Contents lists available at SciVerse ScienceDirect
Tectonophysics
j ourna l homepage: www.e lsev ie r .com/ locate / tecto
Review Article
Models of crustal thickness for South America from seismic refraction, receiverfunctions and surface wave tomography
Marcelo Assumpção a,⁎, Mei Feng b, Andrés Tassara c, Jordi Julià d
a Institute of Astronomy, Geophysics and Atmospheric Sciences, University of São Paulo, Rua do Matão 1226, 05508-090, São Paulo, SP, Brazilb Institute of Geomechanics, Chinese Academy of Geological Sciences, Beijing, 100081, Chinac Department of Earth Sciences, Universidad de Concepción, Victor Lamas 1290, Concepción, Chiled Departamento de Geofísica and Programa de Pós-Graduação em Geodinâmica e Geofísica, Universidade Federal do Rio Grande do Norte, Natal, 59078-970 RN, Brazil
Please cite this article as: Assumpção, M., et aface wave tomography, Tectonophysics (201
a b s t r a c t
a r t i c l e i n f o
Article history:Received 22 April 2012Received in revised form 5 November 2012Accepted 15 November 2012Available online xxxx
Keywords:CrustMohoAndesTomography
An extensive compilation of crustal thicknesses is used to develop crustal models in continental SouthAmerica. We consider point crustal thicknesses from seismic refraction experiments, receiver function anal-yses, and surface-wave dispersion. Estimates of crustal thickness derived from gravity anomalies were onlyincluded along the continental shelf and in some areas of the Andes to fill large gaps in seismic coverage.Two crustal models were developed: A) by simple interpolation of the point estimates, and B) our preferredmodel, based on the same point estimates, interpolated with surface-wave tomography. Despite gaps in con-tinental coverage, both models reveal interesting crustal thickness variations. In the Andean range, the crustreaches 75 km in Southern Peru and the Bolivian Altiplano, while crustal thicknesses seem to be close to theglobal continental average (~40 km) in Ecuador and southern Colombia (despite high elevations), and alongthe southern Andes of Chile–Argentina (elevation lower than 2000 m). In the stable continental platform theaverage thickness is 38±5 km (1-st. deviation) and no systematic differences are observed among Archean–Paleoproterozoic cratons, NeoProterozoic fold belts, and low-altitude intracratonic sedimentary basins. Anexception is the Borborema Province (NE Brazil) with crust ~30–35 km thick. Narrow belts surroundingthe cratons are suggested in central Brazil, parallel to the eastern and southern border of the Amazon craton,and possibly along the TransBrasiliano Lineament continuing into the Chaco basin, where crust thinner than35 km is observed. In the sub-Andean region, between the mid-plate cratons and the Andean cordillera, thecrust tends to be thinner (~35 km) than the average crust in the stable platform, a feature possibly inheritedfrom the old pre-Cambrian history of the continent. We expect that these crustal models will be useful forstudies of isostasy, dynamic topography, and crustal evolution of the continent.
l., Models of crustal thickness for South America from seismic refraction, receiver functions and sur-3), http://dx.doi.org/10.1016/j.tecto.2012.11.014
Mapping variations of crustal thickness in the continents havemanyimportant applications for the study of the continental crust and litho-sphere. Besides giving information on the crustal evolution, degree ofisostatic compensation (e.g., Sacek and Ussami, 2009), and intraplatestress patterns (e.g., Lithgow-Bertelloni and Guynn, 2004), crustalthickness estimates are essential for modeling wave-propagation inglobal and regional seismic studies (e.g., Hjörleifsdóttir and Ekstrom,2010; Lebedev and van der Hilst, 2008), monitoring regional-scaleseismicity (e.g. Ferreira et al., 2008), and for source discrimination inthe framework of the Comprehensive Nuclear Test-Ban Treaty(e.g. Cormier and Anderson, 2004; Fan and Lay, 1998). In addition,models of crustal thickness variation can serve for developing sur-face corrections to investigate upper mantle structure through ei-ther body-wave or surface-wave tomography studies (e.g. Bastowet al., 2008; Park et al., 2008) and/or through the study of theEarth's normal-modes of vibration (e.g., Justowski et al., 2007; Mooneyand Kaban, 2010). Also, the increasing use of shorter wavelengths inglobal seismic modeling requires correspondingly more accuratemodels of crustal thickness variation (e.g., Fichtner and Igel,2008; Fichtner et al., 2009).
In spite of its importance, crustal thickness is still among the leastknown crustal properties of South America. Large areas of the conti-nent, such as the Amazon craton and the Chaco basin in NE Argentina,are sparsely sampled and detailed information on crustal structure islacking (e.g. van der Lee et al., 2001). The best known area is theAndean region, for which a number of passive and active seismic ex-periments have been carried out (e.g. ANCORP, 2003; Gans et al.,2011; McGlashan et al., 2008; Yuan et al., 2002) and detailed modelsof crustal structure have been developed by combining seismic dataand gravity modeling in Venezuela (Niu et al., 2007) and South andCentral Andes (Tassara and Echaurren, 2012). In addition, severaltemporary seismic experiments in NW Argentina and Chile havealso mapped crustal structure over the region of flat-slab subduction(e.g. Calkins et al., 2006; Gans et al., 2011) and the Bolivian altiplano(e.g. Beck and Zandt, 2002; Beck et al., 1996; Swenson et al., 2000). Inthe tectonically stable part of the continent, the most comprehensivecrustal thickness maps were produced by Feng et al. (2007) and Lloydet al. (2010), through constrained tomographic inversion of surface-wave data. Those studies used both average epicenter-station 1Dmodels and group velocities and incorporated local constraints oncrustal thickness from independent receiver function studies and seis-mic refraction profiles. Some additional information based on isostaticassumptions was also included. Nonetheless, in spite of fitting seismicpoint constraints with a root mean square (RMS) deviation of about3–4 km, these crustal thickness models still had errors around 10 kmin areas without point constraints.
We have built on a previous compilation of seismic pointconstraints for crustal thickness in the Brazilian shield and adjacentregions (Assumpção et al., in press) to produce an enlarged compila-tion of 920 point constraints, 730 onshore and 190 offshore, for thewhole of South America. To our knowledge, this is the most compre-hensive compilation of seismic point constraints on crustal thick-nesses for the continent. Previous continental-scale compilations ofcrustal thickness were performed during the development of globalcrustal thickness models, such as CRUST2.0 (Bassin et al., 2000),CRUST5.1 (Mooney et al., 1998), or the global compilation of Solleret al. (1982), but those studies had very sparse coverage in South
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
America and/or focused on a single data type. Models CRUST5.1 andCRUST2.0, for instance, were based entirely on crustal thickness esti-mates from active-source profiling and ignored constraints on crustalthickness from passive-source studies. Soller et al. (1982) used bothactive-source profiling and surface-wave studies, but included just afew studies in the Andean region for South America leaving crustalthickness estimates for the majority of the continent as mere extrap-olations (Tanimoto, 1995). Our dataset largely improves and updatesthe constraints provided in those earlier compilations.
Comparing the newly compiled set of seismic crustal thicknesswith the Bouguer Anomaly we derived an empirical relationshipthat is then used for predicting crustal thickness in areas where noseismic data is available (such as parts of the northern Andes andthe continental shelf). We developed two types of models of crustalthickness variation for South America. The first type consists of aninterpolation of the seismically-constrained and gravity-predictedcrustal thicknesses, while the second type consists of an interpolationbased on surface-wave studies. As expected, the addition of newpoint constraints improves the resolution of crustal thickness esti-mates in previously unsampled areas of the continent, such as North-ern Andes (Ecuador, Colombia and Venezuela) and NortheasternBrazil (Borborema Province and northern part of the São Franciscocraton) and the southern part of the Paraná basin. However, largeportions of stable South America remain poorly sampled and predic-tions from our models within those regions could have significant er-rors (around 5–10 km). We hope that this study will motivate thedeployment of temporary broadband experiments to fill in the gapsin our knowledge. Despite the largely unsampled areas, we showthat derived maps of crustal thickness correlate with first order tec-tonic features of the continent and give new insights into the old geo-logical control on the present-day crustal structure.
Finally, the point constraints compiled in this study have been uti-lized in the validation of an independent model of crustal thicknessvariation for South America based on satellite gravity (van derMeijde et al., submitted for publication).
2. Compilation of published crustal thickness data
Weexpanded the recent compilation of seismic crustal thicknesses forthe Brazilian shield and adjacent regions (Assumpção et al., in press),which included 229 previous point constraints from Feng et al. (2007),244 from Lloyd et al. (2010), 183 from Tassara and Echaurren (2012),and 200 fromPavão et al. (2012), to a total of 920 point constraints. Addi-tional point constraintswere incorporated for Venezuela (Niu et al., 2007;Schmitz et al., 2005), Central and Northern Andes (Dorbath et al., 1993;Robalino, 1977 (apud Feininger and Seguin, 1983; Ocola et al., 1975);Pacific offshore margin (Agudelo et al., 2009; Hussong et al., 1976;Meyer et al., 1976); Western Argentina (Gans et al., 2011), NE Argentina(Rosa et al., 2010), South Central Peru (Phillips et al., 2012), southernPuna (Bianchi et al., in press), and Patagonia (Lawrence and Wiens,2004). Point constraints developed in Assumpção et al. (in press) fromreceiver functions for 19 newly installed seismic stations in Brazil werealso included.
The point constraints come from two different seismic data types:active source experiments (deep seismic refraction lines, or deep seis-mic reflection surveys) and receiver functions (see Fig. 1). Logically,point constraints from active-source profiling required sampling theseismic line at regular intervals. For 2D seismic refraction models,points were selected at every 50 km, on average; for old, 1D models,
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
3M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
only a few points were selected with larger spacing. Receiver func-tions, on the other hand, are sensitive to the structure around the re-cording station and naturally provide point constraints (e.g. Ammon,1991). Also, in the offshore areas, the seismic constraints werecomplemented with published crustal thickness estimates based onBouguer gravity anomalies constrained by seismic data, such asMohriak et al. (2000) and Zalán et al. (2011) for the Atlantic marginand Schmitz et al. (2005) for Venezuela. More details of the compila-tion can be found in Assumpção et al. (in press).
Our compilation also includes uncertainties for each point con-straint. If uncertainties were not provided by the source, a rough esti-mate was obtained based on the quality of the processed data. Forexample, for recent seismic refraction experiments interpreted withray tracing techniques in 2D models (such as Schmitz et al., 2005;Soares et al., 2006), we assigned an uncertainty of 2 km for crustalthicknesses. For old, unreversed seismic experiments interpreted with1D models, such as the Nariño Project in the Ecuadorean/Colombian
Fig. 1. Distribution of data points from seismic refraction/reflection profiles and stationswith Relogical provinces are indicated as follows: GS and CBS=Guyana and Central Brazil shields of theand Mantiqueira NeoProterozoic foldbelt provinces, respectively; intracratonic basins: Am=SOr=Oriente Basin; PTG=Patagonia province (northern limit from Ramos, 2004). Blue solid lin
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
Andes (Mooney et al., 1979; Ocola et al., 1975), the assigned uncertaintycould reach 10 km. For receiver function point constraints based on thehk-stacking technique of Zhu and Kanamori (2000), uncertainties incrustal thickness were generally reported during the analysis; for re-ceiver function point constraints involving modeling of the waveforms,confidence bounds equal to ±one layer thickness were assigned to theestimates (usually ±2 km).
It is important to realize that, for the same point, several estimatesof crustal thickness might be available. For instance, analysis of re-ceiver functions at some stations was performed by several authorsutilizing different techniques or, sometimes, even the same tech-nique. Not surprisingly, a range of estimates for crustal thickness isavailable at those stations and the estimates vary with the choice offilter bandwidth, event selection, and assumed parameters duringthe data processing. Assumpção et al. (in press) compiled all availableestimates of crustal thickness for the Brazilian shield and adjacentareas and calculated the standard deviations (σH) of the mean crustal
ceiver Function. Color scale indicates crustal thickness (from surface to Moho). Major geo-Amazon craton; SFC=São Francisco craton; BB, TT andMQ are the Borborema, Tocantins
olimões and Amazon, Pn=Parnaíba, Pc=Parecis, Pt=Pantanal, Pr=Paraná, Ch=Chaco;e is the 3000 m altitude.
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
Fig. 3. Distribution of the 920 compiled crustal thicknesses for the whole area(RF+refraction/reflection data), in both continental and oceanic areas. Dark shadeshows data with uncertainty ≤3 km; light shade denotes data with larger uncer-tainties. The three groups reflect the preferential distribution of the experiments inthe compiled literature.
4 M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
thickness (H) taking into account the uncertainties of each individualmeasurement. Fig. 2 shows the distribution of the standard deviations(σH) of the crustal thickness (H). Most measurements agree to within±2 km, and scatter (σH) less than 1 km is rare. In the final dataset, aminimum uncertainty of 1 km was assigned to all stations with onlyone reported value.
It has been argued that refraction data and receiver functions maynot sample exactly the same Moho depth (e.g., Stratford et al., 2009).The high frequency (5 to 10 Hz) Pn head-waves, used in seismic re-fraction studies, sample the topmost upper mantle. Lower frequencyreceiver functions (around ~1 Hz), in areas of gradational Moho,may sample roughly the middle of the Moho transition zone, therebygiving slightly lower values for the crustal thickness. However, withinthe uncertainties of our compiled data (roughly 2 to 3 km for both re-ceiver functions and seismic refraction) the two types of data shouldbe consistent. In addition, most of the onshore seismic refraction linesare close to seismic stations with receiver function data; the 1 degreespatial resolution of our model smears out possible systematic differ-ences between the two types of data.
All compiled point constraints in South America with crustal thick-nesses derived from seismic methods (both deep seismic refraction/reflection experiments and receiver functions) are shown in themap of Fig. 1, and a histogram of crustal thicknesses for the completedataset is shown in Fig. 3. Thicknesses less than 20 km are mostlyfrom seismic refraction lines in oceanic areas and have smaller uncer-tainties. Thicknesses larger than 55 km refer to the Andean range. Themean crustal thickness in the stable part of the continent for the com-piled data is 38 km with a standard deviation of 4.8 km (Assumpçãoet al., in press). We should note that this is not the mean thickness ofthe stable continent because the sampling is not geographically uniformthroughout the continental interior.
3. Assessing gravity-based and surface-wave based models ofcrustal thickness
To construct models of crustal thickness for South America usingthe point constraints displayed in Fig. 1, we need to interpolate thedataset within poorly sampled areas of the continent. To accomplishthis goal we explore two possibilities: 1) using crustal models derivedfrom gravity anomalies, and 2) using surface wave tomography. Bothapproaches, however, have advantages and disadvantages. Many
0
5
10
15
20
25
30
0 1 2 3 4 5 6 7 8
st.dev.(km)
Fig. 2. Distribution of standard deviations of crustal thickness for all stations with threeor more different estimates (i.e., different authors or different techniques).From Assumpção et al. (in press)
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
gravity-derived crustal models, for instance, assume that the anoma-lies are due mainly to Moho topography with a constant densitycontrast between lower crust and upper mantle, which may not beaccurate enough, especially in the stable part of the continent.Surface-wave tomography, on the other hand, does not depend on as-sumptions about isostatic mechanisms, but is affected by uneven dis-tribution of epicenter-station paths. In addition, there is a trade-offbetween crustal thickness and upper mantle velocity, which usuallyprevents a resolution better than about 5 km or so when invertingfor crustal thickness. Before interpolating our point constraints,therefore, we first make an assessment of previously gravity-basedand dispersion velocity-based models of crustal thickness for SouthAmerica.
We start by assessing the accuracy of two previous surface-wavebased crustal thickness maps for South America (Feng et al., 2007;Lloyd et al., 2010). Both maps were obtained by joint inversion ofsurface-wave group velocities, waveform modeling and sparse pointconstraints on crustal thickness. Both models fit the crustal thicknessdata used in their joint inversion with an rmsmisfit of about 3 km. Toassess the real accuracy of those models, we use 56 of our newlycompiled point estimates of crustal thicknesses (not utilized in theirsurface-wave studies) as test points for the surface-wave basedmodels. Fig. 4 shows the twomodels and the 56 test points, all locatedmore than 100 km away from any of the point constraints of theirjoint inversion. In the Andean region, Fig. 4 shows that the model ofFeng et al. (2007) fits 17 test points with an rms residual of 9.4 km,while the model of Lloyd et al. (2010) fits the test points with anrms residual of 10.5 km. In the stable part of the continent, Fig. 4shows that 39 test points have an rms misfit of 4.7 and 5.8 km, re-spectively. Recall that for the stable part of the continent (Fig. 1),the compiled data for crustal thickness had an average value of~38 km and a standard deviation of 4.9 km. This means that neitherFeng et al. (2007) nor Lloyd et al. (2010) predict Moho depths inthe unsampled regions much better than the expected variability.That is, a model with a constant thickness of 38 km in the unsampledareas would fit the data almost as well as the models obtained with
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
Fig. 4. (A,B) Misfits of previous models of Moho depth (top: Feng et al., 2007; bottom: Lloyd et al., 2010) to 56 new point constraints far from the ones used to derive the models.Red circles mean the new measurements indicate a Moho 4 km or more deeper than the model; blue circles indicate the Moho is more than 4 km shallower than the model.
5M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
the tomographic inversions. This clearly shows the need for an im-provement in the model derived from the surface wave tomography.
Next, we assess the performance of a gravity-based model for theAndean region. Tassara and Echaurren (2012) developed a detailed 3Dmodel of Central and Southern Andes by forward modeling of gravitydata constrained by 183 seismically derived Moho estimates (from re-ceiver functions and active experiments) and other seismic and thermalinformation to fix the subducted slab and lithosphere–astenosphereboundary. Their model fits these 183 points with an rms misfit of5.8 km. This misfit is due to scatter in the compiled point constraints(uncertainties in crustal thickness estimates from receiver functions)but also due to smoothing of themodel in regionswith strong lateral var-iations. Fig. 5 shows 44 newly compiled points, which were not used intheir gravity modeling. These points have an rms misfit of 5.9 km, verysimilar to themisfit of the actual data.With exception of the northeasternarea (Western Amazon, Brazil, Fig. 5) themodel of Tassara and Echaurren(2012) seems to make a useful prediction of Moho depths for thesub-Andean area, with an expected error of about 6 km.
Given the result of the assessments, we have opted for developingtwo crustal models, both including gravity-derived estimates in theCentral and Southern Andes (model of Tassara and Echaurren, 2012)and in northern Andes (empirically estimated from Bouguer anomalies)to fill large gaps in seismic data: model A — using the seismic pointconstraints shown in Fig. 1, with simple interpolation in the stable partof the continent, and model B — using seismic point constraints, andsurface-wave tomography.
4. Crustal model A — point constraints and gravity-based studies
4.1. Data and method
In this model, we used the crustal thicknesses from the 762 seis-mic constraints (Fig. 1) complemented with the gravity-based crustal
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
model of Tassara and Echaurren (2012) for Central and SouthernAndes to guide the development of our crustal thicknesses in areasof poor seismic coverage, such as the sub-Andean region. This gravitymodel has errors larger than ±8 km near the Amazon, far from theAndean seismic constraints (see Fig. 5). For this reason, points inthe Amazon too far from the Andes were excluded from our model A.
In the northern Andes, where not much data was found in the liter-ature, we use point constraints derived from the following relationship
whichwas developed byfitting those point constraints fromour compi-lation (Fig. 1) that are located in the Andes and the adjacent Pacificmar-gin to Bouguer anomalies taken from the continental-scale map of Sá(2004) (see Fig. 6). The relationship fits the point constraints with anrms residual of 7.4 km. This means that crustal thicknesses estimatedfrom gravity data should have an expected standard error of about7 km. This is less than the rms misfit of the tomographic models ofFeng et al. (2007) and Lloyd et al.(2010) in the Andes (9.4 and10.5 km, respectively, as seen in Fig. 4). Given that crustal thicknessesacross the Pacific margin and the Andean cordillera vary widely (from10 to 20 km in the ocean, to more than 60 km in the high Andeanrange, and back to ~35–45 km in the sub-Andean basins), we thinkthat this empirical relation is useful to fill large gaps in seismic data cov-erage, as in the Northern Andes. For the Atlantic margin, Assumpção etal. (in press) derived the following quadratical empirical relation usingonly data in the oceanic areas:
Fig. 5. Moho depths from gravity model of Tassara and Echaurren (2012). Open circlesare the seismic constraints used in the gravity modeling. Larger circles are new seismicpoints not used in the gravity modeling. Grey circles denote misfits less than ±4 km;red circles denote observations are 4–8 km deeper than the model; blue circles denoteMoho depths 4–8 km shallower than the model; larger circles denote misfits largerthan ±8 km, such as in the northern area. The data used in the modeling (open circles)have an rms misfit of 5.8 km. The rms misfit of the 49 new data points is also 5.8 km.
−85°−80°−75°−70°−65°−60°−55°−50°−45°−40°−35°
−40°
−35°
−30°
−25°
−20°
−15°
−10°
−5°
0°
5°
10°
10
20
30
40
50
60
70
80
Cru
stal
thic
knes
s, k
m
−400 −200 0 200
Bouguer, mGal
Pacific Margin + Andes
Fig. 6. Relationship between crustal thickness and Bouguer anomaly in the Pacific mar-gin and Andean region. Red circles denote points used to establish a 3rd orderpolinomial (blue line in the bottom plot). Blue crosses in the top show the pointswith estimated crustal thicknesses used to complement the seismic data of Fig. 1 andmake the contour map of crustal thicknesses.
6 M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
was also used here to fill data gaps in the oceanic areas of the Atlanticmargin.
For the stable continental interior (onshore, East of the Andes),Assumpção et al. (in press) showed that the correlation betweenelevation and crustal thickness is poor (only 0.20). The correlationbetween crustal thickness and Bouguer anomalies is slightly better(correlation coefficient of −0.38). However, the regression ofcrustal thicknesses and Bouguer anomalies has an rms misfit of4.4 km (Assumpção et al., in press). The 260 crustal thicknessescompiled for the stable part of the continent have a mean valueof 37.7 km and a standard deviation of 4.8 km. For this reason,using gravity to fill large data gaps in the stable part of the
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
continent is not very useful and we preferred to interpolatebetween widely spaced data points (as done by Assumpção et al., inpress) rather than introduce bias from Bouguer anomalies in the con-tinental interior.
4.2. Model results and data misfit
Fig. 7 shows the crustal model constructed by interpolating seismicpoint constraints (receiver functions and active seismics) complementedwith thicknesses estimated from seismically-constrained gravity model-ing in the Andes (Tassara and Echaurren, 2012), and gravity-predictedestimates in the northern Andes and oceanic areas (one point every2.5°). We interpolated the data with the minimum curvature technique
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
7M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
(Smith andWessel, 1990;Wessel and Smith, 1998) after gettingweight-ed mean values every 0.5°. A surface tension factor of 0.5 was used toavoid large maxima and minima outside the data range. The rms fit tothe 920 seismic points is 3.5 km. The rms fit to the 260 seismic pointswithin the stable continental area is 1.6 km, close to the average accura-cy of the seismic data (Fig. 2).
Along the Andean chain, crust thicker than 60 km is seen fromcentral Peru (8°S) to the Puna plateau (30°S). In central Ecuadorand northern Peru (3°S to 7°S) the crust does not seem to bemore than 50 km thick, although no direct seismic data is avail-able in this region. In the stable continental interior, two mainareas of thick crust (40 km or more) can be seen, one in theAmazon craton and the other in the São Francisco and Paranábasin. The northern part of the intracratonic Paraná basin showsthe thickest crust, reaching more than 45 km. In the southernpart of the continent, east of the Andes, the crust is generally thin-ner than 40 km. A belt of thin crust (reaching 35 km or less) isseen along the sub-Andean basins, especially in Southern Colom-bia, Peru and Bolivia. This is based mainly on gravity data, butthe few available seismic points seem to be consistent with thisfeature.
−80° −75° −70° −65° −60° −−45°
−40°
−35°
−30°
−25°
−20°
−15°
−10°
−5°
0°
5°
10°
20
20
20
30
30
3040
40
40
40
40
40
40
50
50
50
50 60
60
70
70
Fig. 7. Contour map of crustal thickness of MODEL A (contour interval 5 km). Red dots showinferred from Bouguer anomalies. White dots denote thickness data given by seismically-conof the main geological provinces (see Fig. 1 for legend). Values of crustal thickness in the o
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
5. Crustalmodel B—point constraints and surface-wave tomography
5.1. Data and method
Here we use the tomographic inversion method of Feng et al.(2007) to derive a crustal model based on joint inversion of differentkinds of data: waveform modeling, group velocity dispersion, andseismic point constraints. The waveform data was the same as usedby Feng et al. (2007): 1537 regional wave trains analyzed with thepartitioned waveform inversion (PWI) method of van der Lee andNolet (1997) and van der Lee et al. (2001). Each waveform modeling(S+Rayleigh waves) produces an average 1D model for the epicenter-station path. To the 5700 group velocity measurements of Feng et al.(2004, 2007), we added 1031 new measurements (Collaço et al.,2012; Rosa et al., 2012) including inter-station paths using ambientnoise cross-correlation in the Paraná and Chaco basins (Fig. 8). Thesame seismic point constraints used in model A were also used here,with large weighting, to constrain model B. Additionally, point con-straints from the gravity model of Tassara and Echaurren (2012), andfrom the empirical Bouguer relations for unsampled areas in the North-ern Andes and oceans, were also used, but with lower weights.
55° −50° −45° −40° −35°
20
30
30
40
40
40
40
10
20
30
40
50
60
70
seismic data (RF and active seismics). Crosses in the oceanic areas denote thicknessesstrained gravity modeling of Tassara and Echaurren (2012). Green lines are boundariescean include the water layer.
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
8 M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
First, a 2Dgroup velocity tomography is carried out on a 1° by 1° geo-graphical grid, for each period, with the ~(5700+1031) paths using theray-theory tomography method of Feng et al. (2004). Ray-theory to-mography can offer nearly identical result as finite-frequency tomogra-phy (Ritzwoller et al., 2002; Sieminski et al., 2004) in continental-scalestudies if appropriate smoothing is applied and the path distribution isgood. The regionalized dispersion curves are then combined with the1D path-averaged models (obtained from the PWI inversions) to obtainperturbations of a 3D initial S-velocity structure. The initial reference 3Dmodel, called SA40, is the IASP91 upper mantle model with a 40 kmcrust in the stable part of South America and 45 to 70 km in the Andes(van der Lee et al., 2001). The two data sets are then jointly invertedwith the following set of equations (Feng et al., 2007; Lloyd et al., 2010):
HwλdHdλRFHRFλI
0BB@
1CCAm ¼
qwλdqdλRFqRF0
0BB@
1CCA ð3Þ
where Hw,Hd andHRF are the sensitivitymatrix for thewaveform, groupvelocity data and seismic point constraints, respectively;m is the modelparameters, i.e., 3D S-wave velocity perturbations (Δβ) andMohodepthperturbations (ΔH) from the referencemodel SA40;qw is thedata vectorrelated to the PWI data; qd is the data vector of group velocity differencebetween the regionalized and predicted group velocities (ΔU); qRF is thevector with crustal thickness data; λ is a damping factor, and λd and λRFcontrol the relative weights between waveform, group velocity andcrustal thickness data. In addition, we weight all the data by their re-spective inverse estimated uncertainty.
5.2. Model results
Wepresent three versions ofmodel B, obtainedwith different combi-nations of seismic data types, to investigate their contribution to the finalmodel: model B1 — based on waveform modeling and Rayleigh-wavegroup dispersion; model B2 — where the point constraints compiled inthis study were added, including those derived from the empirical rela-tionships (Eqs. (1) and (2)); and model B3 – where gravity constraints
Fig. 8. Additional 1031 paths with group velocity measurements used to complement the ~cently installed Brazilian stations and some GSN stations between 2009 and 2011. Center)Interstation paths measured with ambient noise cross-correlation (Bruno Collaço, USP, 201
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
for the Andean region from Tassara and Echaurren (2012) were added.Fig. 9 shows the resulting model based only on waveform modelingand Rayleigh-wave group velocity measurements (model B1). Thickcrust can be seen in Southern Peru, Southern Bolivia and NorthernChile. Interestingly, the thick crust in the Bolivian Altiplano (between15oS and 20oS) is not retrieved by the surface-wave data only. This isprobably due to irregular distribution of ray paths and few ray crossingsover the Bolivian Altiplano. Similarly, in the Ecuadorian/N.Peruvian area,the surface-wave data does not indicate a thick crust. In the stable part ofthe continent, the surface-wave data indicate thicker crust (>40 km) inthe Amazon and thinner in the Atlantic shield.
Adding the seismic depth constraints we get model B2 (Fig. 10). Inthis model extra points were also used in the ocean away from theseismically determined point constraints (based on the empirical re-lationship with Bouguer anomaly, Eqs. (1) and (2)), as well as inthe northern Andes. The thick crust of the Bolivian altiplano is nowbetter defined with the inclusion of receiver function data. In the At-lantic shield, the thick crust (>40 km) spans a larger area, especiallyin the Paraná Basin and near the São Francisco craton. A large regionof thin crust (b35 km) is seen in the Chaco Basin.
Fig. 11A shows our preferred and final tomographic model (B3)where, in addition to the data used in model B2, we added thegravity-derived crustal thicknesses of Tassara and Echaurren (2012),which had been tightly constrained by seismic observations. ModelB3 fits the 730 on-shore seismic point constraints with an rms misfitof 4.7 km (Fig. 11B. For the stable continental interior (260 pointsonshore), the misfit is only 1.7 km. The misfit in the Andes (Fig. 11B)is larger partly because of smoothing of themodel in the inversion pro-cess and partly because of larger variability of the compiled data. Theweights in the tomographic inversion were chosen so that the seismicpoint constraints have relatively more influence than waveform or dis-persion data. Model B3 improves the definition of the thick Andeancrust. However, in the region of Southern Ecuador and Northern Peru(where no seismic data have been compiled), the crust remains rela-tively thin, about 40 km or less.
The map of Fig. 12 shows the S-wave velocity anomalies at 100 kmdepth, which is not the main topic of this work but is briefly describedhere as an interesting by-product of the tomographic inversion. The
5700 paths previously used by Feng et al. (2004, 2007). Left) Earthquake data from re-Paths measured at LPA station (University of La Plata; María Laura Rosa 2012). Right)2).
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
Fig. 9. Model of Moho depths (tomographic inversion method of Feng et al., 2007) using only waveform modeling and group velocity dispersion (Model_B1_WF+DSP). Samewaveform data as in the 2007 model; dispersion data from 2007 with additional paths from Fig. 8. Blue line denotes 3000 m elevation of the Andean plateau. Note generallythick crust in the Andes, except at the Southern Bolivia plateau (near 18°S), probably due to insufficient ray-crossings in that area. No deep crust is suggested in the Ecuadorean/N. Peru area (0° to 5°S). Deep crust in the northern and southern Paraná basin is suggested, but not near the São Francisco craton.
9M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
results are very similar to those of Feng et al. (2007) except in thesouthern part of the Paraná Basin where higher velocities are found.The inclusion of more dispersion paths (Fig. 8) improved the veloci-ties of the lithospheric lid in Eastern Argentina. The region of theChaco Basin is characterized by low velocities in the upper-mantle,which correlate with the generally thin crust (Fig. 11A), as discussedbelow. In the Andes, the velocities at 100 km depth are generally lowreflecting the shallow asthenosphere, except for the high velocitiesin the two areas of flat subduction of the Nazca slab: in southernEcuador/northern Peru near 5°S, and in central Chile/Argentina near30°S.
5.3. Data misfit
Examples of fitting of the surface-wave data for the final model B3are shown in Fig. 13, compared with the initial reference model SA40.Models inverted with only one type of data (DSP=group velocitiesonly, or WF=PWI waveform data only) are also shown for compari-son. The joint model B3 fits the DSP and theWF data almost as well as
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
the model inverted with only the specific data. Interestingly, the WFdata improves the group velocity fit, relative to the initial model(Fig. 13B), but using only the DSP data one cannot improve the wave-form fit of the seismograms (Fig. 13D). The amplitude and phase in-formation contained in the PWI modeling are very important to betterconstrain the final model.
6. Discussion
6.1. Stable continental interior
Both models A and B3 show that the average crustal thickness inthe stable part of the continent is about 40 km. Two regions withcrust generally thicker than 40 km can be seen in the Amazon andin the Atlantic shield. This is seen both in model A (Fig. 7, based onseismic constraints only) and model B3 (Fig. 11A) includingsurface-wave tomography. The more detailed features in the finalmodel B3 indicate that the thick crust in the Amazon correlatesroughly with the oldest Archean provinces as defined by Tassinari
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
Fig. 10.Model B2 of Moho depths (tomographic method of Feng et al., 2007) using waveform modeling and group velocity dispersion (same data as in Fig. 9) plus 762 seismic pointconstraints. In the oceans and Northern Andes, gravity-derived additional points were used to fill large gaps, based on Eqs. (1) and (2) (shown in Fig. 7).
10 M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
and Macambira (1999). In the Atlantic shield (i.e., SE of theTransBrasilian Lineament, Fig. 11A) there is evidence of thick crustassociated with the São Francisco craton but the deepest Moho(reaching more than 45 km) is found in the intracratonic ParanáBasin based mainly on receiver functions (Assumpção et al., 2002;Julià et al., 2008). The basin has generally lower elevations than thesurrounding areas, which led Assumpção et al. (2002) to proposethat the deep crust in this basin could be compensated by high densi-ties in the lithospheric lid. This hypothesis remains consistent withhigh S-wave velocities at 100 km depth shown in Fig. 12.
A SW–NE continental-scale lineament, the TransBrasiliano Linea-ment (“TBL” in Fig. 11A), is believed to indicate a major Neo-Proterozoic suture between the Amazon craton and the other provincesto the SE (São Francisco craton and other old crustal units). Inmodel B3,the TBL seems to be characterized by relatively thin crust, especially be-neath the Parnaiba basin. Another pronounced belt of thin crust is seeninmodel B3 along the eastern and southern borders of the Amazon cra-ton. This feature could indicate an extension of a narrow belt of thincrust found by Assumpção et al. (2004a, 2004b) in the Tocantins prov-ince of central Brazil roughly coincident with the Neo-ProterozoicGoiás magmatic arc.
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
Crustal thicknesses of the Rio de la Plata craton of eastern Argen-tina also seem to be larger than 40 km The crust in the Patagonianprovince, in the southern part of the continent, appears thinnerthan 40 km, but the coverage of our data set is poor in the southerntip of South America and model B3 has poor resolution in this area.
6.2. Andean range
The Moho beneath the Andes can reach more than 70 km depth inthe eastern cordillera of Bolivia, and is generally deeper than 50 km inareas with elevations higher than 3000 m. A notable exception to thisis observed below the northern Puna at 25°S where the crust is thinnerthan 55 km for elevations higher than 4000 m. This result reinforce thehypothesis as to that a recent (b4 Ma) event of lithospheric delamina-tion has affected this region and removed part of the dense lowercrust (Kay and Kay, 1993; Whitman et al., 1996). Interestingly, andnot noted by previous models (Feng et al., 2007; Lloyd et al., 2010;Tassara et al., 2006), this region of thin crust seems to be oriented in aENEdirection that could be continued eastward into the TransBrasilianoLineament. The SW limit of the TBL is presently characterized by a me-chanically weak (Pérez-Gussinyé et al., 2007), thinned lithosphere
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
Fig. 11. A) Model B3 of Moho depths using waveform modeling, group velocity dispersion, all point constraints as model B2 with additional depths from the model ofTassara and Echaurren (2012). Control points are shown in panel B. Contour interval 5 km. Blue solid line denotes the 3000 m altitude in the Andes. Dark green linesare boundaries of major geological provinces. Gray solid lines are the limits of the South American plate. Red line in the Amazon is the oldest Archean geochronologicalprovince (Tassinari and Macambira, 1999). TBL=TransBrasiliano Lineament; Ch=Chaco Basin, Or=Oriente basin. Note better definition of the Andean thick crust. Notethat in South Ecuador and Northern Peru (~3°S), the crust is still thin. B) Fit to MODEL B3. Crustal thickness contours with all point-constraints used in the continent:colored circles are the seismic constraints; rms misfit to all 728 continental data points=4.7 km; fit to 259 points in stable continent=1.7 km). Small open circles denotecontrol points from Tassara and Echaurren (2012) gravity-derived model; crosses show locations of thicknesses derived from the empirical relationship with Bougueranomaly (Eq. (1), Fig. 6).
11M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
(Assumpção et al. (2004b) as suggested by the low S-wave veloc-ities in Fig. 12). Although highly speculative, this hypothesiscould suggest that delamination below the Puna occurred alonga pre-weakened, thinned lithospheric zone inherited since theNeo-Proterozoic.
In southern Ecuador and northern Peru (around 3°S), at the limitbetween central and northern Andes, another region of relativelyhigh elevation (near 3000 m) is underlain by a Moho that seems tobe only about 40 km or less. In this area there are no direct seismicconstraints and the thin crust is controlled by the surface wave dataand some gravity-predicted estimates of crustal thickness (Fig. 11B).Geoid anomalies are mainly positive and higher than 20 m alongthe whole Andean chain, reaching more than 40 m in the Altiplano.However, in the south Ecuatorian Andes this anomaly almost van-ishes and geoid heights are less than 10 m (Blitzkov et al., 2009;King, 2002). Although long wavelength geoid anomalies are mainlycontrolled by deeper structures (subducted plates and density varia-tions at the mid-lower mantle), the lack of a short wavelength positivegeoid anomaly can be connected to the lack of a deep crustal root in thisregion. In this case, thin cordilleran crust can be the consequences of aregional, flexural-controlled compensation of surface topography incontrast to the local, Airy-type compensation that seems to dominatealong the rest of the Andean cordillera as suggested by the correlationbetween topography and crustal thickness. Moreover, estimates of theelastic thickness that parametrizes theflexural rigidity of the lithosphere(Pérez-Gussinyé et al., 2008; Tassara et al., 2007) are anomalously high
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
(>40 km) for the Ecuador–Peru border region compared to the rest ofthe Andes (b20 km).
6.3. Sub-Andean and Chaco basins
Another interesting feature of the Mohomodels is a large region ofthin crust (shallower thanb35 km) between the Andean range andthe cratonic areas. In some areas of the Chaco basin, model B3suggests a Moho depth even shallower than 30 km (Fig. 11A).Although the Moho model in the sub-Andean region (Fig. 11A ismainly controlled by the surface-wave data, the few direct seismicmeasurements in these areas (Fig. 7) and the relatively high gravityanomalies (Tassara and Echaurren, 2012; Tassara et al., 2006) confirmthe generally thin crust east of the Altiplano–Puna plateau of thecentral Andes. Interestingly, this feature has been independentlyfound from the inversion of satellite gravity data (van der Meijdeet al., submitted for publication).
Using surface-wave data to invert for both crustal thickness andS-wave velocity of the lithospheric lid can be a difficult problem be-cause of the trade-off between these two parameters. A solutionwith a thin crust could be compensated by lower velocities in theupper mantle and produce similar dispersion curves. Snoke andJames (1997) analyzed phase and group velocity data in the Chacobasin and showed that this region has very low S-wave velocities inthe upper mantle, which was later confirmed by Feng et al. (2007).In addition, checker board resolution tests by Feng et al. (2007)
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
Fig. 12. S-wave velocity anomalies at 100 km depth from the tomography model B3.Red line in the Amazon denotes the oldest geochronological province of Tassinariand Macambira (1999). TBL is the TransBrasiliano Lineament.
12 M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
showed that the low mantle velocities in the Chaco basin are well de-termined. This means that this feature is robust and should not causean artificially thin crust in our tomographic inversion.
We do not have a definitive explanation for the origin of this zoneof thin crust along the sub-Andean belt, but some points are worthdiscussing. Flexural uplift in the sub-Andean crust caused by Andeanload can only account for a few meters, even when 3D amplificationeffects due to the bend of the Andean chain are taken into account(Sacek and Ussami, 2009). Dynamic topography of other possibleorigin (subducted slabs, lower crustal flow, etc.) normally accountsfor at most a couple of kilometers (Clark et al., 2005; Dávila et al.,2010; Pérez-Gussinyé et al., 2008) and also seems unlikely to explaina shallowing of the Moho by 5–10 km in a large area as observed forthe Chaco basin here. We note that crust of 30–35 km thickness isalso observed in our models A and B3 along the eastern passive mar-gin of the continent where it is of course a consequence of the crustalextension and thinning accompanying the rifting of Gondwana. Thearea of thin crust bordering the Amazon and Rio de la Plata cratonin the west, e.g. lying to the east of the Andes, coincides with whathas been proposed as the main arm of the intracontinental rift zonethat separated Laurasia from proto-Gondwana during the disaggrega-tion of the supercontinent Rodinia and the aperture of the Iapetus
Fig. 13. Data misfit. A) Example of group velocity misfit for one event in the Atlantic. DottedB3 (joint inversion). B) Average misfit of the up to 7000 measurements. SA40 is misfit to theor waveform data only, respectively; JOINT is misfit using all data including seismic constraifor the S and Rayleigh waves of an event in Central Chile. Solid line is the observed vertical cis the calculated trace for the final joint model B3. Note better fit for the shorter periods at thto the traces) for the initial reference model (SA40), model obtained with group velocity onversion of all data sets (JOINT). Number at the top of each plot is the average misfit. JOINT fitexample in (C) is 1.91 and 0.64 for the SA40 and JOINT model, respectively.
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
Ocean at the Neo-Proterozoic (Cawood, 2005; Cordani et al., 2009;Rapela et al., 2007). The region of anomalously thin crust (b30 km)below the Chaco basin occurs at the intersection of this old riftedmargin and the TransBrasiliano Lineament, which also coincidewith the suture left after the closure during the Cambrian of theClymene Ocean that separated Amazonia from Sao Francisco andRio de la Plata cratons (Escayola et al., 2011; Tohver et al., 2012).Thus, it seems very likely that the present-day crustal thicknessconfiguration at the interior of the South American continentreflects an old inheritance left after previous steps of the Wilsoncycle.
7. Conclusions
A new model of Moho depth was derived for continental SouthAmerica based on an increased data set of more than 900 crustalthickness constraints from receiver functions and active seismicexperiments. The surface-wave data originally used by Feng et al.(2007) was increased by ~1000 paths with new group velocity mea-surements improving the coverage in the Paraná and Chaco basins. Ajoint inversion of all these data sets, including extra (weak) con-straints from gravity-derived depths in poorly covered areas, pro-duced a model (model B3, Fig. 11A) fitting all used data reasonablywell, i.e. within their uncertainties.
The preferred model shows two main regions of thick crust in thestable continental interior, one in the Amazon craton (mainly in theoldest geochronological province) and the other in the Atlantic shield(including the São Francisco craton and the Paraná intracratonicbasin) separated by a zone of thin crust which could be correlatedwith the Transbrasilian Lineament.
TheMoho in the Andean chain is usually deeper than 50 km in areaswith elevations higher than 3000 m, reaching more than 70 km in theBolivian Eastern Cordillera.
A shallowing of the AndeanMoho seems to occur below the Puna ofnorth-western Argentina and in the border region between Ecuadorand Peru. Causes for the lack of a deep crustal root compensating highelevations for these regions vary from a possible delamination eventand then a thermal compensation for the Puna, while for the southernEcuatorian Andes it seems likely that compensation is dominated bythe flexural support of a rigid lithosphere.
A belt of thin crust (b35 km thick) that is seen below most of thesub-Andean zone between the Andean orogeny and the stable craton-ic shield, especially beneath the Chaco basin, is probably an oldinherited feature left after the rifting and drifting episodes relatedto the separation of Rodinia during the Neo-Proterozoic and subse-quent closure of ocean basins connected to the first orogenic cyclesthat formed the Phanerozoic Gondwana continent and then the mod-ern South America.
Acknowledgments
This work is supported by CNPq grant 309724/2009-0 (MA), NSFCgrant 41174039 (MF) and Fondecyt project 1101034 (AT). Part of thedata used in this paper was collected in the BRASIS project (IAG-USP)supported by “Rede de Estudos Geotectônicos” of Petrobras. We thank
line is the curve for the SA40 initial model. Solid curve is the prediction with the modelinitial reference model; DSP and WF are misfits to models inverted with group velocitynts. JOINT is similar to DSP, and is better than SA40 andWF. C) Example of waveform fitomponent, dotted line is the calculated trace for the SA40 initial model, and dashed linee end of the Rayleigh wave train. D) Distribution of all waveform fits (normalized rms fitly (DSP), model obtained with waveform only (WF), and final model B3 with joint in-s are close to WF, and are better than DSP and the initial model SA40. The misfit for the
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
13M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
María Laura Rosa (UNLP), Bruno Collaço (IAG-USP) and Gerardo Sánchez(INPRES) for dispersion data from their stations. We also thank EricSandvol (Missouri Univ.) for providing unpublished crustal thicknesses
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
from the Southern Puna experiment, Kristin Phillips (Caltech) forunpublished data from the Southern Peru experiment, and StéphaneDrouet (ON-Brazil) for unpublished data from RSIS network.
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
14 M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
References
Agudelo, W.A., Ribodetti, J.-Y., Collot, S. Operto, 2009. Joint inversion of multichannelseismic reflection and wide-angle seismic data: improved imaging and refined ve-locity model of the crustal structure of the north Ecuador–South Colombia conver-gent margin. Journal of Geophysical Research 114, B02306 http://dx.doi.org/10.1029/2008JB005690.
Ammon, C.J., 1991. The isolation of receiver effects from teleseismic P waveforms. Bul-letin of the Seismological Society of America 81 (6), 2504–2510.
ANCORP Working Group, 2003. Seismic imaging of a convergent continental marginand plateau in the central Andes (Andean Continental Research Project 1996(ANCORP'96)). Journal of Geophysical Research 108 (B7), 2328 http://dx.doi.org/10.1029/2002JB001771.
Assumpção, M., James, D., Snoke, A., 2002. Crustal thicknesses in SE Brazilian shieldby receiver function analysis: implications for isostatic compensation. Journalof Geophysical Research 107 (B1) http://dx.doi.org/10.1029/2001JB000422(ESE2-1—ESE2-14, 2006).
Assumpção, M., An, M., Bianchi, M., França, G.S.L., Rocha, M., Barbosa, J.R., Berrocal, J.,2004b. Seismic studies of the Brasília Fold Belt at the western border of the SãoFrancisco craton, central Brazil, using receiver function, surface wave dispersion,and teleseismic tomography. Tectonophysics 388, 173–185.
Assumpção, M., Schimmel, M., Escalante, C., Rocha, M., Barbosa, J.R., Barros, L.V., 2004a.Intraplate seismicity in SE Brazil: stress concentration in lithospheric thin spots.Geophysical Journal International 159, 390–399.
Assumpção, M., Bianchi, M.B., Julià, J., Dias, F.L., França, G.S., Nascimento, R.M., Drouet, S.,Pavão, C.G., Albuquerque, D.F., Lopes, A.E.V., in press. Crustal thickness map of Brazil:data compilation and main features. Journal of South American Earth Sciences.
Bassin, C., Laske, G., Masters, G., 2000. The current limits of resolution for surface wavetomography in North America. EOS Transactions of the American GeophysicalUnion 81, F897.
Bastow, I.D., Nyblade, A.A., Stuart, G.W., Rooney, T.O., Benoit, M.H., 2008. Upper mantleseismic structure beneath the Ethiopian hot spot: rifting at the edge of the Africanlow-velocity anomaly. Geochemistry, Geophysics, Geosystems 9, Q12022 http://dx.doi.org/10.1029/2008GC002107.
Beck, S.L., Zandt, G., 2002. The nature of orogenic crust in the central Andes. Journal ofGeophysical Research 107, 2230 http://dx.doi.org/10.1029/2000JB000124.
Beck, S.L., Zandt, G., Myers, S.C., Wallace, T.C., Silver, P.G., Drake, L., 1996. Crustal-thicknessvariations in the central Andes. Geology 24, 407–410.
Bianchi, M., Heit, B., Jakovlev, A., Yuan, A., Kay, S.M., Sandvol, E., Alonso, R.N., Coira, B.,Brown, L., Kind, R., Comte, D., in press. Teleseismic tomography of the southernPuna plateau in Argentina and adjacent regions. Tectonophysics. http://dx.doi.org/10.1016/j.tecto.2012.11.016.
Blitzkov, D., Matos, A.C.O.C., Lobianco, M.C.B., Campos, I.O., 2009. The progress of thegeoid in South America under GRACE and EGM08 models. IAG Assembly, BuenosAires, 2009 (Abstracts).
Calkins, J.A., Zandt, G., Gilbert, H.J., Beck, S., 2006. Crustal images from San Juan, Argentina,obtained using high frequency local event receiver functions. Geophysical ResearchLetters 33, L07309 http://dx.doi.org/10.1029/2005GL025516.
Cawood, P.A., 2005. Terra Australis Orogen: Rodinia breakup and development ofthe Pacific and Iapetus margins of Gondwana during the Neoproterozoic and Pa-leozoic. Earth-Science Reviews 69 (3–4), 249–279 http://dx.doi.org/10.1016/j.earscirev.2004.09.001.
Clark, M.K., Bush, J.W.M., Royden, L.H., 2005. Dynamic topography produced by lowercrustal flow against rheological strength heterogeneities bordering the TibetanPlateau. Geophysical Journal International 162, 575–590 http://dx.doi.org/10.1111/j.1365-246X.2005.02580.x.
Collaço, B., Sánchez, G., Rosa, M.].L., Assumpção, M., Sabbione, N., Araujo, M., Rocha, M.P.,2012. Crustal structure of the Paraná Basin from ambient noise tomography. XVICongr. Geological Soc. of Peru, Lima, 23–26 September 2012, extended abstract.
Cordani, U.G., Teixeira, W., D'Agrella-Filho, M.S., Trindade, R.I., 2009. The position of theAmazonian Craton in supercontinents. Gondwana Research 15 (3–4), 396–407http://dx.doi.org/10.1016/j.gr.2008.12.005.
Cormier, V.F., Anderson, T.S., 2004. Efficiency of Lg propagation from SmS dynamic raytracing in three-dimensionally varying crustal waveguides. Pure and Applied Geo-physics 8, 1613–1633.
Dávila, F.M., Lithgow-Bertelloni, C., Giménez, M., 2010. Tectonic and dynamic controlson the topography and subsidence of the Argentine Pampas: the role of the flatslab. Earth and Planetary Science Letters 295 (1–2), 187–194 http://dx.doi.org/10.1016/j.epsl.2010.03.039.
Dorbath, C., Granet, M., Poupinet, G., Martinez, C., 1993. A teleseismic study of theAltiplano and Eastern Cordillera in Northern Bolivia: new constraints on a litho-spheric model. Journal of Geophysical Research 98 (B6), 9825–9844.
Escayola, M.P., van Staal, Cees R., Davis, William J., 2011. The age and tectonic setting ofthe Puncoviscana Formation in northwestern Argentina: an accretionary complexrelated to Early Cambrian closure of the Puncoviscana Ocean and accretion of theArequipa–Antofalla block. Journal of South American Earth Sciences 32 (4),438–459 http://dx.doi.org/10.1016/j.jsames.2011.04.013.
Fan, G.W., Lay, T., 1998. Statistical analysis of irregular wave-guide influences on re-gional seismic discriminants in China. Bulletin of the Seismological Society ofAmerica 88, 74–88.
Feininger, T., Seguin, M.K., 1983. Simple Bouguer gravity anomaly field and the inferredcrustal structure of continental Ecuador. Geology 11, 40–44 http://dx.doi.org/10.1130/0091-7613(1983) 11b40:SBGAFA>2.0.CO;2.
Feng, M., Assumpção, M., Van der Lee, S., 2004. Group-velocity tomography and litho-spheric S-velocity structure of the South American continent. Physics of the Earthand Planetary Interiors 147, 315–331.
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
Feng, M., Van der Lee, S., Assumpção, M., 2007. Upper mantle structure of South Amer-ica from joint inversion of waveforms and fundamental-mode group velocities ofRayleigh waves. Journal of Geophysical Research 112, B04312 http://dx.doi.org/10.1029/2006JB004449.
Ferreira, J.M., França, G.S., Vilar, C.S., do Nascimento, A.F., Bezerra, F.H.R., Assumpção,M., 2008. Induced seismicity in the Castanhão reservoir, NE Brazil — preliminaryresults. Tectonophysics 456, 103–110.
Fichtner, A., Igel, H., 2008. Efficient numerical surface wave propagation through theoptimization of discrete crustal models—a technique based on non-linear disper-sion curve matching (DCM). Geophysical Journal International 173, 519–533http://dx.doi.org/10.1111/j.1365-246X.2008.03746.x.
Fichtner, A., Kennett, B.L.N., Igel, H., Bunge, H.-P., 2009. Full seismic waveform tomog-raphy for upper-mantle structure in the Australasian region using adjointmethods. Geophysical Journal International 179, 1703–1725 http://dx.doi.org/10.1111/j.1365-246X.2009.04368.x.
Gans, C.R., Beck, S.L., Zandt, G., Gilbert, H., Alvarado, P., Anderson, M., Linkimer, L.,2011. Continental and oceanic crustal structure of the Pampean flat slabregion, western Argentina, using receiver function analysis: new high-resolutionresults. Geophysical Journal International 186, 45–58 http://dx.doi.org/10.1111/j.1365-246X.2011.05023.x.
Hjörleifsdóttir, V., Ekstrom, G., 2010. Effects of three-dimensional Earth structure onCMT earthquake parameters. Physics of the Earth and Planetary Interiors 179,178–190 http://dx.doi.org/10.1016/j.pepi.2009.11.003.
Hussong, D.M., Edwards, P.B., Johnson, S.H., Campbell, J.F., Sutton, G.H., 1976. Crustalstructure of the Peru–Chile trench: 8°–12°S latitude. The geophysics of the PacificOcean Basin and its margins. AGU Geophysical Monograph 19, 71–85.
Julià, J., Assumpção, M., Rocha, M.P., 2008. Deep crustal structure of the Paraná Basinfrom receiver functions and Rayleigh-wave dispersion: evidence for a fragmentedcratonic root. Journal of Geophysical Research 113, B08318 http://dx.doi.org/10.1029/2007JB005374.
Justowski, B., Dziewonski, A.M., Ekstrom, G., 2007. Nonlinear crustal corrections fornormal-mode seismograms. Bulletin of the Seismological Society of America 97,1756–1762 http://dx.doi.org/10.1785/0120070041.
King, S.D., 2002. Geoid and topography over subduction zones: the effect of phasetransformations. Journal of Geophysical Research 107 (B1), 2013 http://dx.doi.org/10.1029/2000JB000141.
Lawrence, J.F., Wiens, D.A., 2004. Combined receiver-function and surface wave phase-velocity inversion using a Niching Genetic Algorithm: application to Patagonia.Bulletin of the Seismological Society of America 94 (3), 977–987.
Lebedev, S., Van Der Hilst, Rob D., 2008. Global upper-mantle tomography with the au-tomated multimode inversion of surface and S-wave forms. Geophysical JournalInternational 173, 505–518 http://dx.doi.org/10.1111/j.1365-246X.2008.03721.x.
Lithgow-Bertelloni, C., Guynn, J.H., 2004. Origin of the lithospheric stress field. Journalof Geophysical Research 109, B01408 http://dx.doi.org/10.1029/2003JB002467.
Lloyd, S., van der Lee, S., França, G.S., Assumpção, M., Feng, M., 2010. Moho map ofSouth America from receiver functions and surface waves. Journal of GeophysicalResearch 115, B11315 http://dx.doi.org/10.1029/2009JB006829.
McGlashan, N., Brown, L., Kay, S., 2008. Crustal thickness in the central Andes fromteleseismically recorded depth phase precursors. Geophysical Journal International175, 1013–1022 http://dx.doi.org/10.1111/j.1365-246X.2008.03897.x.
Meyer, R.P., Mooney, W.D., Hales, A.L., Helsey, C.E., Woollard, G.P., Hussong, D.M.,Kroenke, L.W., Ramirez, J.E., 1976. Project Nariño III: refraction observation acrossa leading edge, Malpelo Island to the Colombian Cordillera Occidental. The geo-physics of the Pacific Ocean basin and its margins. AGU Geophysical Monograph19, 105–132.
Mohriak, W.U., Mello, M.R., Bassetto, M., Vieira, I.S., Koutsoukos, E.A.M., 2000. In: Mello,M.R., Katz, B.J. (Eds.), Crustal architecture, sedimentation, and petroleum systemsin the Sergipe–Alagoas Basin, northeastern Brazil: Petroleum Systems of SouthAtlantic Margins: AAPG Memoir, 73, pp. 273–300.
Mooney, W.D., Kaban, M.K., 2010. The North American upper mantle: density,composition, and evolution. Journal of Geophysical Research 115, B12424 http://dx.doi.org/10.1029/2010JB000866.
Mooney, W.D., Meyer, R.P., Laurence, J.P., Meyer, H., Ramírez, J.E., 1979. Seismic refractionstudies of the western cordillera, Colombia. Bulletin of the Seismological Society ofAmerica 69, 1745–1761.
Mooney, W.D., Laske, G., Masters, T.G., 1998. CRUST 5.1: a global crustal model at 5°x5°.Journal of Geophysical Research 103 (B1), 727–747.
Niu, F., Bravo, T., Pavlis, G., Vernon, F., Rendon, H., Bezada, M., Levander, A., 2007. Re-ceiver function study of the crustal structure of the southeastern Caribbean plateboundary and Venezuela. Journal of Geophysical Research 112, B11308 http://dx.doi.org/10.1029/2006JB004802.
Ocola, L.C., Aldrich, L.T., Gettrust, J.F., Meyer, R.P., Ramírez, J.E., 1975. Project NariñoI: crustal structure under southern Colombian–northern Ecuador Andes fromseismic refraction data. Bulletin of the Seismological Society of America 65,1681–1695.
Park, Y., Nyblade, A.A., Rodgers, A.J., Al-Amri, A., 2008. S wave velocity structure ofthe Arabian Shield upper mantle from Rayleigh wave tomography. Geochemistry,Geophysics, Geosystems 9, Q07020 http://dx.doi.org/10.1029/2007GC001895.
Pavão, C.G., França, G.S., Marotta, G.S., Menezes, P.H.B.J., Neto, G.B.S., Roig, H.L., 2012.Spatial interpolation applied to crustal thickness in Brazil. Journal of GeographicInformation System 4 (2), 142–152 http://dx.doi.org/10.4236/jgis.2012.42019.
Pérez-Gussinyé, M., Lowry, A.R., Watts, A.B., 2007. Effective elastic thickness of SouthAmerica and its implications for intra-continental deformation. Geochemistry,Geophysics, Geosystems 8, Q05009 http://dx.doi.org/10.1029/2006GC001511.
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
15M. Assumpção et al. / Tectonophysics xxx (2013) xxx–xxx
Pérez-Gussinyé, M., Lowry, A.R., Phipps Morgan, J., Tassara, A., 2008. Effective elasticthickness variations along the Andean margin and their relationship to subductiongeometry. Geochemistry, Geophysics, Geosystems 9, Q02003 http://dx.doi.org/10.1029/2007GC001786.
Phillips, K., Clayton, R.W., Davis, P., Tavera, H., Guy, R., Skinner, S., Stubailo, I., Audin, L.,Aguilar, V., 2012. Structure of the subduction system in southern Peru from seismicarray data. Journal of Geophysical Research 117, B11306 http://dx.doi.org/10.1029/2012JB009540.
Ramos, V.A., 2004. La plataforma patagónica y sus relaciones con la plataformabrasilera. In: Matess-Neto, V., et al. (Ed.), Geologia do Continente Sul-Americano:Evolução e Obra de Fernando Flávio Marques de Almeida. Editora Beca, SãoPaulo, Brazil, pp. 371–381.
Rapela, C.W., Pankhurst, R.J., Casquet, C., Fanning, C.M., Baldo, E.G., González-Casado,J.M., Galindo, C., Dahlquist, J., 2007. The Río de la Plata craton and the assemblyof SW Gondwana. Earth-Science Reviews 83 (1–2), 49–82 http://dx.doi.org/10.1016/j.earscirev.2007.03.004.
Ritzwoller, M.H., Shapiro, N.M., Barmin, M.P., Levshin, A.L., 2002. Global surface wavediffraction tomography. Journal of Geophysical Research 107 (B12), 2335 http://dx.doi.org/10.1029/2002JB001777.
Robalino, F., 1977. Espesor de la corteza en Quito mediante al análisis del espectro delas ondas P de periodo largo. XI Asamblea General, Instituto Panamericano deGeografia e Historia, Quito (Abstract).
Rosa, M.L., Assumpção, M., Sabbione, N., 2010. Crustal structure in the eastern marginof Argentina with receiver functions at LPA and TRQA: evidence for crustal thin-ning near Río de La Plata. AGU Meeting of the Americas, Brazil. Abstract.
Rosa, M.L., Collaço, B., Sánchez, G., Assumpção, M., Sabbione, N., Araujo, M., 2012. Crust-al thickness beneath the Chaco–Paraná Basin. NE Argentina. Using surface-wavesand ambient noise tomography. XVI Congr. Geological Soc. of Peru, Lima, 23–26September 2012, extended abstract.
Sá, N.C., 2004. O campo de gravidade, o geóide e a estrutura crustal na Amé ́rica do Sul:Novas estratégias de representação. Tese de Livre-Docência, Universidade de SãoPaulo. 121pp.
Sacek, V., Ussami, N., 2009. Reappraisal of the effective elastic thickness for the sub-Andes using 3-D finite element flexural modelling, gravity and geological con-straints. Geophysical Journal International 179, 778–786 http://dx.doi.org/10.1111/j.1365-246X.2009.04334.x.
Schmitz, M., Martins, A., Izarra, C., Jácome, M.I., Sánchez, J., Rocabado, V., 2005. The majorfeatures of the crustal structure in north-eastern Venezuela from deep wide-angleseismic observations and gravity modelling. Tectonophysics 399, 109–124.
Sieminski, A., Lévêque, J.-J., Debayle, E., 2004. Can finite-frequency effects be accountedfor in ray theory surface wave tomography? Geophysical Research Letters 31,L24614 http://dx.doi.org/10.1029/2004GL021402.
Smith, W.H.F., Wessel, P., 1990. Gridding with continuous curvature splines in tension.Geophysics 55 (3), 293–305.
Snoke, J.A., James, D.E., 1997. Lithospheric structure of the Chaco and Paraná Basins ofSouth America from surface-wave inversion. Journal of Geophysical Research 102,2939–2951.
Soller, D.R., Ray, R.D., Brown, R.D., 1982. A new global crustal thickness map. Tectonics1 (2), 125–149.
Soares, J.E., Berrocal, J., Fuck, R.A., Mooney, W.D., Ventura, D.B.R., 2006. Seismiccharacteristics of central Brazil crust and upper mantle: a deep seismic refractionstudy. Journal of Geophysical Research 111, B12302 http://dx.doi.org/10.1029/2005JB003769.
Please cite this article as: Assumpção, M., et al., Models of crustal thicknessface wave tomography, Tectonophysics (2013), http://dx.doi.org/10.1016
Stratford, W., Thybo, H., Faleide, J.I., Olesen, O., Tryggvason, A., 2009. New Moho mapfor onshore southern Norway. Geophysical Journal International 178, 1755–1765http://dx.doi.org/10.1111/j.1365-246X.2009.04240.x.
Swenson, J.L., Beck, S.L., Zandt, G., 2000. Crustal structure of the Altiplano from broad-band regional waveform modeling: implications for the composition of thick con-tinental crust. Journal of Geophysical Research 105, 607–621.
Tanimoto, T., 1995. Crustal structure of the Earth. Global Earth Physics, A Handbook ofPhysical Constants. In: Ahrens, T.J. (Ed.), AGU reference shelf Series, vol 1. AmericanGeophysical Union, Washington, DC, pp. 214–224 (ISBN 0-87590-851-9).
Tassara, A., Echaurren, A., 2012. Anatomy of the Andean subduction zone: three-dimensional density model upgraded and compared against global-scale models.Geophysical Journal International 189 (1), 161–168 http://dx.doi.org/10.1111/j.1365-246X.2012.05397.x.
Tassara, A., Götze, H.-J., Schmidt, S., Hackney, R., 2006. Three-dimensional densitymodel of the Nazca plate and the Andean continental margin. Journal of Geophys-ical Research 111, B09404 http://dx.doi.org/10.1029/2005JB003976.
Tassara, A., Swain, C., Hackney, R., Kirby, J., 2007. Elastic thickness structure of SouthAmerica estimated using wavelets and satellite-derived gravity data. Earth and Plan-etary Science Letters 253 (1–2), 17–36 http://dx.doi.org/10.1016/j.epsl.2006.10.008.
Tassinari, C.C.G., Macambira, M.J.B., 1999. Geochronological provinces of the Amazo-nian craton. Episodes 22 (3), 174–182.
Tohver, E., Cawood, P.A., Rossello, E.A., Jourdan, F., 2012. Closure of the Clymene Oceanand formation of West Gondwana in the Cambrian: evidence from the SierrasAustrales of the southernmost Rio de la Plata craton, Argentina. GondwanaResearch 21 (2–3), 394–405 http://dx.doi.org/10.1016/j.gr.2011.04.001.
van der Lee, S., Nolet, G., 1997. Upper mantle S velocity structure of North America.Journal of Geophysical Research 102, 22,815–22,838.
van der Lee, S., James, D., Silver, P., 2001. Upper mantle S velocity structure ofcentral and western South America. Journal of Geophysical Research 106,30821–30834.
van der Meijde, M., Julià, J., Assumpção, M., submitted for publication. Satellite gravityderived Moho for South America. Tectonophysics, Special issue.
Wessel, P., Smith, W.H.F., 1998. New, improved version of Generic Mapping Tools re-leased. EOS Transactions of the American Geophysical Union 79 (47), 579.
Whitman, D., Isacks, B.L., Mahlburg Kay, S., 1996. Lithospheric structure and along-strike segmentation of the Central Andean Plateau: seismic Q, magmatism, flexure,topography and tectonics. Tectonophysics 259 (1–3), 29–40 http://dx.doi.org/10.1016/0040-1951(95)00130-1.
Yuan, X., Sobolev, S.V., Kind, R., 2002. Moho topography in the central Andes and itsgeodynamic implications. Earth Planetary Science Letters 199, 389–402 http://dx.doi.org/10.1016/S0012-821X(02)00589-7.
Zalán, P.V., Severino, M.C.G., Rigoti, C.A., Magnavita, L.P., Oliveira, J.A.B., Vianna, A.R.,2011. An entirely new 3D-view of the crustal and mantle structure of a SouthAtlantic passive margin — Santos, Campos and Espírito Santo Basins, Brazil. AAPGAnnual Convention, Houston TX, 2011, Extended Abstracts, Search and Discoveryarticle #30177 (2011).
Zhu, L., Kanamori, H., 2000. Moho depth variation in southern California from teleseismicreceiver functions. Journal of Geophysical Research 105 (B2), 2969–2980.
for South America from seismic refraction, receiver functions and sur-/j.tecto.2012.11.014
