Doc/nshm model relational.md #11
Conversation
|
|
||
| The NZ NSHM project publishes **Seismic Hazard Model** versions in the [GNS_Science/nzshm-model]() github repository. | ||
|
|
||
| Models may be Published or Partial. |
There was a problem hiding this comment.
I'd go with Published or Unpublished
| Models may be Published or Partial. | ||
|
|
||
| - Published models are for use by 3rd parties e.g. `NSHM_v1.0.4`. | ||
| - Partial/Unpublished models are typically used for comparisons against complete models, e.g sensitivity tests. They usually are most informative when compared to a published model. |
There was a problem hiding this comment.
"Unpublished models are typically used for scientific experimentation, e.g. sensitivity tests or testing new concepts."
Remove sentence about "usually are most informative . . ."
| class HazardModelRealisation { | ||
| vs30s() integer | ||
| LocationList() string[] | ||
| IMTList() string[] |
There was a problem hiding this comment.
add IMTLList() string[]
we specify both the IMTs (intensity measure types) and the IMTLs (intensity measure type levels)
| tag() string e.g. `Hik TL, N16.5, b0.95, C4, s0.42` | ||
| } | ||
|
|
||
| class SourceLogicTreeSource { |
There was a problem hiding this comment.
I don't like the nomenclature that uses "Source" twice. What if we changed SourceLogicTreeSource to be simply Source? Just like a WalnutTree has Walnuts not WalnutTreeWalnuts
|
|
||
| In `NSHM_v1.0.4` the model contains 4 SourceLogicTreeGroups: one each for Crustal, Hikurangi, Puysegur and Slab sources. | ||
|
|
||
| Each SourceLogicTreeSource is a re-useable arrangment of fault sources, so these are uniquely identified to support reuse in independent **Seismic Hazard Models**. |
There was a problem hiding this comment.
If we follow the above change to Source then this could be changed to:
Each Source is a re-useable arrangement of ruptures (individual earthquake sources), so these are uniquely identified to support reuse in independent Seismic Hazard Models.
|
|
||
| NSHM and many other SeismicHazardModels will include different types of SourceLogicTreeSource model. | ||
|
|
||
| In NZ NSHM we have three ... |
There was a problem hiding this comment.
I think this is too specific and restrictive. We can describe in generic terms that we can combine Source objects or have just a single one to create a SourceLogicTreeBranch. We can use the current model as an example of how sources are split by region and/or TRT and that some have multiple contributions (IFM + DSM) and others are just a DSM
| notes() string | ||
| } | ||
|
|
||
| class SourceLogicTreeGroup { |
There was a problem hiding this comment.
Could we make our terminology less of a mounthful by using SRM instead of Source?
SRMLogicTree
SRMLogicTreeGroup
...
Though, this may then not match with what we've done in nzshm-model (does that matter?)
| - Subduction sources contain a single fault source describing the ruptures that can occur across the subduction fault system. | ||
| - A single Slab source that describes source rates occuring in the interface region (25-) | ||
|
|
||
| We use the NRML (XML) file format for SourceLogicTreeSource sources. This format is required to run hazard calculation in Openquake, and it's well known throughout our user community. |
There was a problem hiding this comment.
This is now a combination of XML and hdf5
| } | ||
|
|
||
| SeismicHazardModel -- "0..*" HazardModelRealisation | ||
| HazardModelRealisation -- "0..*" HazardCurveResult |
There was a problem hiding this comment.
Not sure I follow the difference between HazardModelRealization and HazardCurveResult
|
|
||
| ???+ "Accessing the NSHM hazard realisations" | ||
|
|
||
| The NZ NSHM project makes Hazard Model Realisations available via a graphql web API. There is also a python client library that wraps the web API and also allows the user to cache results locally for better performance in many situations. |
There was a problem hiding this comment.
Is this true? We use THS to store/retrieve component realizations. The composite realizations are not kept (at the moment).
commit WIP on domain model (which includes relational work to date)