Import of high order turbulence closure scheme

[mo-snishimoto]

Contact Details

shusuke.nishimoto@metoffice.gov.uk

Are there any linked Issues or Pull Requests?

#734

Brief description

This issue imports high order turbulence closure scheme (3DTE Mk1 scheme) in UM to LFRic.

Development branch

https://github.com/mo-snishimoto/lfric_apps/tree/3dte_mk1

Documentation

• Namelists and diagnostics added in this branch : Namelist_and_diagnostics.pdf
• Details of the code change : Code_change_detail.pdf
  • 6 sections in this document correspond to the first 6 revisions of the branch, respectively.
• Details of the new matrix solver added in this branch : heptadiag/equation.pdf
• Performed tests : Tests_performed.pdf

Description of the change

• To implement 3DTE scheme in LFRic, following modifications were added to the UM's legacy source code.

  • Elimination of UM’s unique (and not available in LFRic) subroutines or functionality
  • Elimination of global array variables
  • Fix of obvious bugs
  • Removal of redundant options (same change as Remove redundant options and restructure code in ex_coef to be easier to follow and modify further in future #101)
  • Code style change along with the LFRic's coding rule
  • Real precision change (r_umphys -> r_bl) and necessary changes so that 3DTE scheme can work in single precision
  • Implementation of new length scale and TKE-based dynamical leonard term coefficient option, which are not included in the UM's trunk but necessary for complete 3DTE Mk1 scheme

• Tracer transport scheme is modified so that one of the prognostic variables (cov) is transported appropriately.

  • Current transport scheme assumes positive tracer and corrects every tracer 0 or greater after the calculation.
  • Since cov can be both positive and negative, this correction causes a problem.
  • After the consultation with people in Dynamic research team, we decided to consolidate current ‘advective’ and ‘consistent’ tracers
    into a single ‘positive’ tracer and to create a new ‘general’ tracer -capable of taking negative values- to handle the transport of cov.
    • In the latest science configuration, ‘advective’ and ‘consistent’ tracers have an identical configuration. So, this consolidation doesn't affect.

Please refer Code_change_detail.pdf for more details.

• A new regional nesting suite, using 3DTE scheme (run_lfric_atm_3dte-seuk_MG) is added in the rose-stem suite so that least performance will be guaranteed in the future.

KGO changes

• Checksum files for new tests (run_lfric_atm_3dte-seuk_MG) are added.
• Checksum file for run_lfric_coupled_nwp_gal9-C48_ex1a_cce_fast-debug-64bit has changed.
  • This job still uses old tracer transport configuration (i.e. equation_form=2) for advective tracers (an oversight when making changes in ticket#845 in trac?).
  • Since in this branch, ‘advective’ and ‘consistent’ tracers are consolidated into a single 'positive' tracer, which has the latest scientific configuration (i.e. equation_form=3) , the result changes.

Supplement

Currently, 3DTE scheme cannot work with:

• bimodal or PC2 cloud scheme
• Comorph convection scheme
• TKE based RHc option (cloud:rh_crit_opt='tke')
• Microphysics option to consider mixed phase cloud production by turbulence (microphysics:turb_gen_mixph)

as necessary inputs for them are not calculated in this scheme.

[james-bruten-mo]

[Automatic Update]

The Code Review deadline for the Spring 2026 has passed. As this issue does not have a linked pull request it has been removed from the milestone. Please review this issue and either select a new milestone or close it as appropriate. Please contact @MetOffice/ssdteam if you think there has been an error.

Thanks