Run a GFDL climate experiment with the Earth System Curator interface.
Author: Julien Chastang
Date: August 23, 2006
Goals
Discover how ESC harvests metadata while a user runs a climate experiment through ESC interface.
Actors
Jane Modeler
ESC front-end interface.
Trigger
Jane wants to run a GFDL climate experiment through ESC interface so that her experiment may be recorded, stored, and shared with other climate scientists.
Summary
Jane runs a GFDL climate experiment through ESC interface. All relevant metadata for that experiment are stored in ESC.
Precondition
Curator has a user interface and back-end system sophisticated enough to run and store metadata for a GFDL climate experiment.
There is a Curator metadata XML lingua franca that has been adopted by GFDL.
Jane has access to the Internet via a web browser and has a Curator account.
Post-conditions (Expected Outcome)
From within ESC, Jane successfully runs a climate experiment. ESC stores the metadata relevant to that experiment.
Main success scenario
Jane logs into ESC.
Jane navigates to the configure experiment page.
Jane fills out the “Configure Experiment” form:
Jane enters information in the following fields:
- Experiment Name
- Experiment Description
- Components & configuration (Grid specifications, etc).
- Ocean
- Land
- Sea-ice
- Atmosphere
- Couplers & configuration
- Initial conditions
- Forcing fields
- Post-processing
Once the experiment is assembled, Jane launches the climate experiment.
Jane logs out of ESC.
Several hours later Jane receives an email indicating that the climate experiment is finished. [The notion of a jobqueue is an interesting one here. Jane should be able to login to ESC and check on the status of her job.]
Jane logs back into Curator system.
Jane navigates to the “Model Run Results” page.
ESC has recorded all the metadata details of her experiment.
Finding the details of her experiment noteworthy. Jane decides to publish her experiment to the ESC data repository.
This action leads to the metadata details of this experiment being searchable by other climate scientist through the ESC interface. There is enough information for Jane’s collaborators to reproduce this experiment.
Issues
The question I wonder about is whether it is sufficient for Curator to just be a nice front-end to the RTS.xml.
What role will CMOR have with this internal view of Curator?
Jane should be able to check on the status of her model run. How will that happen?
Obstacles
What if Jane wants to bring in her own custom code into the experiment? How will that happen?
What if Jane wants sophisticated parameterizations that are currently not available? How will that happen?
In order to make the design and implementation feasible, will we have to limit the options that are available for climate model configuration?
Further Use cases
Programmatically query repository.
Link to this Page