Discussion on collecting reasons for omissions from data provider contributions

[martinjuckes]

Matthew Mizielinski
There are limitations as to what a data provider can do; certain models may not have the capability to produce certain variables or the cost of implementing them is too high given other priorities. A large number of high priority, high frequency variables were dropped by UKESM1 as the cost (in computing time) of producing them was too high.

Martin N Juckes
Thanks, that is interesting information. It would be good to try to capture it. I'm not sure what the best approach is. How about inviting modeling centres to submit some information to a github repo, in CSV or JSON, with one row/record for each variable deliberately omitted? Would UKMO be able to submit such a list? Would there be any sensitivity about how widely such information was distributed (should it be a private repository)?

Matthew Mizielinski
I don't see any objections to this coming from UKMO and we do already have preliminary lists of this form embedded in our code for internal use. It might take a couple of hours to arrange, but I'm happy to look into this.

Martin N Juckes
Should we discuss it with the WIP and send out a request to all groups ... as a request to support improvements in the process rather than as a requirement?