Disagg Realisations

Produce Disaggregration realizations

flowchart LR
    classDef nshm stroke:lightgreen, stroke-width:3px
    classDef AWS stroke:orange, stroke-width:3px

    RHR[/run disagg realizations/]:::nshm
    runzi:::nshm
    oq[openquake]
    DR[(disagg realizations)]:::AWS

    RHR --> |with| runzi -->|using| oq -->|produces| DR 

NSHM SRM LT has 49 parts (ref NSHM)

For each component branch of the SRM logic tree, site, poE, IMT, VS30:

Run a job with

inputs:

• inversion solution and distributed seismicity rate models (multiple) for the fault systems. i.e. one part of the Source LT)
• which Hazard Model
• Needs Hazard Aggregregation to work out IMTL for requested PoE (352 bytes, via THS query) -> Hazard DB query
• GMCM Logic Tree
• site config: 1 site, poE
• user config: 1 IMT, 1 VS30
• disagg calc config

outputs:

• one HDF5 file, size (TBA): 1 site, poE, IMT, VS30 => 680kB
• CSV files: 80kB
• JSON index of disagg realisations produced

resource/cost/metrics:

In general this job is v similar to hazard realizations, just much bigger.

• 49 * 680kB of realisations per site/VS30/IMT/PoE NSHM: (35,1,5,7) => 850mB

• Currently taking 1 - 4 m (* 1225) on AWS M5 instances 8 CPU.