CMIP5 Conformance wiki for ticket #250

2009/05/26 Charlotte wrote:

I have created ticket #255 to be the place to work out how to allow users to either upload or describe their forcing data

2009/05/26 Sebastien wrote:

Hi Gerry,

I agree that the mindmap approach would be the good one to define what the conformance would be regarding CMIP5 experiment. I agree it will help us to fix ideas.

If we can understand the way Rupert plan to use mindmap to go to the questionnaire we then can try not to diverge too much from Rupert's approach.

Charlotte's example are good to identify which forcing (initial conditions/boundary conditions) define an experiment. I feel that the hard work will be to describe where those forcings came from.

2009/05/26 Gerry wrote:

Hi Sebastien,

I'm starting to go through the CMIP5 document now to get a feel for what is needed, conformance-wise, for each of the different experiments. Charlotte's example is useful for this. We should maybe discuss what is the best way to direct this among ourselves? Last week, Charlotte and myself decided to go down the mindmap route as opposed to directly creating an xml, for constructing the conformance part of the questionnaire. We thought this would be quicker (but obviously entails more software, Ruperts programs etc). To make sure our mindmap/questionnaire instance stays valid with the CIM, we decided we should create some select CIM instances from some common mindmap pathways. You may feel differently about this, and decide that it may be better to create xmls directly without using a mindmap? We can discuss this a little further in the telco if needs be.

2009/05/26 Charlotte wrote:

I have created a couple of tickets (#250 and #251) to document the way I expect us to work on conformance. At this stage I want to document the particular requirements for each CMIP5 experiment.

After a user has described their model their next task is to link simulations (run with that model) to particular CMIP5 experiments. When they do this they will be asked to describe how the simulation conforms to the particular requirements of that experiment.

Sebastien and Gerry will write mindmaps which can be processed to generate the conformance questions (#250).
I will write CIM instances for selected mindmaps as a "sanity check" and to make sure that we don't depart from the CIM concepts (#251).

Once we have captured the experiment requirements in our mindmaps we can think about how we ask questions in an intelligent way so that we can make use of conformance descriptions from similar/related experiments and also make use of software descriptions from the model documentation.