IO strategy

[sjboeing]

Mean profiles (1D: z-dependent). Typically good to have at 1 to 5 minutes, depending on computing cost.

• Field averages, field variances and some co-variances (especially those corresponding to vertical fluxes).
• Cloud fraction directly from cloudy/clear parcels.
• Cloud fraction from gridded ql?
• So called mass-fluxes for cloud, updraught, cloudy updraught, buoyant cloudy updraught, and buoyant cloud (based on parcels or fields)? More widely: conditionally averaged statistics (e.g. cloud-averaged values)

2D fields (xy) Typically good to have at 1-minute frequency

• Liquid water path (later to be expanded to rain).
• Water vapour path.
• Near-surface wind and thermodynamic fields (later also surface rainfall flux): could be in cross-sections below though.
• Cloud base
• Cloud top
• Mixed layer height

2D cross-sections (xz: typically just one) Typically good to have at 1-minute frequency

• Prognostic fields

2D cross-sections (yz: typically just one) Typically good to have at 1-minute frequency

• Prognostic fields

2D cross-sections (xy: possibly more than one) Typically good to have at 1-minute frequency

• Prognostic fields

3D full fields Typically just at 30 minute to hourly frequency

• Prognostic fields

3D full parcels Typically less than 10 dumps per simulation.

• Prognostic fields

3D restricted parcels Maybe at 5-minute frequency, or more frequntly bursts? Used for detailed cross-sections and small "boxes".

• Prognostic fields

• Dump parcels/fields restricted to a box or slice.

Note: some of these make more sense for CBL/BOMEX than for e.g. moist bubble.