Add summary statistic (mean, max, min,...) plots and introduce additional aggregation options

[ukmo-huw-lewis]

Contribution checklist

Aim to have all relevant checks ticked off before merging. See the developer's guide for more detail.

• Documentation has been updated to reflect change.
• New code has tests, and affected old tests have been updated.
• All tests and CI checks pass.
• Ensured the pull request title is descriptive.
• Conda lock files have been updated if dependencies have changed.
• Attributed any Generative AI, such as GitHub Copilot, used in this PR.
• Marked the PR as ready to review.

[github-actions]

[ukmo-huw-lewis]

See sample output: https://wwwspice/~huw.lewis/CSET/CSET_RAL3_LFRic_Means2/

Based on /home/users/huw.lewis/CSET/CSET/cset-workflow/rose-suite.conf

[ukmo-huw-lewis]

Example cset-workflow output to support review: https://wwwspice/~huw.lewis/CSET/CSET_RAL3LFRic/WestAfrica_4p4_708_1/

[ukmo-huw-lewis]

Latest update to time_series_case_aggregation_all.yaml recipe now exploits SEQ processing to unpack input cubelists across multiple models/cases, and then output a common 1D cube that includes all times. Example output file below (noting this is only 2 cases, so obvious gap between case dates etc). In general use-case here would be if chunking some input data into cycles and wishing to output continuous timeseries, or to visualise multiple case study outputs on same output:

Example cset-workflow outputs for sub-set of requested variables:
https://wwwspice/~huw.lewis/CSET/CSET_testing/SWUK_pr1377/

[ukmo-huw-lewis]

Example workflow outputs with fuller set of variables and output plot types:
https://wwwspice/~huw.lewis/CSET/CSET_RAL3LFRic/WestAfrica_4p4_708_LOCALDATAb/
This captures 'SEQ' and 'MEAN' outputs for each case, and illustrates additional 'all' aggregation output to distil outputs into single summary plot for each plot type.

[Sylviabohnenstengel]

Rose gui:
A minor change, but I would suggest for clarity purposes to distinguish more clearly between collapsing per datetime forecast and the aggregation section below that aggregates across datetime cases. As I currently see it the "SPATIAL_SURFACE_FIELD_METHOD" allows collapsing or sequence for each forecast individually.

So how about something along those lines:
description=Select analysis method(s) for output mapped plots. Add all options required.
Leave blank or set to "SEQ" (sequence) for plots each diagnostic output time.
For time-collapsed outputs for each datetime, set to "MEAN", "MAX", "MIN", "STD" etc.

[ukmo-huw-lewis]

Thanks for comment on gui description. It highlights a need to be clearer, so agree could do with updating. Confess I don't fully know what to call the unit of "chunk" as in general been aiming for avoid "case" also. Not sure datetime is quite right wording either given i) python library and ii) may suggest each datetime in output - i.e. hour-by-hour again.

How about.....

description=Select analysis method(s) for output mapped plots. Add all options required.
Leave blank or set to "SEQ" (sequence) for plots each diagnostic output time.
For time-collapsed outputs for each analysis period, set to "MEAN", "MAX", "MIN", "STD" etc.

This at least ties into the ANALYSIS_LENGTH env variable.

[Sylviabohnenstengel]

approved subject to addressing minor comments