access_WPIUseCases - ACCESS-NRI/accessdev-Trac-archive GitHub Wiki
NeCTAR ACCESS simulation and modelling lab use cases
This page gives an outline of the use cases that form part of the work package testing and evaluation. They're intended to cover typical uses of the ACCESS modelling system.
WP1 U1 Reproducing a standard experiment
Goal: Reproduce results from a validated standard experiment
Workflow
Process to reproduce validated standard experiment.
- Researcher searches list of standard atmospheric model experiments in virtual lab documentation and chooses one of interest (e.g. an AMIP experiment with one of the CMiP5 model configurations, ACCESS 1.0 or 1.3).
- Finds selected experiment via model user interface
- Copies experiment and runs test case (model executable and all input data from standard shared locations)
- Checks results against archived output
Details
The standard experiment library is documented at AccessModelExperimentLibrary. See ACCESS_AMIP_testcases for details of this particular test case.
WP1 U2 Modifying a standard experiment
Goal: Modify a standard experiment and analyse impact
Workflow
Process to modify a standard experiment.
- Researcher searches list of standard experiments in virtual lab documentation and chooses one of interest
- Finds selected experiment via model user interface
- Copies experiment
- One or more of the following (in increasing order of sophistication)
- Requests an additional model diagnostic
- Changes a model physics option
- Selects an alternate ancillary file
- Modifies an existing ancillary file
- Checks out model code from repository, modifies code and rebuilds model
- Runs model
- Checks results against archived output and analyses impact of the change
WP1 U3 Testing a climate hypothesis
Goal: Perform a newclimate simulation experiment starting from a validated standard experiment
Workflow
- Researcher forms hypothesis
- Tests using CMIP5 data from ESG
- Thinks of model experiment to further test idea
- Selects ACCESS CMIP5 standard experiment (ACCESS1.0 or 1.3) as a starting point
- Runs test case and checks against archived output
- Modifies code/data appropriately. E.g., one or more of the following possible changes
- Change model forcing (through UI choices or specification of alternate ancillary files)
- Creates a modified branch in code repository
- Uses standard tools to modify ancillary files
- Uses coupled model UI to build model and control experiment
- Runs experiment with data appearing in CMIP5 form in some ESG like catalogue
- Metadata includes details of code branches and ancillary files
- Analysis using same tools as in step 2.
- Publish data to some more permanent storage or go back to step 5
Notes
This is intended to represent a work process that might take several months rather than a single interaction with the CWSLAB.