Creation of Command Line Prototype Modeller's Toolkit

Registered by Graeme Weatherill on 2011-09-08

This is a prototype version of the Modeller's Tookit, which will be operated from the command line with the modelling settings defined using a configuration file. The toolkit will firstly take input data including source geometry, earthquake catalogue, fault data and geodetci strain data. Then it will apply pre-processing steps (e.g. declusting, completeness analysis of catalogue, geometrical definitions of fault etc.) and modelling steps to define completely the parameters needed to define a PSHA source. The toolkit will then output the complete PSHA source model in the nrml format.

Blueprint information

Status:
Started
Approver:
None
Priority:
High
Drafter:
None
Direction:
Approved
Assignee:
None
Definition:
Approved
Series goal:
None
Implementation:
Good progress
Milestone target:
None
Started by
John Tarter on 2011-09-21

Whiteboard

The aim of the Modeller’s Toolkit is to prepare the seismic source model that is needed as an input into PSHA [OpenQuake]. A typical flow would be something like:

1) Read configuration file
2) Load in the geometry of a set of sources (e.g. areaSource, simpleFault, etc.)
3) Load in data from: the earthquake catalogue, geodetic strain etc.
4) Apply pre-processing tools to data - different algorithms can be chosen by the user
5) Apply modelling calculations so that each model has all the essential attributes to define a PSHA source (in the nrml schema) - different modelling algorithms can be chosen by the user
6) Export the model to the nrml format

The user should be able to choose different algorithms, or different settings, for the pre-processing tools and modelling calculations. They may also want some flexibility so they do not necessarily apply every step of the workflow (e.g. the data they upload may be pre-processed already and they may only want to apply the modelling algorithms). The choice of tools and the settings for certain algorithms will need to be set in a configuration file, which will be called inside the program.

For now we can assume the input data are in certain fixed formats, for which we will need to construct parsers (See bugs).

1) Define the source model class

2) Geometry:

For the first version of the MT we assume that the geometry of the source (i.e. area source, fault source etc.) is already defined - we may implement algorithms later on to do this too, but that will be for a later version. To get started it would be useful to have parsers that can read read a fully defined source model in nrml format, as well as a “minimal” format (“skeleton”) which contains only a few key attributes (i.e. geometry, tectonic region, typology) that the user can specify a priori.

Parser 1: Parse full source model from nrml format to object

Parser 2: Parse “skeleton” source model from nrml format to object

Parser 3: Parse “skeleton” source model from shapefile format to object

Parser 4: Parse full source model from shapefile format to object

3) INPUT DATA

a) Earthquake Catalogue

This data will be stored in one of several formats:

   i) Full earthquake catalogue in csv format: This contains all the attributes defined by the Instrumental Catalogue global components (including non-essential ones) to describe an earthquake. An example file is available.

  ii) Earthquake catalogue in reduced csv format: This is another minimal format which contains only the “essential” attributes needed for basic pre-processing and modelling
  iii) Earthquake catalogue in IMS 1.0 (ascii) format (low priority): This format represents how an earthquake is described when a query is performed of the International Seismological Centre web database - an example file is available.

b) Geodetic Strain Data

This information usually defines all the necessary parameters to describe strain at a particular location. A simple csv format is used.

c) Fault Data

This will require a more complex form of input file that can include the geometry as well as attributes that characterise the fault behaviour. As with the zone geometry files, the inputs for fault data could be in one of two formats: xml (preferred) or shapefile. The nrml seismic input that is currently used to define a fault source does not contain all the attributes that are needed to define a “fault source input” for the MT. Missing attributes include: slip, slip uncertainty, shear modulus, displacement-length ratio and magnitude-scaling relation. A new general fault schema will need to be defined for this purpose.

Parser 5: Load in earthquake catalogue in full csv format

Parse 6: Read in earthquake catalogue in “skeleton” csv format

Parser 7: Read in earthquake catalogue in IMS1.0 format (low priority)

Parser 8: Read in strain data in csv format (medium priority)

Parser 9: Read in fault data in xml format (high priority)

Parser 10: Read in fault data in shapefile format (medium priority)

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.