Difference between revisions of "CyberShake Review Process"

From SCECpedia
Jump to navigationJump to search
Line 39: Line 39:
 
*All seismogram files non empty
 
*All seismogram files non empty
 
*All peak amplitude files non empty
 
*All peak amplitude files non empty
 +
 +
*Check of Processing Stages
 +
**Input Checks
 +
* GF Check
 +
* SRF Check
 +
* SRC Check
 +
* Site List Check
 +
** Output Checks
 +
* SRF
 +
* Seismograms
 +
* Amp files
 +
* Amp db entries
 +
* hazard curves
 +
* Hazard maps
  
 
'''Reasonable Bounds Range Checks:
 
'''Reasonable Bounds Range Checks:

Revision as of 04:23, 6 September 2015

We are defining a CyberShake Review Process that will define a series of verification checks that we perform on a CyberShake result before it is released to public for use.

CyberShake Data Products

Computational products that need review include:

  • Hazard Map (at periods)
  • Hazard Curve (multiple sites)
  • SRFs
  • Rupture Variations
  • Seismograms
  • Amplitudes
  • Hazard Curves

Review Stages

We'll define a series of states that data products must pass through to be released.

  • non started
  • started->incomplete
  • complete->preliminary
  • complete->reviewed
  • reviewed->released
  • released->published

Confirming CyberShake Results

To migrate from complete->preliminary, a result must go through these checks to be

Completion check

  • Individual DAQ’s confirmed as completed by DAGmanager
  • No empty files

DB Consistency

  • NAN check
  • Consistency check confirms integrity of post-processing

Runtime logs:

  • Runtime logs greped for error indocators, must return empty

Checks that are comprehensive:

  • All seismogram files non empty
  • All peak amplitude files non empty
  • Check of Processing Stages
    • Input Checks
  • GF Check
  • SRF Check
  • SRC Check
  • Site List Check
    • Output Checks
  • SRF
  • Seismograms
  • Amp files
  • Amp db entries
  • hazard curves
  • Hazard maps

Reasonable Bounds Range Checks:

  • Comprehensive PGA > 5g

Comprehensive:

  • Alls Hazard curves inspection

Random Samples:

  • 10 (of 400k) Seismograms inspected as “seismic”
  • 1 forward simulation for a randon SRF

Map Evaluation Averaging-Based Factorization comparing new map to

    • National Seismic Hazard Map
    • Previous CyberShake Maps