Skip to content

Source Models

Peter Powers edited this page Aug 10, 2017 · 18 revisions

Sources are representations of earthquakes that occur with some rate. They can be well defined faults with a specific geometry, or point sources derived from historic earthquake catalogs for which planar pseudo-fault representations may be used. In either case, there may be multiple earthquake sizes associated with a given source.

Outline

Source File Structure

An earthquake source model, or HazardModel, is specified using XML files grouped in folders by SourceType. The currently supported source types are:

[ Area, Cluster, Fault, Grid, Interface, Slab, System ]

One level of nested folders is permitted to facilitate logical 'source groups' and the use of different ground motion models (GMMs or model initialization settings. Source files, except those for Fault System Sources, and nested 'source group' folders may be given any name. Ground motions are computed using the models defined in the gmm.xml file in each source directory. If a 'source group' does not contain gmm.xml, one is required in the parent 'source type' folder. See file structure example below.

Sample File Structure:

ModelDirectory/
  ├─ config.json
  ├─ Fault/                           <─── source type
  │    ├─ faultSources1.xml
  │    ├─ faultSources2.xml
  │    ├─ ...
  │    ├─ gmm.xml
  │    │
  │    └─ FaultGroup/                 <─── source group
  │         ├─ otherFaultSources.xml
  │         └─ gmm.xml
  │
  ├─ Grid/
  │    ├─ gridSources.xml
  │    ├─ ...
  │    ├─ gmm.xml                     <─── uses parent GMMs ─┐
  │    │                                                     │
  │    ├─ GridGroup1/                                        │
  │    │    └─ otherGridSources.xml                ──────────┘
  │    │
  │    └─ GridGroup2/
  │         ├─ otherGridSources.xml
  │         └─ gmm.xml
  └─ ...

Common Source Elements

Magnitude-Frequency Distributions (MFDs)

A source requires a description of the sizes and rates of all earthquakes it is capable of generating, otherwise known as a magnitude-frequency distribution. For conciseness, all source files may supply default MFDs so that individual sources need only specify those attributes that are different from the default. Standard MFD types:

XML examples:

<!-- Single magnitude MFD
        - Used to represent the rupture of a specific size event
          on a fault; historically this may have been referred to
          as a 'characteristic' rupture.
        - Can float over or fill source. -->
<IncrementalMfd type="SINGLE" weight="0.25" id="NAME"
    rate="1.0" floats="[true | false]" m="6.5" />

<!-- Gutenberg-Richter MFD
        - Used to represent a range of evenly discretized magnitude
          events with systematically varying rates.
        - Always floats when used for a fault source; never floats
          for grid sources.
        - 'a' is the incremental log10(number of M=0 events). -->
<IncrementalMfd type="GR" weight="0.25" id="NAME"
    a="1.0" b="1.0" dMag="0.1" mMin="5.0" mMax="7.0" />

<!-- Tapered Gutenberg-Richter MFD
        - Same as Gutenberg-Richter, above, but with an exponential
          taper applied to the cumulative number of events with
          seismic moment greater than M.
        - Only used for grid sources. -->
<IncrementalMfd type="GR_TAPER" weight="0.25" id="NAME"
    a="1.0" b="1.0" dMag="0.1" mCut="6.5" mMin="5.0" mMax="7.0" />

<!-- Incremental MFD
        - General purpose MFD that consists of varying magnitudes
          and rates.
        - Can float over or fill source. -->
<IncrementalMfd type="INCR" weight="0.25" id="NAME"
    mags="[5.05, 5.15, ...]" rates="[1.0e-2, 0.9e-2, ...]" 
    floats="[true | false]" />

Rupture Scaling Relations

Fault-based sources commonly require a model that describes the geometry of partial (or floating) ruptures. Likewise, grid sources require a model of fault-length or -area when building pseudo-faults or performing point source distance corrections that consider magnitude-dependent rupture sizes with unkown strike. Such models are composed of published magnitude-length or -area relations and restrictions on aspect-ratio and/or maximum rupture width that are independent of the published model.

See the API Docs for details on specific model combinations and implementations. A rupture scaling model is always sepcified as an attribute of <SourceProperties />; see examples in next section.