GEM1 seismic hazard branch: completed work and future targets

1
GEM1 seismic hazard branch completed work and
future targets

• Marco Pagani, Laurentiu Danciu, Damiano Monelli,
  Stefan Wiemer, and Domenico Giardini
• Swiss Seismological Service ETH Zürich
• Ned Field, Peter Power
• USGS, Pasadena (CA)

2
Presentation overview

• Objectives of the GEM1 hazard branch
• Recommendations achieved during the Canberra
  kick-off meeting (beginning of March 2009)
• Overview of some tasks were currently working on
  
• Work to be done

3
GEM1 hazard goals
GEM1 is a focused pilot project to generate
GEMs first products and develop GEMs initial IT
infrastructure, GEM1 Implementation plan
January 2009

• IT infrastructure (PSHA engine data storage)
• First compilation of PSHA input models at a
  global scale
• Initial time-independent reference global hazard
  model

4
GEM1 hazard goals (contd)

• Critical points/requirements
• Store of a large amount of information
• Deal with a multitude of input hazard models
  (standardization and flexibility of data models
  and codes are important issues in this context)
• Use of high performance computing resources
  (probably)
• Provide information necessary to perform
  loss/risk analyses

5
Kickoff meeting recommendations

• Input data
• Declustered global earthquake catalog (with
  regionally dependent completeness periods)
• Data necessary to create a global earthquake
  rupture forecast based on the following seismic
  source models source zones, gridded seismicity
  and fault-based.
• A set of representative GMPEs
• Site effects proxy/ies to be used in hazard
  computations (e.g. topography derived Vs,30 model)

6
Kickoff meeting recommendations (contd)

• Methods
• Use at least two codes one conventional code,
  one capable to generate stochastic event sets
• Requirements for hazard codes
• Ability to include different models for different
  regions
• Ability to do time dependent hazard
• Ability to account for local site conditions
• Ability to generate hazard based on smoothed
  seismicity
• Ability to implement logic-trees
• Urgent needs for
• Standardized inputs and outputs
• Parsers for existing model file inputs
• Codes for pre-processing earthquake catalogs and
  create inputs

7
Kickoff meeting recommendations (contd)

• Outputs
• Spinning globe (Google Earth interface or
  similar) zoomable hazard map at a global scale
• Hazard values (Sa(T) 5 damping, Macroseismic
  intensity, PGV and PGA)
• Disaggregation on demand (for magnitude, distance
  and epsilon)
• Ability to generate synthetic catalogs and to
  choose scenario events for risk analysis
• Availability of source models used in hazard
  calculations (first version for internal use
  only)
• For source models used, documentation of origin

8
Kickoff meeting recommendations

• Model uncertainties
• GEM1 may implement regional source models with
  logic tree and associated weights if supplied as
  separate input files
• Aleatoric uncertainties in GMPEs must be included
• Epistemic uncertainties in GMPEs maybe included

9
GEM1 major tasks

• Design and implement the GEM seismic hazard
  computational infrastructure
• Data storage infrastructure
• Seismic hazard computational engine
• PSHA data structure
• Tools for
• Checking the quality and consistency of input
  data
• Plotting hazard input and output information
• Creating input models (e.g. tool assisting in the
  definition of seismic sources and in the
  compilation of logic trees)
• Gather available seismic hazard models and codes
  at a global scale
• Compute a global reference time independent
  seismic hazard model

10
GEM1 major tasks (contd)

• Design and implement the GEM seismic hazard
  computational infrastructure
• Gather available seismic hazard models and codes
  at a global scale
• Currently we have about 10 hazard input models
• Simple models GSHAP East Asia Model (it contains
  more than 400 seismic sources)
• More complex models like the USA NSHMP 2008
• More are coming
• Compute a global reference time independent
  seismic hazard model

11
GEM1 major tasks (contd)

• Design and implement the GEM seismic hazard
  computational infrastructure
• Gather available seismic hazard models and codes
  at a global scale
• Compute a global reference time independent
  seismic hazard model
• Primary goals (within GEM1) are
• To verify the complexity of our calculations and
  to check the needs in terms of computing power
• To test the implemented computational
  infrastructure
• The global hazard map will be patchy
• In this first implementation reliability and
  homogeneity are not strict requirements

12
What we did until now

• Collected seismic hazard data and codes
• Designed the GemSHE GEM Seismic Hazard Engine
  (DB computational infrastructure)
• Defined a standard XML to exchange information
  related to PSHA (draft)
• Hazard computations
• Tests on hazard codes already available
• Tests on input hazard model gathered during the
  initial part of the project
• Tests on possible tools assisting in the creation
  of input hazard models or in the visualization of
  results (e.g. GIS, Google Earth, Databases, DB
  specific modules e.g. PostGIS)

13
Collected PSHA models and codes
?
Gathered about 10 input hazard models (more to
come within the next months) Received all the
requested codes (OpenSHA, CRISIS, EQRM, NZ Hazard
code)
14
What we did until now

• Collected seismic hazard data and codes
• Designed the GEM Seismic Hazard Engine GemSHE
  (DB computational infrastructure)
• Defined a standard XML to exchange information
  related to PSHA (draft)
• Hazard computations
• Tests on hazard codes already available
• Tests on input hazard model gathered during the
  initial part of the project
• Tests on possible tools assisting in the creation
  of input hazard models or in the visualization of
  results (e.g. GIS, Google Earth, Databases, DB
  specific modules e.g. PostGIS)

15
GemSHE seismic hazard engine

• Guidelines adopted for the design
• Based on State of the art PSHA techniques
• Direct connection with the risk/losses component
• Strong interaction with DBs (given the large
  amount of stored data)

16
GemSHE seismic hazard engine (contd)

• Main components
• Global/Regional ERF constructor
• PSHA calculator
• Stochastic event set calculator (using ERF or
  using Hazard Disaggregation)
• Shaking map calculator

17
GemSHE seismic hazard engine (contd)

• Main results provided
• Probabilistic seismic hazard maps
• Hazard curves
• Uniform hazard spectra
• Seismic hazard disaggregation (M-R-e and
  Geographic)
• Stochastic event set (SES)
• Sets of shaking maps related to SES

18
PSHA data storage

• DB schema view

• Input data
• Information relative to the analyses performed
• Intermediate results (e.g. stochastic event sets)
• Output data

19
What we did until now

• Collected seismic hazard data and codes
• Designed the GemSHE GEM Seismic Hazard Engine
  (DB computational infrastructure)
• Defined a standard XML to exchange information
  related to PSHA (draft)
• Hazard computations
• Tests on hazard codes already available
• Tests on input hazard model gathered during the
  initial part of the project
• Tests on possible tools assisting in the creation
  of input hazard models or in the visualization of
  results (e.g. GIS, Google Earth, Databases, DB
  specific modules e.g. PostGIS)

20
GEM1 data exchange format

• A standard XML for PSHA related information
• (a markup language for PSHA data)
• Motivations
• We need to deal with a multitude of different
  models in a standard an homogenized way
• We want to work easily with different PSHA codes
  (at least at the beginning)
• We want to distribute information (e.g. input
  and output models) that is clear and easily
  understandable
• The proposed standard should be appropriate for
• The available PSHA input models and outputs
• The models accepted by the most widely used PSHA
  codes

21
GEM1 data exchange format (contd)
Current situation
PSHA Input model from an external research
institution
GEM1
22
GEM1 data exchange format (contd)
Ideal GEM1 situation

• Data check
• QA check
• Plotting tools

GEM1
GEM outreach meeting 10 June 2009
23
GEM1 data exchange format (contd)

• This is an example of a seismic source
  zone144.10 9.0
• 44.10 9.944.80 9.944.80 9.0
• 5.0 10.0 80.0 180.0 90.0
• 0.05 -1.0 5.0 6.5 0.1

Classic example of a seismic source description
contained in a PSHA code input file

• Pros
• Easy to read by a code (when the format is
  known)
• Cons
• Not flexible
• Not human-readable

24
GEM1 data exchange format (contd)

• ltSeismicSourcegt lt!-- This is an example of a
  seismic source zone --gt ltSeismicSourceGeometrygt
  ltSeismicSourceZonegt ltSeismicityDepthDescriptio
  ngt ltDepthBoundaries seismicityTop"5.0"
  seismicityBottom"10.0" units"km"/gt lt/Seismici
  tyDepthDescriptiongt ltPolygongt ltLocation
  Lat"44.10" Lon"9.0"/gt ltLocation Lat"44.10"
  Lon"9.9"/gt ltLocation Lat"44.80"
  Lon"9.9"/gt ltLocation Lat"44.80"
  Lon"9.0"/gt lt/Polygongt lt/SeismicSourceZonegt
  lt/SeismicSourceGeometrygt ltSeismicSourceSeismicity
  gt lt!-- Faulting style definition
  --gt ltFaultingStylesgt ltSeismicityFocalMechanis
  m dip"80.0" strike"180.0" rake"90.0"/gt ltSeis
  micityMFDescriptiongt ltMFDistGR bGR"-1.0"
  mMin"5.0" mMax"6.5" dM"0.1"/gt lt/SeismicityMF
  Descriptiongt ltSeismicityTDescriptiongt
• ltTPoisson rateOccurence"0.05" timePeriod"1"
  timePeriodUnits"yr"/gt
• lt/SeismicityTDescriptiongt lt/FaultingStylesgt
  lt/SeismicSourceSeismicitygtlt/SeismicSourcegt

• Pros
• Flexible format
• Human-readable ( for those familiar with the
  PSHA jargon)
• Cons
• Relatively easy to read by a code but
  requires an advanced programming knowledge

25
GEM1 data exchange format (contd)

• Were working on XML schemes for the description
  of
• Input data
• Logic trees
• Hazard analysis
• Output data

This is an on-going open project! Currently, our
proposal is under internal review. Any of you
interested in being involved can send me an
e-mail to manifest your interest. In the future
we would like to promote the use of this standard
with the deliver of a suite of software tools
with input PSHA models.
26
What we did until now

• Collected seismic hazard data and codes
• Designed the GemSHE GEM Seismic Hazard Engine
  (DB computational infrastructure)
• Defined a standard XML to exchange information
  related to PSHA (draft)
• Hazard computations
• Tests on hazard codes already available
• Tests on input hazard model gathered during the
  initial part of the project
• Tests on possible tools assisting in the creation
  of input hazard models or in the visualization of
  results (e.g. GIS, Google Earth, Databases, DB
  specific modules e.g. PostGIS)

27
GSHAP model Asia

• Testing a workflow based on the use of
  information stored in a DB (semi-automatic
  approach)
• Definition of the area of interest
• Selection of the input model/s and, consequently,
  of the seismic sources of interest for computing
  the hazard
• Creation of the input file
• Calculation of the hazard
• Plotting of the results

28
GSHAP model Asia (contd)

• Seismic source selection

29
GSHAP model Asia (contd)

• Seismic source selection

30
GSHAP model Asia (contd)

• PGA (10peT50yr)
• GMPE Abrahamson Silva, 1997

31
Smoothed seismicity approach

• Smoothed seismicity approach
• Data ISC catalogue (1964-1999)
• Method triple-S algorithm (Zechar Jordan,
  2009)
• Two-dimensional Gaussian smoothing kernel
  governed by a single length-scale parameter.
• Retrospective forecast experiments are performed
  to determine the optimal value of the
  length-scale parameter.

32
Smoothed seismicity approach (contd)
Annual seismicity rate (for 5Mw9) from INGV
catalog (1901-2005)
PGA (in g) with 10 probability of exceedance in
50 years
33
Work to be completed

• Conclude the first implementation of the GemSHE
  (both in terms of the data storage infrastructure
  and the set-up of the computational engines)
• Improve the standards for data exchange data
  collection
• Hazard computations for different parts of the
  Globe (approx starting in the middle of September)

Munich Outreach Meeting
End of March 2010
September 2009
1 GemSHE
2 Data models
3 Hazard computations
34

• Thank you
• marco.pagani_at_sed.ethz.ch