May 7, 2015

Performance

new model configurations (as exploratory options) are coming online
i/o issues are also important to us
performance may be better optimized if we use multples of 16 levels
- (advice has been updated: sufficient if number of levels is a multiple of 8)
ensemble strategies may present new challenges. Would like some assistance in optimization

Phil Rasch (pnl.gov), Shaocheng Xie --> prioritized list of tasks for Performance Team 15 May 2015

Land

- Phil Rasch (pnl.gov), will provide CERES-EBAF-Surface dataset to land team 29 May 2015

land group recommend test configuration of ALM with prescribed phenology for testing

- Peter Thornton will identify the right individual on land team to help the atmosphere group 15 May 2015

atmosphere team will begin providing many more climate changing modifications to master.
appropriate strategy for testing, evaluation and providing new baseline sims, etc need to be established
perhaps start testing atmosphere only with data ocean and data atmosphere?

time to begin doing short simulations with MPAS-Ocean (prescribed sea ice)
interest in common datasets for verification/validation
interest in exploring "alternate data atmospheres"(not CORE.v2) to drive ocean model
- ACME atmosphere
- alternate observational datasets (e.g. CERES-EBAF-Surface) for radiative fluxes.

improved communications are needed. There is some uncertainty in the atmosphere team about how "showstoppers" identified at /wiki/spaces/ATM/pages/15073766 are being handled by the workflow team. It is clear that some have already been fixed and others are being worked on. But it is not yet clear whether all are being worked on or what is the schedule for completion. We will have some follow up phone calls to resolve this.
One of the showstoppers specifies the need for an observational data repository, with datasets fixed to be CF Compliant and that satisfy minimal ACME standards. (see /wiki/spaces/WORKFLOW/pages/19464769). Jeffrey Painter (Unlicensed) notes that it would take a lot of work to fix all the datasets currently used by the NCAR NCL scripts. Phil Rasch (pnl.gov) agreed, and indicated that the atmosphere team were not specific enough in our request. What in fact we hoped for was:
- a single official obs dataset repository (which could be mirrored) that was backed up, and supported versioning, etc
- as new or updated datasets were committed to that respository that those datasets satisfied the miniminal ACME standards described at /wiki/spaces/WORKFLOW/pages/19464769
part of improved communication is a better method for issue tracking. Dean N. Williams (Unlicensed) anticipates a better proposed solution in about 3 months time, in the meantime we will improvise.