122:5591–5621
Turcotte DL, Schubert G (1982) Geodynamics: applications of continuum physics to geological prob-
lems. Wiley, New York, p 450
Vasconcelos AM, Ribeiro JAP, Colares JQS, Gomes IP, Forgiarini LL, Medeiros MF (2004) Folha ter-
esina SB.23. In: Schobbenhaus C, Gonçalves JH, Santos JOS, Abram MB, Leão Neto R, Matos
GMM, Vidotti RM, Ramos MAB, Jesus JDA (eds) Carta Geológica do Brasil ao Milionésimo, Sis-
tema de Informações Geográficas. Programa Geologia do Brasil. CPRM (CD-ROM), Brasília
Vaz PT, Rezende NGAM, Wanderley Filho JR, Travassos WAS (2007) Bacia do Parnaíba. Boletim de
Geociências da Petrobras 15:253–263

13
Surv Geophys

White RS, Mckenzie DP (1989) Magmatism at rift zones: the generation of volcanic continental margins
and flood basalts. J Geophys Res 94:7685–7729
Xiong S, Yang H, Ding Y, Li Z, Li W (2016) Distribution of igneous rocks in China revealed by aeromag-
netic data. J Asian Earth Sci 129:231–242
Zalán PV (2004) Evolução Fanerozóica das Bacias Sedimentares Brasileiras. In: Mantesso-Neto V, Bar-
torelli A, Carneiro CDR, Brito Neves BB (eds) Geologia do continente sul-americano: evolução da
obra de Fernando Flávio Marques de Almeida. Editora Beca, São Paulo, pp 407–421

13