South America (SAM)


J. Garcia-Pelaez, M. Pagani, K. Johnson, R. Styron, V. Poggi


The first version of the South America Model (SAM) was developed in the framework of the SARA project (SARA project v1.0, Garcia et al., 2017) funded by Swiss Re Foundation and benefits of the contribution of an important group of South American Institutions:

Here, we present an updated version of the SAM model made by GEM hazard team. This model covers the whole South American continent, with the exception of the Faulkland and the Galapagos Islands. Panama and the northeastern part of the Caribbean have been updated using newer information from the CCARA project, a GEM collaboration project funded by USAID.

Information about the OQ model versions and input files can be found on the Results and Dissemination page.

Interactive Viewer

The viewer below depicts the seismic sources and hazard results in terms of PGA for a return period of 475 years. Click on the menu in the upper right corner to select the layer.


Regional Tectonics

Seismicity in South America primarily results from the subduction of oceanic lithosphere beneath the western margin of the continent and from the eastward translation of the Caribbean plate along the continent's northern margin. In the west, the Nazca plate subducts beneath the central and northern part of the continent, while the Antarctic plate subducts beneath the southern part of the continent. This subduction occurs at the Nazca Trench, and has produced several of the largest earthquakes ever recorded, including the the 1960 Mw 9.4 Valdivia, Chile earthquake and the 2010 Mw 8.8 Maule, Chile earthquake. This subduction has also caused substantial deformation of the upper plate as well, creating the Andes mountains. This deformation is ongoing and distributed throughout the Andean zone, and may be mostly accommodated on large, rapidly-slipping thrust faults in the Subandean Zone at the eastern rangefront of the Andes. However, oblique subduction and gravitational forces result in strike-slip and normal faulting throughout much of the Andes as well.

The Caribbean plate dips south beneath northern South America, though convergence rates are low; this plate boundary may be best characterized as strike-slip, with important left-lateral faults located on the densely-populated Caribbean coast of Colombia and Venezuela. Offshore eastern Venezuela, the Caribbean plate terminates and Caribbean-South American relative motion is accommodated on the Lesser Antilles subduction zone in the Atlantic Ocean.

To the east and south of the Andes, the South American continent may be considered stable. Nonetheless, like many stable continental regions, faults do exist--many of these date to the assembly and breakup of the supercontinent Pangea hundreds of millions of years ago. Stresses transmitted through the cold, strong South American shield produce some earthquakes in Brazil and the surrounding nations, though the rates of seismicity in central and eastern South America is a small fraction of the western and northern active margins.

Basic Datasets

The South American seismic model was created using a compilation of updated and harmonised databases needed for PSHA (e.g. historical and instrumental catalogue, a compilation of active fault data, a database of strong motions recordings) which were created following, to the greatest extent possible, common standards and transparent procedures. The construction of these databases was completed mainly by a number of South American scientists and engineers. In the next lines we briefly describe the main Datasets created in the framework of the SARA project and how some of them those were recently updated.

Earthquake Catalogues

The availability of a homogeneous earthquake catalogue is a fundamental requirement for any seismic hazard analysis. This updated version of the SARA model (called SAM), benefits of the first version of the catalogue obtained in the framework of the SARA project (SARA catalogue v1.0), where, using information from a wide collection of earthquake databases and scientific publications, two groups of scientifics, each focusing on differents eras (historical-pre-instrumental and instrumental), created a parametric harmonized earthquake catalogue to be used in PSHA for the South America region.

However, for this new version, even when the procedure to obtain the catalogue follows a similar approach (Weatherill et al., 2016; Garcia et al., 2017; Poggi et al., 2016), we introduced some modifications:

  • The historical part finished in 1906, then from the historical catalogue created in the framework of SARA we used only those events reported before this date,
  • As a consecuence, now the period between 1906 and 1964 is treated in the same way than the instrumental part in Garcia et al., 2017,
  • Newer or updated information was added, in particular from the local southamerican agencies.

The final updated (and harmonized) version of the catalogue is presented below (a- non declustered, b- declustered). This version, as the previous ones, covers the whole South America region, part of Central America (Panama) and the Caribbean (Trinidad and Tobago). The non-declustered version consists of 106798 events from 1513 to 2017 and in the magnitude range 3.0 ≤ Mw ≤ 9.6. The harmonized catalogue was then purged from fore- and aftershock sequences and possible seismic swards, using the Gardner and Knopoff (1974) declustering algorithm and a spatial-time Uhrhammer (1985) window (see details in Characterization and processing of seismic catalogs) This version contains 56220 events from the same period and magnitude range.

SAM18 catalogue (un-declustered) SAM18 catalogue (declustered)
SAM18 cata undec SAM18 cata dec
N. events (Mw ≥ 3.0): 106798 N. events (Mw ≥ 3.0): 56220

We also used the Global Centroid Moment Tensor (GCMT) focal mechanisms from 1976-2017 (Dziewonski et al., 1981; Ekström et al., 2012). This catalogue was used in the characterization of the sources as well as the definition of the geometry of whole model.

Seismotectonic Classification

The classification of seismicity taking into account its possible tectonic environment is a crucial and fundamental step in the development of a seismic hazard model following the (GEM standards methodologies).The classification consists in the application of an automatic procedure, which is capable to assign a tectonic class to each earthquake in the catalogue, following criteria defined a priori by the hazard modeller and using the main tectonic settings of the region were the model is developed. Specific details about this procedures can be found (here). The results of the classification are presented below (a: shallow, b: interface, in-Slab and deep main events).

SAM18 tectonic classification (shallow) SAM18 tectonic classification (interface + in-slab + deep)
SAM18 shallow SAM18 no shallow

Fault Database

A group of South American geologists undertook the complex and challenging task of assimilating and homogenising active fault data across the South American region. Data came from many published sources including the International Lithosphere Program II-2 project (Costa et al., 2003), the Multinational Andean Project (Getsinger and Hickson, 2000) and existing national neotectonic databases and fault data collections. Quaternary active structures known up to present in the region, particularly considering their possible seismogenic relevance, have been compiled under common criteria, building a harmonised database. Due to the fact that data availability and accuracy is very heterogeneous across the region, harmonisation efforts on selecting priority faults were needed.

The following alternative selection criteria were applied:

  • Slip rates equal or larger than 0.1 mm/yr (this applies to rates already available or estimated by compilers using quantitative-semiquantitative criteria)
  • Evidence of Late Pleistocene tectonic activity (confirmed and suspected)
  • Confirmed or suspected sources of earthquakes with magnitudes greater than Mw 5.5.

The first version of the fault database can be seen below (v. 2016). It was created in the framework of the SARA project (Garcia et al., 2017, Costa et al., 2016), and it contained XXX neotectonic structures. Around 35% of the overall faults collected had enough information to properly characterize their geometry and activity and only 486 of them satisfied the selection criteria. Additional investigations was necessary in order to properly characterise the faults compiled in this first version. During the last year, with the contribution of local experts and the GEM hazard team, it was possible to obtain an improved version (v. 2018) of this database (see figure below), which it also benefits of compilations obtained during currently GEM collborations projects (GEM/CGS collaboration projects) and recent national efforts (Yepez et al., 2017).

SARA active fault database (v. 2016) SARA active fault database (v. 2018)
SARA fault database SARA fault database, version 2018

Ground Motion Database

In the framework of SARA project the creation of the first database of ground motion recordings for the South America was possible thanks to the contribution of a group of scientists from Colombia, Bolivia, Brazil, Chile, Ecuador and Venezuela. This database (see Figures below) contains about 2100 records for the different tectonic regimes present in South America (ACC: active shallow crust, SCC: stable continental crust, Subduction: interface and in-slab and Deep: Bucaramanga nest), all of them processed following the schema proposed by Boore et al. 2012 and organized according to Weatherill (2014).

SARA Ground motion database SARA Ground motion database
Events Stations
a) events classified according its tectonic context b) stations


To build the model and manage the basic information and datasets described above, we used diverse GEM Toolkits (Hazard Modelled Toolkit, HMTK; Subduction Toolkit; Ground Motion Toolkit, GMPE-SMTK and Model Building Toolkit, MBT).

Seismic Source Characterisation

The SAM Seismic hazard model (SHM) was divided into three main components:

  1. the shallow seismicity, modelled using an integrated model of distributed seismicity (kernel based grid sources) and "simple" crustal fault sources,
  2. the subduction interface seismicity, modelled as large "complex" fault sources with a 3D geometry,
  3. the subduction in-slab and deep seismicity, modelled as nonparametric ruptures.

To know about GEM modus operandi or methodologies/approaches used to build the different components of a SHM, please see here. In the following sections we briefly describe the main characteristics of each component.

Shallow seismicity: Distributed Seismicity

This component covers both, the active and stable shallow crust in South America. In the fisrt version of the SAM model developed in SARA, the distributed seismicity model was created using a number of polygons (or volumes) delineating regions with homogeneous temporal and spatial characteristics of seismicity, tectonic and geodynamic setting.

The catalogue was the main resource used to characterize the sources and for each of them was defined:

  • the maximum magnitude (adding an increment of 0.5 to the largest observed event),
  • the seismogenic thickness (following the procedure implemented to classify the seismicity),
  • the focal depth distribution of events and its probability (histograms of hypocentral depth distribution),
  • the orientation and faulting styles of ruptures (focal mechanisms distribution and fault geometry characterization),
  • the activity rate or magnitude frequency distribution, MFD (represented by a truncated Gutenberg-Richter distribution assuming a magnitude binning of 0.1, a minimum magnitude around 4.0; and b and a values computed using the maximum likelhood estimator (Wiechert, 1980).

For this version we decided to model in a different manner this component. The distributed seismicity model is based on the previous ones, but the number of sources was reduced by merging a few zones with the same or similar seismotectonic settings, which certainly improved the source characterization (e.g. robust MFDs). The characterization of the sources was made in the same way, but for this version we combined the area sources with a smoothed based kernel methodology as in (Frankel, 1995). The resulting sources are now gridded sources (spacing ~10 kms) with a constant b value and a spatially variable a predicted by the smoothed Gaussian kernel used. In addition, we adjusted the grid points around the fault sources to avoid a double counting of the ocurrences. In the interactive viewer can be seen the shallow sources for both, the active and stable shallow regions.

Shallow seismicity: Crustal fault

The methodology implemented to model a simple fault can be see in Characterizing and modeling fault sources. The most important issues to be noted here are the assumptions used to build the model:

  • The occurrence of the events on the fault follows a GR distribution, and the total seismic moment rate from the magnitude frequency distribution equals the geological moment rate derived from the fault dimension and slip rate,
  • The b-value on the fault is the same as the b-value of the area source within which the fault is located, and,
  • For each fault a lower and upper-bound magnitudes were assigned.

Occurrence rates on faults are in general modelled either following a characteristic, a truncated Gutenberg Richter model or a combination of the two; we opted for the use of a truncated Gutenberg-Richter (GR) model (e.g. Youngs and Coppersmith, 1985).

The lower bound assigned is around M6.5 since the small magnitude earthquakes are modelled using distributed seismicity sources, whilst the upper-bound magnitude is constrained by the fault dimension (length and downgoing wide) and the scaling relation used to compute this value (Leonard, 2010, 2014). In the interactive viewer can be seen the fault sources considered for both, the active and stable regions.


Source geometry

The subduction source geometries are built using the GEM Subduction Toolkit.

We segment each subduction zone or subduction-style seismicity zone accodring to past megathrust earthquakes, current seismicity patterns, trench convergence rates and kinematics, and assistance from thorough structural and tectonic regional studies. The sources are built such that ruptures do not propagate across the defined boundaries. We include the Nazca subduction zone, the North Panama deformation belt, and the Lesser Antilles subduction zone.

Nazca Subduction zone

We test multiple segmentations, and ultimately choose the following eight-segment model, described from south to north (note that use of countries in segment geographical descriptions is approximate):

  • Segment 1 (Southern Chile): Extends from the southernmost slab to the projection of the Juan Fernandez ridge to the trench, and is close to that proposed by Salliard et al. (2017). This segment hosted both the 1960 MW 9.6 Valdivia and 2010 MW 8.8 Maule earthquakes.

  • Segment 2 (Northern Chile): Terminates approximatively where the Iquique ridge intersects the trench. This limit agrees with Saillard et al. (2017) and Villegas-Lanza et al. (2016), as well as interface segmentation for the 2017 USGS seismic hazard model of Petersen et al. (2018).

  • Segment 3 (Southern Peru): Terminates at the Nazca ridge, and is also used as a segment boundary by Villegas-Lanza et al. (2016) and Brizzi et al. (2018). The ridge itself is a low-coupling region along the interface (Brizzi et al., 2018).

  • Segment 4 (Central Peru): Terminates at the Mendana Fracture Zone, where the crustal age changes and seismic coupling decreases (Villegas-Lanza et al., 2016).

  • Segment 5 (Northern Peru): Terminates north of the approximate intersection of an oblique strand of the Viru Fracture Zone with the trench, where seismic coupling further decreases (Villegas-Lanza et al., 2016).

  • Segment 6 (Southern Ecuador): Terminates at the bouyant Carnegie Ridge intersection with the trench, close to a limit used in the highly segmented model by Beauval et al., (2018).

  • Segment 7 (Northern Ecuador/Southern Colombia): Terminates at the spreading center just north of Malpelo ridge. Approximately coincident with the northern boundary of subduction used by Petersen et al. (2018).

  • Segment 8 (Northern Columbia): Extends to the northern limit of the Nazca plate.

Segments 2-8 are assigned a locking depth of 50 km, representing the slip observed in many finite rupture models of the Finite-Source Rupture Model Database. For Segment 1 extends to 60 km, consistent with the Maule earthquake, and to yielding fault areas large enough to compute MW 9.6 using subduction geometry scaling relationships (Strasser, 2010; Thingbaijam and Mai, 2017; and Allen and Hayes, 2017) These locking depths are used as the cutoff depth between interface and subduction sources.

North Panama

For the North Panama Deformation Belt, we created subduction-style geometry for a single surface, following the same procedure as for the western South America Nazca subduction zone. At this convergent boundary, there is little shallow seismicity, and so we create only a slab model from six profiles spaced at 100 km (see next section).

Lesser Antilles The Lesser Antilles subdction sources was created in the framework of the CCARA project.


Interface rates

We derive a magnitude-frequency distribution (MFD) for each interface segment using a hybrid approach that combines statistics from observed seismicity with a characteristic component derived from tectonics. Source characteristics with references are summarized in Table 1. The listed characteristic magnitude (Mchar) is the median magnitude computed from the scaling relationship Thingbaijam and Mai (2017).

Subduction zone Segment a-Value b-Value Mmax,obs Mchar Convergence rate (mm/yr) Coupling
Nazca 1 4.26 0.83 9.6 9.17 68 0.80
2 4.56 0.76 8.31 8.99 79 0.95
3 5.66 1.03 8.69 8.95 62 0.75
4 5.18 1.01 8.40 8.67 61 0.80
5 4.48 0.88 7.85 8.62 58 0.55
6 5.90 1.11 7.32 8.73 58 0.55
7 3.12 0.64 8.75 8.69 58 0.55
8 3.21 0.70 7.10 8.12 58 0.55

The Lesser Antilles subduction zone is discussed in the model for the Carribean and Central America.

In-Slab rates

In our source model, segmentation boundaries from the interface extend into the slab, and are not meant to suggest barrier to rupture within the downgoing slab volume, but instead to allow spatial variability in observed seimsicity while still using non-parametric ruptures. We model the rupture geometries and rates following the standard GEM methodology for slab earthquakes, described here for the range M6.5-8.5. Results are in Table 2.

Subduction zone Segment a-Value b-Value
Nazca 1 5.955 1.126
2 6.733 1.115
3 5.802 0.954
4 4.758 0.882
5 3.304 0.676
6 3.287 0.668
7 4.568 0.828

Ground Motion Characterisation

The harmonized database of earthquake recordings (see Figure above), which covers the major tectonic regions found in South America (subduction interface, subduction in-slab, active shallow crustal and stable continental) was used for the selection of suitable ground motion prediction equations for application in the different tectonic regions of South America, from the OpenQuake-engine’s extensive library of GMPEs.

In the table below is presented the ground motion model (and weigths) obtained from the selection procedure for the different tectonic regions considered for the firts (SARA) version of the SAM model.

Active Shallow Crust Weight Stable Shallow Crust Weight
Akkar et al. (2014) 0.33 Atkinson and Boore (2006) 0.25
Bindi et al (2014) 0.33 Tavakoli and Pezeshk (2005) 0.50
Boore et al. (2014) 0.34 Drouet (Unpublished) 0.25
For the Lesser Antilles (LAN) region 2018 Selection
Abrahamson et al. (2014) 0.33 Atkinson and Boore (2011) 0.33
Akkar et al. (2014) 0.33 Pezeshk et al. (2011) 0.33
Cauzzi et al. (2014) 0.34 Silva et al. (2002) 0.34
Subduction Interface Weight Subduction In-slab Weight
Zhao et al. (2006) 0.33 Abrahamson et al. (2015) 0.50
Abrahamson et al. (2015) 0.33 Montalva et al. (2015) 0.50
Montalva et al. (2015) 0.34

For the second version of the SAM model some modifications were needed. In particular, we modified the ground motion model for the Stable Shallow Crust region (see GMPEs in bold above) and added a new model for the Subduction sources in the Lesser Antilles region (see below):

Subduction Interface Weight Subduction In-slab Weight
Zhao et al. (2006) 0.34 Abrahamson et al. (2015) 0.50
Abrahamson et al. (2015) 0.33 [Montalva et al. (2015)][mo2015s] 0.50
Youngs et al. (1997) 0.33


Hazard curves were computed with the OQ engine for peak ground acceleration (PGA) and spectral acceleration (SA) at 0.2s, 0.5s, 1.0s, and 2s. The computation was performed on a grid of approximately 10 km-spacing with reference soil conditions corresponding to a shear wave velocity in the upper 30 meters (Vs30) of 760-800 m/s.

The hazard map for PGA corresponding to a 10% probability of exceedance in 50 years (475 year return period), can be seen using the interactive viewer. For a more comprehensive set of hazard and risk results, please see the GEM Visualization Tools.


Abrahamson N., N. Gregor and K. Addo (2015). BC Hydro Ground Motion Prediction Equations For Subduction Earthquakes Earthquake Spectra, in press

Akkar S., M. A. Sandikkaya, and J. J. Bommer (2014). Empirical Ground-Motion Models for Point- and Extended- Source Crustal Earthquake Scenarios in Europe and the Middle East, Bulletin of Earthquake Engineering (2014), 12(1): 359 - 387

Atkinson Gail M. and David M. Boore (2006). Earthquake Ground-Motion Prediction Equations for Eastern North America; Bulletin of the Seismological Society of America, Volume 96, No. 6, pages 2181-2205

Bindi D., M. Massa, L.Luzi, G. Ameri, F. Pacor, R.Puglia and P. Augliera (2014). Pan-European ground motion prediction equations for the average horizontal component of PGA, PGV and 5 %-damped PSA at spectral periods of up to 3.0 s using the RESORCE dataset, Bulletin of Earthquake Engineering, 12(1), 391 - 340

Boore David M., Jonathan P. Stewart, Emel Seyhan and Gail Atkinson (2014). NGA-West2 Equations for Predicting PGA, PGV, nd 5 % Damped PGA for Shallow Crustal Earthquakes; Earthquake Spectra, Volume 30, No. 3, pages 1057 - 1085.

Boore, D.M., Azari Sisi, A. and Akkar, S. (2012). Using Pad-Stripped Acausally Filtered Strong-Motion Data, BSSA 102(2), 751-760.

Drouet S. (2015). Unpublished for Brazil based on the method described in Douet & Cotton (2015)

Drouet, S., Cotton, F. (2015): Regional Stochastic GMPEs in Low‐Seismicity Areas: Scaling and Aleatory Variability Analysis—Application to the French Alps. - Bulletin of the Seismological Society of America, 105, 4, pp. 1883—1902.

GEM/CGS collaboration projects

Carlos C., Garcia J., Alvarado A., Audemard F., Audin L., Benavente C., Bezerra F. H., Cembrano J., González G., López M., Minaya E., Paolini M., Pérez I., Santibanez I., Arcila M., Delgado F. and Pagani M. (2016). From neotectonic data to seismogenic sources in South America: Results and lessons learned from the SARA project, Proceeding of 35th General Assembly of the European Seismological Commission, ESC2016-479, Trieste, Italy, 2016

Costa, C., H. Cisneros, M. M. Machette, R. L. Dart, (2003). A new database of Quaternary faults and folds in South America. ILP Task Group II-2 (western Hemisphere). Proceedings of the A. G. U., Fall Meeting 2003.

Dziewonski, A. M., T.-A. Chou and J. H. Woodhouse, Determination of earthquake source parameters from waveform data for studies of global and regional seismicity, J. Geophys. Res., 86, 2825-2852, 1981. doi:10.1029/JB086iB04p02825

Garcia et al., 2017

Gardner, J. K., and L. Knopoff (1974). Is the sequence of earthquakes in southern California, with aftershocks removed, Poissonian? Bull. Seismol. Soc. Am. 64: 1363–1367.

Leonard, 2010, 2014

Montalva et al. (2015). Unpublished, adaptation of the Abrahamson et al. (2015) BC Hydro GMPE, calibrated to Chilean strong motion data.

Tavakoli B. and S. Pezeshk in 2005 and published as “Empirical-Stochastic Ground-Motion Prediction for Eastern North America” (2005, Bull. Seism. Soc. Am., Volume 95, No. 6, pages 2283-2296).

Uhrhammer, R. (1986). Characteristics of Northern and Central California Seismicity, Earthquake Notes, 57(1): 21

Yepez et al., 2017

Youngs, R. R. and K. J. Coppersmith (1985). “Implications of fault slip rates and eaarthquake recurrence models to probabilistic seismic hazard estimates”. In: Bull. Seism. Soc. Am. 75.4, pages 939–964

Weatherill, G. A. (2014). OpenQuake Ground Motion Toolkit - User Guide. Global EarthquakeModel (GEM). Technical Report.

Weichert, D. H. (1980). Estimation of the Earthquake Recurrance Parameters for Unequal Observation Periods for Different Magnitudes. In: Bulletin of the Seismological Society of America 70.4, pages 1337 –1346

Zhao, J. X., Zhang, J., Asano, A., Ohno, Y., Oouchi, T., Takahashi, T., Ogawa, H., Irikura, K., Thio, H. K., Somerville, P. G., Fukushima, Y., & Fukushima, Y. (2006). Attenuation relations of strong ground motion in Japan using site classification based on predominant period. Bulletin of the Seismological Society of America, 96(3), 898–913.