Investigation of the Goiás Alkaline Province, Central Brazil: Application of gravity and magnetic methods

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Investigation of the Goiás Alkaline Province, Central Brazil: Application of gravityand magnetic methods

Alanna C. Dutra a, Yara R. Marangoni a,*, Tereza C. Junqueira-Brod b

aGeophysics Department, Institute of Astronomy, Geophysics and Atmospheric Sciences, São Paulo University, Rua do Matão 1226, Cidade Universitária,05508-090 São Paulo, SP, Brazilb Instituto de Estudos Sócio Ambientais, Universidade Federal de Goiás, Campus II, Samambaia, Goiânia 74000-000, GO, Brazil

a r t i c l e i n f o

Article history:Received 30 March 2010Accepted 17 June 2011

Keywords:Alkaline intrusions3D inversionAneuler methodGravityMagnetic

a b s t r a c t

We investigate the strong magnetic and gravity anomalies of the Goiás Alkaline Province (GAP), a regionof Late Cretaceous alkaline magmatism along the northern border of the Paraná Basin, Brazil. The alkalinecomplexes (eight of which are present in outcrops, two others inferred from magnetic signals) arecharacterized by a series of small intrusions forming almost circular magnetic and gravimetric anomaliesvarying from �4000 to þ6000 nT and from �10 to þ40 mGal, respectively. We used the Aneuler methodand Analytical Signal Amplitude to obtain depth and geometry for mapped sources from the magneticanomaly data. These results were used as the reference models in the 3D gravity inversion. The 3Dinversion results show that the alkaline intrusions have depths of 10e12 km. The intrusions in thenorthern GAP follow two alignments and have different sizes. In the anomaly magnetic map, dominantguidelines correlate strongly with the extensional regimes that correlate with the rise of alkaline mag-matism. The emplacement of these intrusions marks mechanical discontinuities and zones of weaknessin the upper crust. According to the 3D inversion results, those intrusions are located within the uppercrust (from the surface to 18 km depth) and have spheres as the preferable geometry. Such sphericalshapes are more consistent with magmatic chambers instead of plug intrusions. The Registro do Araguaiaanomaly (w15 by 25 km) has a particular magnetic signature that indicates that the top is deeper than1500 m. North of this circular anomaly are lineaments with structural indices indicating contacts on theiredges and dikes/sills in the interiors. Results of 3D inversion of magnetic and gravity data suggest thatthe Registro do Araguaia is the largest body in the area, reaching 18 km depth and indicating a circularlayered structure.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Alkaline magmatism is characterized by a wide range ofcompositions, ranging from ultramafic to felsic, including a fewcases where rocks in the same complex are extremely different,such as the presence of both silica-undersaturated and silica-oversaturated rocks. There are many evolved rock-types anda wide variety of igneous forms, with intrusive ones prevalent. Themechanism of emplacement and the relative volumes in any givencomplex are not well understood. Gravity studies of alkalinecomplexes indicate that geophysics can be useful in establishingthe relative volumes of the rocks in the suite by using knowndensity contrasts from the literature to model the complexes asvertical cylinders.

Forward modeling of the gravity or magnetic signal of alkalineintrusive complexes have been done by Bott and Tantrigoda (1987)in the intrusive complex located at Mull in NW Scotland, by Dindiand Swain (1988) in the Jombo Alkaline Complex, a largecomplex in Kenya, by Arzamastsev et al. (2000) in Kola AlkalineProvince in the northeastern Scandinavian Shield, by Rugenski et al.(2001) whomodeled the Pariquera-Açu alkaline complex emplacedin southeast Brazil, and by Chandrasekhar et al. (2002) who studiedvolcanic plugs associated with the Deccan Volcanic Province inSaurashtra, India. These complexes display strong gravity andmagnetic anomalies. The Bouguer anomaly for each intrusion issemi-circular or circular. The main magnetic anomalies resemblemagnetic dipoles. Density measurements range from 2.8 to3.4 g/cm3 and susceptibility ranges from 0.02 to 0.05 SI, with strongremnant magnetization (Dindi and Swain, 1988; Arzamastsev et al.,2000; Chandrasekhar et al., 2002). Dutra and Marangoni (2009)measured densities from 2.6 to 3.1 g/cm3 and susceptibility from0.01 to 0.06 SI for fresh rock outcrops in the Goiás Alkaline Province.

* Corresponding author. Tel.: þ55 11 3091 4741; fax: þ55 11 3091 5034.E-mail address: [email protected] (Y.R. Marangoni).

Contents lists available at ScienceDirect

Journal of South American Earth Sciences

journal homepage: www.elsevier .com/locate/ jsames

0895-9811/$ e see front matter � 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.jsames.2011.06.004

Journal of South American Earth Sciences 33 (2012) 43e55

Author's personal copy

The fresh rocks also have a strong remnant magnetization witha Konigsberger ratio, Q, from 3.00 to 65.0, based on laboratorymeasurements. Authors have used geometrical forms resemblingcone trunks or plugs for density models in 2.5D or 3D forwardmodeling (Bott and Tantrigoda, 1987; Chandrasekhar et al., 2002).The body depths in the works mentioned above range from 6 kmwith a density contrast of 0.4 g/cm3 to 12 km with lower densitycontrasts (0.3 g/cm3).

Dutra and Marangoni (2009) obtained a model for two alkalineintrusions from the Goiás Alkaline Province (GAP): Morro doEngenho (ME) and a covered possible body, A2, using inversion ofmagnetic and gravity data to estimate density and magneticsusceptibility at depth.MEandA2 reach amaximumdepth of 10 km,and the density contrast distribution shows a NEeSW preferredorientation of the ME intrusive complex. Because the remnantmagnetization is very important in the area, they inverted the

Fig. 1. Geologic map of the GAP adapted from Brod et al. (2005). Inset shows GAP location. Square marks the study area; numbers refer to the alkaline intrusions.

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e5544

Author's personal copy

magnetic data, testing two scenarios: only an induced field and aninduced field plus remnant magnetizations. The results of theinversion using only the induced field could not explainthe observed magnetic field, which was about 1000 nT larger thanthe calculated one. The induced plus remnant field inferred froma wandering polar path fitted the observations better. Both 3Dmagnetic inversions and 2.5D forwardmodeling recover deeper andsmaller intrusions compared with gravity results, suggestinga concentration ofmagneticmaterial at the centers of the intrusions.

In this paper, we return to the north of GAP province to deter-mine the subsurface structures of the alkaline complexes. In thisstudy, we performed only 3D inversions for the gravity data andused the magnetic data to get geometrical information about thesource, such as its edges, structural index and depth. The edgeswere obtained from the horizontal gradient and analytical signalamplitude maps. The depths and structural indices were obtainedfrom the combination of the Euler deconvolution method with theanalytical signal applied to the total-field anomalies reduced to thepole, using the known direction of the induced plus the remnantfield. This information was part of the initial model for the gravityinversion.

2. Geology and tectonic setting

The Goiás Alkaline Province (GAP, Goiás State) and Alto Para-naíba Alkaline Province (APAP, Minas Gerais State) are a result ofmafic-alkalic magmatism that occurred in Late Cretaceous alonga NWeSE lineament. The Goiás Alkaline Province rocks penetratethe Brasília Belt (Valeriano et al., 2008) basement and the GoiásMagmatic Arc (Pimentel et al., 2000), and they intrude and overliebasalts and sedimentary rocks from the northern border of theParaná Basin, Brazil (Carlson et al., 2007). The Goiás AlkalineProvince includes mafic-ultramafic alkaline complexes in thenorthern portion, subvolcanic alkaline intrusions in the centralregion, and volcanic products to the south with several dikesthroughout the area (Junqueira-Brod et al., 2005). The provinceintrusions are predominantly along a NWeSE track in a region250 km long and 70 km wide, related to mapped faults.

Previous work using seismic tomography and geoid heightshowed possible links between anomalies and the alkaline suite.Molina and Ussami (1999) studied the geoid height in SE Brazil andfound a positive anomaly in the southern São Francisco Craton nearthe sea. The western boundary of this anomaly includes the APAP.According to the authors, this feature could be related to thepresence of density anomalies caused by higher temperatures inthe region. Seismic tomography studies of southwestern andcentral-western Brazil also detected lower velocities in the APAP(Minas Gerais State) and GAP (Goiás State) that may be related toa decrease in material density, a possible result of an increase intemperature (van Decar et al., 1995). According to Assumpção et al.(2004), there is a strong correlation between the locations ofanomalies in the upper mantle and the outcrops of APAP and GAP.

A remarkable feature of the Goiás Alkaline Province is the spatialdistribution of the magmatic components. The province is charac-terized by the presence of plutonic intrusions in the north, diat-remes in central zone, and abundant kamafugite lava flows in thesouth. According to Junqueira-Brod et al. (2005), the northern GAPwas probably uplifted during or after the Late Cretaceous,exhuming the ultramafic intrusions. In the case of ultramaficcomplexes of the Iporá region, the magma probably came straightfrom the mantle, experienced some volatile loss along the way,and was emplaced at the Precambrian/Phanerozoic unconformity.The GAP volcanic rocks are the result of an intricate interplayof differentiation processes, including fractionation, liquid immis-cibility, and magma mixing, which progressively changed the

primitive magma. The way and order in which these processesinteracted determined how the volcanism occurred at each locality.

The extreme northwest limit of the GAP is distinguished by theMorro do Engenho complex (#1 in Fig. 1) in the geophysical studyin Dutra and Marangoni (2009). The mafic-ultramafic alkalineSanta Fé complex (#2) is an ellipsoidal body dating to 86.7� 1.8 Mawith KeAr data from biotite from a missourite (Sonoki and Garda,1988). The Montes Claros complex (#3 in Fig. 1) is composed ofzoned intrusions with dunite in the center and clinopyroxenite,peridotite, alkaline gabbro and syenite at the borders. The Córregodos Bois complex (#5) comprises two intrusions covering an area ofapproximately 33 km2 and the Morro do Macaco complex (#6)comprises a zoned intrusion composed of dunite, wehrlite, olivinepyroxenite and clinopyroxenite from core to rim. The Fazenda Buriticomplex (#7) occupies an area of approximately 35 km2 and hasolivine clinopyroxenite, melagabbro, syenogabbro and syeniteintrusions. The Arenópolis complex (#8) is located to the west ofthe main cluster of plutonic complexes and comprises a NeSelongate elliptical intrusion covering an area of 12 km2. Theseplutonic complexes represent the northern GAP (Brod et al., 2005).

The central GAP includes Amorinópolis and Águas Emendadassubvolcanic areas (#9 and #10, respectively, in Fig. 1). In thesouthern GAP, the volcanic rocks of the Santo Antônio da Barraregion (#11) alternate lavas and pyroclastic deposits. The lava flowsoccupy an area of 371 km2 (Junquiera-Brod et al., 2002) and havea geochemical signature consistent with the plutonic complexesoccurring in the northern GAP. The magmas ascended throughthe N40e50W Santo Antônio da Barra-Iporá tectonic-magmaticlineament.

Published radiometric ages for some of the complexes set themat ages of 75e90Ma. KeArmeasurements for five samples from theSanta Fé complex (Sonoki and Garda, 1988) give an age range of76e80 Ma. Three samples fromMontes Claros de Goiás give an agerange of 89e94 Ma (Sonoki and Garda, 1988). The age range for theIporá Complex is larger than the other complexes: 83e53 Ma(Sonoki and Garda, 1988). Whole-rock measurements for seven

Fig. 2. Total-field magnetic anomaly data from GAP igneous rocks. The white numbersindicate the locations of the alkaline intrusions shown in Fig. 1, and gray contoursrepresent the mapped alkaline intrusions.

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e55 45

Author's personal copy

rocks from Santo Antonio da Barra give an age range from 90 to26 Ma. Two measurements from analcitites give very young agescompared to the rest of the area: 40.1 � 1.2 Ma and 26.3 � 2.9 Ma.Carlson et al. (2007) set an age range of 80e85 Ma for the area andGibson et al. (1995) set an age of 78e90 Ma based on theirunpublished data. Based on published ages, we can consider theGoiás Alkaline Province to be from the Late Cretaceous.

3. Magnetic data

The aeromagnetic data of the Goiás Aerogeophysical Project(LASA, 2004) used here were acquired with NeS lines spaced at500 m and EeW control lines 5 km apart. The survey was at

a constant height of 100 m flown between June and November2004. Observations were taken at a rate of ten measurements persecond positioned with GPS coordinates (accurate to �10 m). Moredetails of the processing of the raw data can be found in the surveyreport (LASA, 2004). Due to aliasing effects, we do not have a goodrepresentation of anomalies smaller than 1 km, two times thespatial sample interval (Telford et al., 1990; Blakely, 1996). Thealiasing effect expected from the ratio clearance/line spacing isabout 20% (Reid, 1980) because most of the intrusions reach thesurface. In Fig. 2, we can observe that the magnetic anomaliesrelated to the alkaline intrusions are usually more than 10 km indiameter and that the linear features extendmore than 20 km; bothare large structures rooted in the upper crust and are crossed byvarious flight lines. Although some signals can be lost, we considerthat the anomalies are well represented for our purpose.

Fig. 2 shows the residual magnetic field after correction fordiurnal variation and subtraction of the International Geomagnetic

Fig. 4. Euler solutions for (a) magnetic source depth and (b) structural index (geom-etry) at the northern GAP. Red lines represent the mapped alkaline intrusions. (Forinterpretation of the references to color in this figure legend, the reader is referred tothe web version of this article.)

Fig. 3. (a) Analytical Signal Amplitude (ASA) and (b) horizontal gradient derivative ofthe reduced to the pole anomalies of the alkaline intrusions in the northern province.The black lines are geological lineaments and faults. White numbers refer to thealkaline intrusions of Fig. 1, and red contours represent the mapped alkaline intrusions.(For interpretation of the references to color in this figure legend, the reader is referredto the web version of this article.)

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e5546

Author's personal copy

Reference Field (IGRF). The observed rough magnetic relief mayindicate that the spacing of the flight lines and the survey altitudeare good enough to represent the major anomalies in the area.

The anomalies related to the alkaline intrusions have normalpolarization (i.e., positive lobe in the north) and appear as distinct

and isolated features in a smooth background. A few NEeSWmagnetic lineaments are also observed.

Quantitative magnetic data analysis was performed in thefrequency domain using Fast Fourier Transform (FFT) filters. Toavoid edge effects, we expanded the grid by 10% of its size, with

Fig. 5. Depth and structural index solutions plotted on the vertical gradient map with the edges of the main GAP intrusions marked: (a) Registro do Araguaia; (b) Santa Fé; (c)Montes Claros de Goiás; (d) Morro do Macaco, Buriti, Córrego dos Bois, Amorinópolis; (e) Arenópolis. Circle color indicates depth, and size indicates structural index. Red linesrepresent the mapped alkaline intrusions, and black numbers indicate the locations of the alkaline intrusions shown in Fig. 1. (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e55 47

Author's personal copy

regular values extrapolated from the original grid. The frequenciesabove the Nyquist wavenumber (1.9 cycles/km) were eliminatedbecause they most likely represent noise.

3.1. Magnetic data processing

The potential field T(x, y) reduced to the pole was used toperform the magnetic analysis discussed below. To perform thereduction to the pole, the direction of the inducing field and theremnant field were summed. The remnant field was measured atthe Paleomagnetic lab of IAG-USP.

The Analytical Signal Amplitude (ASA) delimits the horizontalposition of magnetic sources. To determine the analytical signal weused the concepts explained in Nabighian (1972), Thompson (1982)and Blakely (1996). The analytical signal amplitude obtained fromthe magnetic anomalies reduced to the pole in Fig. 3a shows nearlycircular features correlated with alkaline rocks numbered in white.

Assuming that each magnetic anomaly is produced by an iso-lated body, such as a vertical cylinder, the horizontal diameters ofthe sources can be estimated from the figures as varying from 4 to10 km. We also notice a structural pattern characterized bymagnetic lineaments north of 16�S alignedmainly in the NNEeSSWdirection, and NWeSE in the rest of the area. Some of them arecoincident with mapped faults (black lines in Fig. 3).

The horizontal derivatives were used to emphasize horizontalgradients of the potential field anomaly reduced to the pole thatmay indicate abrupt lateral changes in physical properties (Fig. 3b).In this figure, the circular shape of the intrusions is very clear. InFig. 3, the covered Registro do Araguaia (#12) anomaly hasa particular signature. The anomaly seems to be divided intosegments: an almost smooth oval area, 15 by 25 km, surrounded bysmall anomalies forming a rough border, and an extended beltoriented NNEeSSW, from the oval segment to the limit of thesurveyed area. To the north of Santa Fé anomaly (#2) is the biggestmagnetic lineament, almost 20 by 5 km, correlated with mappedfault. Detailed results for these features are discussed later.

3.2. Depth and geometry of the magnetic sources

The “AN-EUL”, based on the combination of the Euler decon-volutionmethodwith the analytical signal (Salem and Ravat, 2003),allows the estimation of the structural index and depth of theanomalous sources at the coordinate (x0, y0) of the maximumamplitude. AN-EUL solutions were obtained for each intrusion(Fig. 4) to correlate the depth of the tops of the magnetic sourcesand geometry with mapped fault locations. We used a 125 m cellsize with awindowof 10 times the cell size. Themaximum distanceacceptable from the window center was 50 km, and the maximumdepth tolerance was 15%; this small tolerance allows more reliablesolutions. Anomaly sources are shallow, varying from less than100 me400 m, with less than 10% at a greater depth. The edges arewell marked only in the Santa Fé (Fig. 5b) and Registro do Araguaia(Fig. 5a) areas. The rest of the alkaline intrusions also includesolutions inside the bodies. The lineaments noted in Fig. 3 havedepths shallower than 400 m.

The structural index (SI) (Thompson, 1982) is a measure of therate of change with the distance of a potential field. The magneticfield of a punctual dipole falls off as the cube root of distance, givingan index of three, while an effective vertical line source such asa narrow, vertical pipe gives rise to a square root field falloff, or anindex of two. Extended bodies are assemblages of dipoles and haveindices ranging from zero (infinite sheet) to three. We notice thatstructural index solutions with depths greater than 800 m havea structural index of 2 or 3 (Fig. 4). Because we are interested inspherical or cylindrical body shapes, we assume an index equal to 2

or 3 for the alkaline intrusions. Solutions indicating a contact (index0) appear along the edges of the anomalies. Structural indices of 1(dikes/sills) appear along the lineaments (northern anomalies) andcrossing inside the anomalies in the southern area, coincident withthe NWeSE lineaments, which are well marked by the horizontalgradient.

In Fig. 5a, the Registro do Araguaia anomaly goes from 400 m to1000 m depth and a structural index of 3. In the Morro do Engenhoanomaly (#1 in Fig. 5a), most solutions, such as a cylinder, areshallow and appear centralized at the main circular feature. TheArenópolis magnetic anomaly (#8 in Fig. 5e) has concentratedsolutions with a structural index of 2 and depth greater than 600m.Based on these results, we set the initial reference model for the 3Dgravity inversion, with the parameters listed in Table 1.

4. Gravity data set

The gravimetric datawere acquired in the 1980s and three othersurveys were performed between 2004 and 2008. Gravity acquisi-tion and reduction was done according to Dutra and Marangoni(2009). The gravity stations cover an area of about 150 � 100 kmdistributed along roadswith spacing varying from 2 to 5 km in areascloser to the alkaline intrusions or other anomalies. Fig. 6 shows theBouguer anomaly map with the gravity station distribution. Wecorrelated the high values in circular gravity anomalies with knownoutcropping alkaline intrusions or with magnetic anomalies prob-ably produced by buried alkaline intrusions as Registro do Araguaia.Most anomalies have a corresponding magnetic signal.

The lowest negative values are in the southwest (Fig. 6) over theParaná Basin and range from �70.0 to �100.0 mGal. The positiveanomalies are associated with some known outcropping alkalineintrusions mapped by geophysical methods. Their amplituderanges from �10.0 to þ40.0 mGal, and they are roughly circular atthe surface, as expected for massive intrusive cores (plugs or stock).

The squares in Fig. 6mark the alkaline complexes of thenorthern,central and southern GAP. Though these areas are part of the sameprovince, the gravity signals are different. Onepossible causemaybedifferences in mineral composition for each area, as explainedbelow. Other possibilities are related to the geological factors, suchas different terrains in the emplacement areas, leading to diversedensity contrasts, magma volumes and composition diversity of thealkaline products, such as density, viscosity and mineral composi-tion, that would result in bodies with varying shapes.

Regarding the mineral composition, we need to consider that inthe northern intrusions, peridotite, pyroxenite, dunite and gabbroare the dominant rocks. Density values for these rocks vary from 3.0

Table 1Parameters of the initial reference model for the 3D gravity inversion.

Alkaline body Xc (km)a Yc (km)a Extension(km)b

Top depth(km)

Area 1:1 e Morro do Engenho 38.0 74.0 8.0 0.13 e Montes Claros de Goiás 70.0 18.0 10.0 0.112 e Registro Araguaia 40.0 46.0 10.0 2.013 e Anomalia 2 56.0 88.0 6.0 2.0

Area 2:3 e Montes Claros de Goiás 22.0 62.0 10.0 0.14 e Diorama 42.0 38.0 4.0 0.15 e Córrego dos Bois 62.0 20.0 8.0 0.17 e Buriti 44.0 18.0 4.0 0.1

Area 3:2 e Santa Fé 22.0 40.0 6.0 0.1

a Xc and Yc refer to Figs. 8e10.b X and Y extension are equal.

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e5548

Author's personal copy

to 3.4 g/cm3 (Arens and Johnson, 1995). These plutonic bodies wereintruded at the contact between basement and sedimentary cover,usually in direct contact with granite rocks and limited at the top bythe sedimentary rocks from the Paraná Basin and alluvium from theAraguaia River, with lower density values ranging from 2.7 g/cm3 to2.3 g/cm3 for granite and sediments. This geological scenarioprovides a strong positive density contrast between the intrusivebodies and the surrounding rocks. Among the subvolcanic andvolcanic rocks in the central GAP, alkaline basalts (2.7e2.9 g/cm3)(Arens and Johnson, 1995), trachiandesite (2.5e2.7 g/cm3) (Arensand Johnson, 1995) and clinopyroxenite (3.1 g/cm3) are present.

In the central GAP, the lower density subvolcanic rocks are setaround the Paraná Basin, which led to lower density contrasts forthis region. In the south, the lava densities range from 2.34 to3.33 g/cm3, with an average of 2.85 g/cm3 (Junqueira-Brod et al.,2005). The lavas are emplaced at the border of the Paraná Basin,and we did not observe a strong gravity anomaly over them. It isnecessary to point out that in the north, GAP gravity stations aregreater in number and closer to each other than in the south. Thismay cause some loss in anomaly definition.

We used the robust polynomial fitting method (Beltrão et al.,1991) to remove the regional effects. Robust polynomial fitting is

Fig. 6. Bouguer anomaly map with gravity stations (dots) and contours at 10 mGal intervals. Squares mark the alkaline complexes of the northern province, central and southernGAP. Red lines represent the mapped alkaline intrusions. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e55 49

Author's personal copy

similar to the least squares method, but it allows anomalies withonly positive or negative signal to be better defined because it is notnecessary that the residual sum goes to zero, as in the least squaresmethod. In this method, a high-order polynomial resulted innegative anomalies at the profile borders that could not beexplained by the local geology.

The choice of the polynomial order was based on the analysis ofprofiles crossing the anomalies. Fig. 7 shows the residual Bougueranomalies obtained by subtracting a fitted surface of order 4 (notshown) from the total Bouguer anomaly. The residual amplitudesrange from 10.0 to 80.0 mGal in circular anomalies centered onalkaline intrusions. To perform a three-dimensional inversion, itwas necessary to divide the GAP region in three areas (area 1, 2 and3 in Fig. 7) due to software limitations.

Dutra and Marangoni (2009) showed a forward model formagnetic and gravity data of Morro do Engenho and A2 anomalies.Their results show that simple geometrical forms, such as spheres,match datasets well for magnetic property sources smaller than themass source (readers are referred to Fig. 10 of Dutra and Marangoni(2009)).

5. 3D gravity inversion

The model configuration explains both gravity and magneticdata because the estimate based only on potential fields iscommonly affected by the non-uniqueness of the solution. Toimprove the estimate and reduce uncertainty we combine theavailable information in the area: gravity and magnetic data,measurements of density and magnetic properties, depth and

structural index estimates, locations of edges of bodies andgeological information.

In the case studied here, the inversion of gravity anomaliesconsists of using observations to generate a model that tries toreproduce the subsurface geological structures that match gravityobservations calibrated with geological and magnetic information.We use the 3D inversion technique developed by Li and Oldenburg(1996, 1998), GRAV3D (2002). The model consists of a grid of 3Dprisms with density contrasts to be determined. A smooth initialreference model is set up with density defined for each prism.

The gravity observations are approximated by a continuousfunction expressing the relationship between density and the cor-responding gravity observations. The Li and Oldenburg methoduses a linear formulation to the inverse problem, representing thesubstrate by a homogeneous grid of known size and position withunknown density contrast. A subsurface distribution is calculatedand the resulting anomaly is compared to the observed data.Changes can be made in the reference model, and a new inversioniteration is performed.

The subsurface was discretized in cells sized 2 km in the x, y andz directions for areas 1 and 2 (in Fig. 7), and with cells of 1 km in thex, y and z directions for area 3 (in Fig. 7). The imposed limits todensity contrast wereDrmin¼ 0.0 g/cm3 outside of the anomaly andDrMax ¼ 0.30 g/cm3 inside it. The maximum density contrast isbased on gabbro, basalt, diabase, and ultramafic rocks, which aretypical in the study area.

Li and Oldenburg (1998) minimize the moduli of the first-orderderivatives of the weighted density distribution along the hori-zontal and vertical directions to counteract the tendency of mass

Fig. 7. Bouguer residual anomaly map of the northern area after subtraction of a surface of the 4th order. The squares mark the study areas for inversions in Figs. 9e11. Whitenumbers indicate the alkaline intrusion locations. Red lines represent the mapped alkaline intrusions. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e5550

Author's personal copy

concentration at the borders of the interpretation region. Theweight function does not allow the assignment of large values tothe density distribution estimates between the Earth’s surface andthe source’s top. Thus, the efficacy of the Li and Oldenburg strategylies in a priori knowledge about the depth of the top of the source.The weight function is w(z) ¼ (z � z0)�b, where b is the rate ofchange with distance of the gravity field and z0 depends on thelength of cell discretization and the altitude of the observation data.We considered b equal to 3. The initial reference model combineinformation based on geology (Gomes et al., 1990; Danni,1994) thatsuggested that the intrusive bodies are plug-like intrusions and onour results from the magnetic analysis presented earlier and shownin Table 1.

These inversions have their origins (0,0) (Fig. 7) at the coordi-nates 16.2�S, 52�W (area 1), 16.6�S, 51.6�W (area 2) and 16�S,51.4�W (area 3). The initial reference model consists of compactbodies 4e10 km wide, ranging from 0 to 8 km depth, and centeredon Xc and Yc. The parameters are shown in Table 1.

Figs. 8e10 show depth sections with subsurface densitycontrasts obtained by the inversion procedure. In Fig. 8 (area 1), thedensity contrast of the surface rocks to those lying deeper than2 km ranges from 0.20 (g/cm3) to 0.30 (g/cm3). Morro do Engenho(#1 in Fig. 1) reaches a depth of 12 km, and Registro do Araguaia(#12 in Fig. 1) extends to 18 km depth. Only intrusions 1 and 13 inFig. 8 appear to be connected at some depth; the other bodiesappear separate. Magnetic data in the neighbor area (anomalies 1and 12 in Fig. 2) pointed out at a set of extensive linear magneticanomalies in the NEeSW direction that can be related to dikes,although we have not noticed any feature similar to that in the fieldbecause the region is covered with Araguaia River alluvium. We

also note that in Fig. 8, the presence of non-zero density contrasts atthe east and south borders are responses from neighboring bodieslocated in areas 2 and 3.

Bodies in Fig. 9 that represent area 2 show a density contrast of0.20 g/cm3 to a depth of 4 km, and 0.30 g/cm3 from 4 to 11 km; the

Fig. 8. Density contrast distribution in depth slices resulting from the inversion of theanomalies of area 1 shown in Fig. 7. Origin of the area is at 16.2�S and 52.0�W. Blacknumbers indicate the alkaline intrusion locations.

Fig. 9. Density contrast distribution in depth slices resulting from the inversion of theanomalies of area 2 shown in Fig. 7. Origin of the area is at 16.6�S and 51.6�W. Blacknumbers indicate the alkaline intrusion locations.

Fig. 10. Density contrast distribution in depth slices resulting from the inversion of theanomalies of area 3 shown in Fig. 7. Origin of the area is at 16.0�S and 51.4�W. Blacknumbers indicate the alkaline intrusion locations.

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e55 51

Author's personal copy

density contrast then diminishes again past 11 km. The mainfeatures reach a maximum depth of 12 km and are separate.Intrusions 5 and 7 in Fig. 9 are connected at depths greater than6 km. Fig. 10 represents area 3 with the Santa Fé Complex, whichdisplays a NeS major axis and EeW minor axis. The distributiondensity contrast of 0.25 (g/cm3) reaches a depth of 10 km. Most ofthe trends in density contrast have a N�S orientation at interme-diate depths.

The 3D inversion results show that the alkaline intrusions havedepths of 10e12 km, except for the Registro do Araguaia that rea-ches a depth of 18 km. In areas 1 and 2, DrMax was 0. 30 g/cm3; inarea 3, DrMax was 0.25 g/cm3. The density depth sections point tospheres and cylinders with comparable vertical and horizontalsizes as preferable geometries for the mass distribution. It is

evident that the centers of the larger bodies are located at inter-mediate depths of 4e6 km; shallower and deeper sections havesmaller diameters for the bodies. The exception is the Registro doAraguaia (#12 in Fig. 8). Density variation inside the models,including some increase in density contrast with depth, may berelated to fractional crystallization because most of the intrusionspresent a zoned distribution of magma products, with dunite in thecenter and less dense lithological types at the border (Radaelli,2000).

We calculated the gravity field by forward modeling using theinversion results, and compared it with the observed Bouguerresidual anomaly in Fig. 11 for each area. In these figures, thegravimetric anomaly fits the data amplitude and shape. Theamplitude differences at the right side of Fig. 11 are almost zero.

Fig. 11. Difference between observed and estimated residual Bouguer anomaly of area 1 (bottom), area 2 (middle), and area 3 (top). Red lines are the results from model inversionintrusions and black lines are from the residual map of Fig. 7. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of thisarticle.)

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e5552

Author's personal copy

6. Registro do Araguaia anomaly

The Registro do Araguaia anomaly (#12 in Fig. 1) is the largestgravity and magnetic anomaly in the area, although there are nooutcrops in the area or any geological feature indicating this

intrusion. It has a particular magnetic signature and followsaNNEeSSWdirection (Fig. 5a). TheRegistro doAraguaia hasAN-EULmethod solutions with a top depth of 600 me800 m. The lines thatoccur north of the anomaly have structural indices ranging from0 to2 and are highly aligned, suggesting contacts or dikes.

Fig. 12. Cross-sections of the magnetic susceptibility contrast distribution for the Registro do Araguaia anomaly. Profile AB was used to visualize results of the 3D inversion ofmagnetic data in Fig. 13.

Fig. 13. Cross-sections of (a) magnetic susceptibility contrast distribution (Dkmin¼ 0.0 SI, DkMax¼ 0.08 SI) and (b) density contrast distribution (Drmin ¼ 0.0 g/cm3, DrMax¼ 0.3 g/cm3)obtained from 3D inversion. Location of the profile AB is shown in Fig. 12.

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e55 53

Author's personal copy

To observe how the magnetic material is distributed in theNNEeSSW belt, we calculated the magnetic susceptibility contrast(Dk) in depth. We applied the technique developed by Ku and Sharp(1983) to calculate Dk from the total-field magnetic anomalyproduced by a thin two-dimensional vertical dike, according to thestructural indices shown in Fig. 5a. We used the data from the flightline after corrections for diurnal variation and subtraction of IGRF.In this inversion, shown in Fig. 12, only the geometrical shape is set;we do not use any other constraint. This can result in a differentresponse than that from 3D inversion, this different response canbe verified when the results in Fig. 12 are compared with the 3Dinversion cross sections in Fig.13a. The different results are partiallydue to the non-uniqueness of the potential field and from thedifferent datasets; the data from flight lines are very close (10 m),while the cell size is 2 km for the 3D inversion.

Each section in Fig. 12 represents the distribution of magneticsusceptibility contrast at the flight line. The high Dk values arerecorded in intrusions 1 and 12 varying from 0.03 to 0.06 SI. TheNNE-SSW belt has a magnetic susceptibility distribution with thesame high value. The Dk values extend to 10 km depth but canpossibly reach greater depths. It is possible that the 2D inversion isbetter at recovering thin lineaments. Their positions can be tracedalong the flight lines, but is almost impossible to trace them in the3D inversion.

We also estimate the magnetic susceptibility contrast of theRegistro do Araguaia through the 3D magnetic anomaly inversiondeveloped by Li and Oldenburg (1996). The limits of the magneticsusceptibility contrast were Dkmin ¼ 0.0 SI outside of the anomalyand DkMax ¼ 0.08 SI inside it.

Fig. 13a and b represents the magnetic susceptibility and densitycontrast from the inversion results for profile AB (marked inFig. 12). Magnetic mineralogy and mass reaches depths of 18 km.Magnetic susceptibility is more concentrated, such as the resultsfrom Morro do Engenho discussed in Dutra and Marangoni (2009).This may suggest that the Registro do Araguaia is an axial zonedbody, as are some of the outcrops in the GAP.

7. Conclusions

The map of analytical signal amplitude calculated using themagnetic data reduced to the pole (Fig. 3a), characterized by lowmagnetic amplitude and moderate magnetic relief, shows a morehomogeneous signal in the study area. The NNW lineaments areexamples of such low-amplitude linear features. The larger distur-bances inmagnetic field are due to the presence of dike swarms andsmall magnetic bodies positioned in the N�S and NeE direction. Adetailed analysis of the analytical signal amplitude shows SWeNEand SEeNW lineaments above 10 km depth. In the areas covered byParaná Basin units, there is a decrease in lineament density. This canbe seen in the attenuation of high frequencies, which characterizethe lineaments. Borehole information about 100 km south of thearea, inside the Paraná Basin, shows at least 2000 m of sediments(Costa, 2006). At the basin border in the GAP area, we expectsediment coverage less than that observed in the boreholes to thesouth, but sediment coverage of at least 400 m could also smooththe signal of small dikes and fractures.

The study area has lineaments defined by an arrangement offaults and fractures filled by extensive mafic-ultramafic dikes thatare predominantly SEeNW and secondarily SWeNE. This is theregion where the regional lineaments of the Bom Jardim Arch andTransbrasiliano Lineament cross each other. These are crustalweakness zones that were reactivated in the Cretaceous by thealkaline magmatism.

The SEeNW structures are frequent and well characterized bythe analytical signal amplitude; it seems that these are the shallow

expression of the emplacement path. The SEeNW lines are asso-ciated with the general alignment of the magma intrusions atemplacement levels linked to the main directions of extension ofthe Bom Jardim Arch. This structure opened enough space for thevoluminous magma of the intrusions along the Paraná basin border(anomalies 12, 3, 4, 5, 6, 7, 8, 9). The structures related to theSWeNE lineament control the location of the Registro do Araguaiaand Morro do Engenho complexes (anomalies 1, 12 and 13 inFig. 5a).

The GAP intrusions have roots at depths ranging from 10 to12 km with density contrasts of 0.25e0.3 g/cm3, except for theRegistro do Araguaia, which reaches a depth of 18 km witha density contrast of 0.3 g/cm3. Themagnetic susceptibility contrastranges from 0.06 to 0.08 SI. The intrusions located at the northernboundary of the GAP are larger, and those to the south are smallerand are aligned with a system of faults and dikes. The Registro doAraguaia anomaly (#12) is at the center of the main area of alkalinecomplexes. It is deeper than the other alkaline intrusions and dipsto the east according to the density contrast model (Figs. 8 and 13).

The calculated density contrast is consistent with chambersfilled mainly with silicate magmas such as peridotite, pyroxeniteand gabbro that may have been subject to some degree of fractionalcrystallization that could lead to zoned intrusions. The magmaticsusceptibility data also corroborate a relatively narrow composi-tional range in terms of iron content. Indeed, the exhumed plutonicGAP complexes are dominated by silicates. Another interestingcharacteristic shown by the Registro do Araguaia anomaly is thecomplexity of the intrusion, which may represent a multistageemplacement process.

According to our 3D inversion results, we interpret the densitycontrast distribution as intrusions located within the upper crust,with shapes varying from spherical (intrusions 1, 2 and 3) tocylindrical (body 5) with intermediate forms of a cone sheet for theRegistro do Araguaia, whose top lies at a depth of 2 km, anda possible dike (body 13). Thus, the intrusions in this area lookdifferent from other alkaline intrusion studies discussed in ourintroduction. Some of those studies (Bott and Tantrigoda, 1987;Arzamastsev et al., 2000; Rugenski et al., 2001; Chandrasekharet al., 2002) modeled the alkaline complexes as relatively narrowvertical cylinders at greater depths, and others as sub-verticalconcentric bodies widening downwards. Spheres and cylinderswith comparable lateral and vertical dimensions were found, inagreement with the field exposures and the idea that at least somepart of these intrusions were emplaced as crystal mushes, assem-bling as sphere-like bodies at the Precambrian/Phanerozoicunconformity. As alkaline magmatism involves small volumes ateach pulse, it is more difficult to connect the bodies, although thereare significant numbers of them distributed in the area.

From north to south in the GAP, there is a considerable change interms of stratigraphic level where alkaline rocks are exposed.Junqueira-Brod et al. (2005) proposed a model for the magmachamber positions in the GAP where plutonic bodies were intrudedalong the contact between Precambrian basement and the Paranásedimentary basin, usually in direct contact with granitic rocks andthe top limited by sedimentary rocks of the Devonian FurnasFormation. This unconformity allowed enough space to accom-modate relatively large amounts of magma. The lineamentsoriented NWeSE and NEeSW seem to be the preferred site of LateCretaceous alkaline magmatism. The dominant guidelines correlatestronglywith the extensional regimes that contributed to the rise ofalkaline magmatism. The region was uplifted during or after theLate Cretaceous, and the mafic intrusions were exposed (as inMorro do Engenho, Santa Fé, Montes Claros de Goiás, Buriti, Cór-rego dos Bois, Morro do Macaco, and Arenópolis) or remained closeto the surface (such as A2 and Registro do Araguaia).

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e5554

Author's personal copy

Our results suggest that intrusion geometry may be betterexplained as magmatic chambers emplaced in the upper crust,some of them eroded to the top and currently exposed ascomplexes. We inferred that the magma used zones of weaknessrelated to the Transbrasiliano Lineament, an ancient suture in thecrust that marks the collision of the Amazon and São Francisco-Congo-Kalahari cratons. The NEeSW border of the Paraná basin isalso an important zone of magma emplacement in the central GAP.The suture of the São Francisco craton and basement block beneaththe Paraná Basin was also a weak zone that allowed emplacementof kamafugite flows in the southern part of the GAP. This suture ismarked on the Bouguer anomaly map (Fig. 6) by the positive-negative anomaly, aligned along a NEeSW direction.

Acknowledgments

This work was supported by São Paulo Scientific Foundation(FAPESP) grants 2006/00201-2 and 2007/53179-7. The authorsacknowledge Clarino Vieira for field support and Nelsi de Sá forfinal GPS processing. We are thankful for the reviewers’ comments.

References

Arens, T.J., Johnson, M.L., 1995. Shock wave data for rocks. In: Rock Physics & Phaserelations: a handbook of physical constants. AGU.

Arzamastsev, A.A., Glaznev, V.N., Raevsky, A.B., Arzamastseva, L.V., 2000.Morphology and internal structure of the Kola alkaline intrusions, in Fenno-scandian Shield: 3D density modeling and geological implications. Journal ofAsian Earth Sciences 18, 213e228.

Assumpção, M., Schimmel, M., Escalante, C., Barbosa, J.R., Rocha, M., Barros, L.V.,2004. Intraplate seismicity in SE Brazil: stress concentration in lithospheric thinspots. Geophysical Journal International 159 (1), 390e399.

Beltrão, J., Silva, J., Costa, J., 1991. Robust polinomial fitting method for regionalgravity estimation. Geophysics 56 (1), 80e89.

Blakely, R.S., 1996. Potential Theory in Gravity and Magnetic Applications. Cam-bridge University Press, USA.

Bott, M.H.P., Tantrigoda, D.A., 1987. Interpretation of the gravity and magneticanomalies over the Mull Tertiary intrusive complex, NW Scotland. Journal of theGeological Society 144, 17e28.

Brod, J.A., Barbosa, E.S.R., Junqueira-Brod, T.C., Gaspar, J.C., Diniz-Pinto, H.S.,Sgarbi, P.B.A., Petrinovic, I.A., 2005. In: Comim-Chiaramont, P., Gomes, C.B.(Eds.), Mesozoic to Cenozoic alkaline magmatism in the Brazilian Platform.Editora da Universidade de São Paulo, Fundação de Amparo à Pesquisa doEstado de São Paulo, São Paulo, pp. 261e316.

Carlson, R.W., Araujo, A.L.N., Junqueira-Brod, T.C., Gaspar, J.C., Brod, J.A.,Petrinovic, I.A., Hollanda,M.H.B.M., Pimentel, M.M., Sichel, S., 2007. Chemical andisotopic relationships between peridotite xenoliths and maficeultrapotassicrocks from southern Brazil. Chemical Geology 242, 415e434.

Chandrasekhar, D.V., Mishra, D.C., Poornachandra Rao, R.G.V.S., MallikharjunaRao, J., 2002. Gravity and magnetic signatures of volcanic plugs related toDeccan volcanism in Saurashtra, India and their physical and geochemicalproperties. Earth and Planetary Science Letters 201, 277e292.

Costa, T. N., 2006. Estudo de espessura sedimentar na Bacia do Paraná com funçãodo receptor de alta frequência. Dissertação de mestrado no Instituto deAstronomia, Geofísica e Ciências Atmosféricas. Universidade de São Paulo, SãoPaulo.

Danni, J.C.M., 1994. Os picritos alcalinos da região de Iporá: implicações na gênesedos complexos tipo central do sul de Goiás. Revista Brasileira de Geociências 24(2), 112e119.

Dindi, E.W., Swain, C.J., 1988. Joint three-dimensional inversion of gravity andmagnetic data from Jombo Hill alkaline complex, Kenya. Journal of theGeological Society 145, 493e504.

Dutra, A.C., Marangoni, Y.R., 2009. Gravity and magnetic 3-D inversion of Morro doEngenho complex, central Brazil. Journal of South American Earth Sciences 28,193e203.

GRAV3D, 2002. A Program Library for Forward Modeling and Inversion of GravityData Over 3D Structures, Version 2.0.

Gibson, S.A., Thompson, R.N., Leonardos, O.H., Dickin, A.P., Mitchell, J.G., 1995. TheLate Creatceous impact of the Trindade mantle plume: evidence from large-volume, mafic, potassic magmatism in SE Brazil. Journal of Petrology 36 (1),189e229.

Gomes, C.B., Ruberti, E., Morbidelli, L., 1990. Carbonatite complexes from Brazil:a review. Journal of South American Earth Sciences 3 (1), 51e63.

Junqueira-Brod, T.C., Gaspar, J.C., Brod, J.A., Jost, H., Barbosa, E.S.R., Kafino, C.V., 2005.Emplacement of kamafugite lavas from the Goiás Alkaline Province, Brazil:constrains from whole-rock simulations. Journal of South American EarthSciences 18, 323e335.

Junquiera-Brod, T.C., Roig, H.L., Gaspar, J.C., Brod, J.A., Meneses, P.R., 2002.A Província Alcalina de Goiás e a extensão de seu vulcanismo kamafugítico.Revista Brasileira de Geociências 32 (4), 559e566.

Ku, C.C., Sharp, J.A., 1983. Werner deconvolution for automated magnetic inter-pretation and its refinement using Maquardt’s inverse modeling. Geophysics48, 754e774.

LASA Engenharia e Prospecções S.A., 2004. “Projeto Levantamento Aerogeofísico doEstado de Goiás e 1aEtapa e Arco Magmático de Arenópolis e ComplexoAnápolis - Itauçu e Seqüência Vulcano-Sedimentar de Juscelândia. RelatórioFinal do Levantamento e Processamento dos Dados Magnetométricos eGamaespectrométricos”, Convênio de Cooperação Técnica entre a SGM/MME/CPRM e SIC/SGM/FUNMINERAL/Estado de Goiás, Relatório Final, vol. 22 Texto eAnexos (mapas), Rio de Janeiro.

Li, Y., Oldenburg, D.W., 1996. 3D inversion of magnetic data. Geophysics 61,394e408.

Li, Y., Oldenburg, D.W., 1998. 3D inversion of gravity data. Geophysics 63, 109e119.Molina, E.C., Ussami, N., 1999. The geoid in southeastern Brazil and adjacent

regions: new constraints on density distribution and thermal state of thelithosphere. Journal of Geodynamics 28, 357e374.

Nabighian, M., 1972. The analytic signal of two dimensional magnetic bodies withpolygonal cross-section: it’s properties and use for automated anomaly inter-pretation. Geophysics 37, 507e517.

Pimentel, M.M., Fuck, R.A., Gioia, S.M.C.L., 2000. The Neoproterozoic GoiásMagmatic Arc, central Brazil: a review and new SM-Nd isotopic data. RevistaBrasileira de Geociências 30 (1), 35e39.

Radaelli, V.A., 2000. Programa levantamentos geológicos básicos do Brasil. Geologiae Recursos minerais do Estado de Goiás e do Distrito Federal. Informe derecursos minerais, Níquel do Morro do Engenho - Estado de Goiás. Companhiade Pesquisa de Recusos Minerais, Goiânia, 10 pp.

Reid, A.B., 1980. Aeromagnetic survey design. Geophysics 45 (5), 973e976.Rugenski, A., Mantovani, M.S.M., Shukowsky, W., 2001. Modelo gravimétricos do

Complexo Alcalino de Pariquera-Açu (SP). Revista Brasileira de Geociências 31(3), 365e374.

Salem, A., Ravat, D., 2003. A combined analytic signal and Euler method (AN� Euler)for automatic interpretation of magnetic data. Geophysics 68, 1952e1961.

Sonoki, I.K., Garda, G.M., 1988. Idades K-Ar de rochas alcalinas do Brasil Meridional eParaguai Oriental: compilação e adaptação às novas constantes de decaimento.In: Boletim IG-USP: Série Científica, vol. 19, pp. 63e85.

Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990. Applied Geophysics, second ed.Cambridge University Press, New York.

Thompson, D.T., 1982. Euldph: a new technique for making computer-assisteddepth estimates from magnetic data. Geophysics 47, 31e37.

Valeriano, C.M., Pimentel, M.M., Heilbron, M., Almeida, J.C.H., Trouw, R.A.J., 2008.Tectonic Evolution of the Brasília Belt, Central Brazil, and Early Assembly ofGondwana. In: Geological Soc. London, Special Publication, vol. 294, pp.197e210.

van Decar, J.C., James, D.E., Assumpção, M., 1995. Seismic evidence for a fossilmantle plume beneath South America and implications for plate driving forces.Nature 378, 25e31.

A.C. Dutra et al. / Journal of South American Earth Sciences 33 (2012) 43e55 55