This document summarizes the hazard component of the Caribbean and Central America Risk Assessment (CCARA) project. The hazard component aims to collect and homogenize earthquake data and develop a new probabilistic seismic hazard model for the region. Key aspects summarized include compiling catalogues of earthquake data, developing an active fault database, creating a ground motion database, developing source models for shallow and subduction seismicity, and selecting appropriate ground motion models. The new hazard model will provide a synthesis of current scientific knowledge on earthquake hazards in the region.

1. Caribbean and Central
America Risk Assessment:
hazard component
GEM Hazard Team
hazard@globalquakemodel.org
CCARA final meeting
March 21th, 2018 – Santo Domingo, Dominican Republic

2. This presentation is made possible by the generous
support of the American people through the United
States Agency for International Development (USAID).
The contents are the responsibility of the GEM
Foundation and do not necessarily reflect the views of
USAID or the United States Government.

3. Why seismic hazard [and risk]?
3

4. Seismic Hazard Analysis: definition and goals
- The goal of seismic hazard analysis
is to compute for a specific site
the expected level of shaking
within a given interval of time.
- One of the principal uses of
hazard results are:
- The design of buildings and
infrastructures. For residential
buildings, seismic hazard
results are incorporated into
building provisions. For special
buildings and critical facilities,
hazard analysis can be
performed ad-hoc.
- To perform risk analysis and
calculation of the expected
losses
4

5. CCARA project: hazard component
Past PSHA studies

6. Recent hazard models in the region
- AA
Bozzoni et
al. (2011)
Frankel et al.
(2011)
Is another model really needed?
YES!
Muller et al.
(2003)
Salazar et al.
(2014)

7. Why another model?
- Datasets used for the construction of most of these
hazard models are only partially available (= lack of
reproducibility)
- The models are inhomogeneous (i.e. they are based on
substantially different modelling choices) hence do not
produce consistent hazard results
- Over the past years new achievements in monitoring,
understanding and modelling tectonics in the region
- In some countries the seismic hazard models use for the
definition of seismic actions in building codes are
outdated or completely missing (i.e. Haiti)
- In order to compute homogenously risk in the region, an
harmonized hazard model is required

8. CCARA project – Hazard Component
The activities under the hazard component will be
organized into five main modules:
1. Compilation of earthquake related catalogues
2. Active faults and tectonic databases
3. Ground motion database and hazard ground motion
model
4. Source model building and hazard calculation
The primary goal of the hazard component is to collect
and homogenize the current earthquake knowledge into
a set of datasets and to build a new hazard model, which
will constitute a synthesis of the current best scientific
knowledge across the region.

9. Methodological aspects - PSHA
A probabilistic seismic hazard model generally contains
two parts :
- Seismic source model
- Ground motion model

10. CCARA project: hazard component
Regional Datasets

11. CCARA Hazard Component – Earthquake Catalogue
Appraisal of available informationSpatial coverage of existing regional and national
catalogues

12. CCARA Hazard Component – Earthquake Catalogue
CCARA catalogue v.1.0
CCARA harmonized catalogue (v.01)
The catalogue contains 81538 events spanning 3.0 ≤ Mw ≤ 8.1 for the
period 1500 – 2016. The catalogue contains 81538 events spanning 3.0 ≤ Mw ≤
8.1 for the period 1500 – 2016.

13. CCARA Hazard Component – FMS Catalogue
CCARA fms catalogue v.1.0
The fms catalogue contains 2580 events (3429 solutions)
spanning 3.0 ≤ Mw ≤ 8.1 for the period 1500 – 2016.
Pag. 9 CCARA focal mechanism database
del San Salvador”
INET 1 “Instituto Nicaragüense de Estudios Territoriales” Nicaragua
In Figure 4 is presented a map of the focal mechanism distribution in
the study area using the ISC database.
Figure 4 – Focal mechanism solutions in the CCARA database after 1976. All solutions are
plotted following the Harvard CMT convention (best double couple defined from the
seismic moment tensor, SMT). The beach balls are coloured according the Alvarez-Gomes
(2009) classification presented in Figure 5.
base.
he CCARA database after 1976. All solutions are
ention (best double couple defined from the
balls are coloured according the Alvarez-Gomes
.

14. Catalogue tectonic regionalization
Mexico Panama
Lesser Antilles

15. CCARA Hazard Component – ASC fault database
All this information have been updated, homogenised
and expanded (when it is possible), within a project led
by GEM and supported by the U.S. Agency for
International Development (USAID)
We have compiled a database of active faults to perform a
fault-based PSHA
The database is released as open-source, under a Creative
Commons Attribution license (CC-BY 4.0)
Detailed information in:
https://github.com/GEMScienceTools/central_am_carib_faults
https://blogs.openquake.org/hazard/2017/05/31/c-am-car-faults/

16. CCARA Hazard Component – ASC fault database
540 faults traces, with contributions from Colombia,
Venezuela (SARA project) and Mexico (Villegas et al., 2017)
datasets

17. CCARA Hazard Component – GM database
Mexico (UNAM
and CICESE)
Costa Rica
French Antilles
El Salvador
Colombia
Nicaragua

18. CCARA Hazard Component – GM database
Mexico (UNAM
and CICESE)
French Antilles
El Salvador
Colombia

19. CCARA Hazard Component – GM database
Negative residuals
observations < models

20. CCARA project: seismic hazard model
Seismic source model

21. Earthquakes types in the CCARA region
- Intra-plate earthquakes in active regions
- Intra-plate earthquakes in stable regions
- Subduction Earthquakes
- Mid-ridge earthquakes
- Deep earthquakes in non-subduction regions
- Volcanic earthquakes

22. CCARA Hazard Component – PSHA model
a) shallow seismicity
• distributed
seismicity
• crustal fault
sources
b) subduction
interface
c) subduction in-slab
seismicity
Model components:

23. CCARA project: seismic hazard model
Shallow model

24. Methodological aspects - PSHA
A probabilistic seismic hazard model generally contains
two parts :
- Seismic source model
- Ground motion model

25. Distributed seismicity
Area sources based on:
o Previous PSHA studies
o Seismicity
o Surface Faults
o Plate boundaries
o Tectonic - Strain models

26. Catalogue processing and source definition
Declustering
Completeness
MFD
Mobs + 0.3
Mmax
42 independent area sources
with homogenous temporal
and spatial characteristics of
seismicity, tectonic and
geodynamic setting.

27. Catalogue processing and source definition
Strike distribution
Faults
Focal mechanims
Style of faulting (rake)
Depth distribution

28. Shallow fault source model
Red : fault traces
Green: their surface projection

29. CCARA project: seismic hazard model
Subduction model

30. Subduction
Stubailo et al., (2012)
Stern (2002)
pter 4. Estimation / constrain of some parameters for the construction of PSHA
el in South America. 109
Yeats (2012)

31. Subduction geometry: top of the slab

32. Subduction geometry: top of the slab [CAM]
Mexico
Panama

33. Subduction geometry: interface/in-slab
SouthNorth

34. Subduction Inslab ruptures: CAM
Mexico
Panama

35. Characterization of subduction sources
- Characterization of occurrence
+ Using past seismicity (instrumental,
historical and paleo-seismicity)
+ Using information from tectonics and
geodesy (future versions)
MFD: Double Truncated GR
+ (future versions we will also consider
combinations of characteristic and GR
distribution)
Mmax: based on global datasets and
local information
Completeness
MFD

36. CCARA project: hazard component
Ground motion model

37. Methodological aspects - PSHA
A probabilistic seismic hazard model generally contains
two parts :
- Seismic source model
- Ground motion model

38. Pre-selection
We preselected a small set of models using criteria
published in the recent scientific literature (Cotton et al.,
2006; Bommer et al., 2010).
No. Cotton et al., (2006)
1 The model is clearly from an irrelevant
tectonic regime
2 The model is not published in an
international peer-reviewed journal
3 The documentation of the model and its
underlying data set is insufficient
4 The model has been superseded by more
recent publications
5 The frequency range of the model is not
appropriate for engineering application
6 The model has an inappropriate functional
form
7 The regression method or regression
coefficients are judged to be inappropriate
Siterupture

39. ACTIVE SHALLOW
Model Abrev Area Mag range Dist range R metric Sites Notes
AbrahamsonEtAl2014 A14 NGAW2 3.0-8.5 0-300 Rrup Vs30 hanging-wall model
AkkarEtAlRjb2014 A10 RESORCE (Europe + ME) 4.0-8.0 0-200 Rjb Vs30
BooreEtAl2014 B14 NGAW2 3.3-7.0 0-400 Rjb Vs30 no hanging wall model
CauzziEtAl2014 C14 Global; 77% from Japan 4.5-7.9 0-150 Rrup Vs30
ChiouYoungs2014 CY14 NGAW2 3.5-8.0 0-300 Rrup Vs30 hanging-wall model
ZhaoEtAl2006Asc Z06 95% from Japan 5.0-8.3 0-300 Rrup 5 classes high GM at long periods, [R]
INTERFACE
Model Abrev Area Mag range Dist range R metric Sites Notes
AbrahamsonEtAl2015SInter A15 Global 6.0-8.4 5-551 Rrup Vs30 fast decay
AtkinsonBoore2003SInter AB03 Global 5.5-8.3 5-420 Rrup 4 classes slow decay
Kanno2006Shallow K06 Japan 5.2-8.2 1-400 Vs30 (crustal +sub) <30km D&M09 find this GMPE
best for LA
LinLee2008SInter LL08 Mostly Taiwanese 5.3-8.1 20-300 Rhypo 2 classes Vs30 only distinguishes rock vs soil
McVerry2006SInter MV06 New Zealand + some foreign 5.25 - 7.5 6-400 Rrup/Rvol 3 classes
YoungsEtAl1997SInter Y97 Global 5.0-8.2 8.5-550 Rrup 2 classes [R]
ZhaoEtAl2006SInter Z06 95% from Japan 5.0-8.3 0-300 Rrup 5 classes high near field GM M>8.5
INTRASLAB
Model Abrev Area Mag range Dist range R metric Sites Notes
AbrahamsonEtAl2015SSlab A15 Global 5.0-7.9 34-991 Rhypo Vs30 fast decay
AtkinsonBoore2003SSlab AB03 Global 5.0-7.9 34-300 Rrup 4 classes
GarciaEtAl2005SSlab G05 Central Mexico 5.2-7.4 40-400 Rrup, Rhypo 1 class derived only for NEHRP B sites
Kanno2006Deep K06 Japan 5.5-8.0 30-500 Vs30 30-180km; D&M09 find this GMPE best for LA
LinLee2008SSlab LL08 Mostly Taiwanese 4.1-6.7 40-600 Rhypo 2 classes Vs30 only distinguishes rock vs soil
McVerry2006SSlab MV06 New Zealand + some foreign 5.25 - 7.5 6-400 Rrup/Rvol 3 classes
YoungsEtAl1997SSlab Y97 Global 5.0-7.8 45-774 Rrup 2 classes [R]
ZhaoEtAl2006SSlab Z06 95% from Japan 5.0-8.3 0-300 Rrup 5 classes [R]
Table of pre-selected GMPEs (highlighted ones used)

40. CCARA project: hazard component
Results

41. Hazard calculations
- 1 Source model
- 27 End-Branches
• 9 Intensity Measure Types [PGA,
SAs: 0.1, 0.2, 0.3, 0.5, 0.7, 1.0, 1.5,
2.0 sec]
- 4 Return Periods [475 to
20000 yr]
- Mean, Median and Quantiles
- Grid [0.1˚ × 0.1˚ spacing]
- Site condition: rock [760 m/s]
‣ Hazard Maps
‣ Hazard Curves
‣ Uniform Hazard Spectra
‣ Disaggregation: selected
cities

42. OpenQuake
Catalogue Toolkit
OpenQuake
Earthquake Source
Toolkit a.k.a hmtk
OpenQuake
Ground Motion
Toolkit
OpenQuake Risk
Modeller’s Toolkit
OpenQuake
Platform
OpenQuake engine
OpenQuake IRMT
(experimental)
OpenQuake
Subduction Toolkit
OpenQuake
Model Building
Toolkit

43. PSHA 475 yr [PGA]

44. PSHA results

45. PSHA results

46. A community-based hazard model for the region

47. A community-based hazard model for the region

48. Model building process
- Involved the local scientific community since the
beginning (representatives from various countries in
Central America and the Caribbean)
- One external reviewer (Prof. M. B. Benito – UPM –
Spain)
- Shared tasks with local experts (ground-motion
database construction)
- Hold three workshops
+ Kick-off workshop
+ Mid-term review meeting and OQ-engine training (1 week in
San Jose, Costa Rica)
+ Final review meeting (done yesterday)

50. CCARA project: hazard component
Lesson learned and next steps

51. Building models with (and not for) communities
- Building relationships and trust across boundaries on a
global/regional scale takes time
- Local ownership and involvement is essential
- Uniformity of models requires harmonisation of data,
tools, methods and procedures first
- Procedures and methods should be transparent +
reproducible
- Global datasets need to be integrated with local data

52. SGC collaboration The SARA regional model is
currently used as the backbone
for the new national hazard
model for Colombia. Will be
used to update building code
and calculate risk at national
level

53. SGC collaboration

54. Please attribute to the GEM Foundation with a link to
www.globalquakemodel.org
Except where otherwise noted, this work is licensed under
https://creativecommons.org/licenses/by-nc-nd/3.0/

55. PSHA 475 yr [PGA]