Minutes of the code management group meeting 26/06/2014

UKCA Code Management Meeting, 26 June 2014 (tele-conference)

Participants:
Met Office: Nick Savage (NS), Mohit Dalvi (MD), Colin Johnson(CJ), Fiona O'Connor (FC)
Leeds: Graham Mann (GM)
Cambridge: Luke Abraham (LA), Alex Archibald (AA)

Apologies: Oxford: Zak Kipling (ZK)
Met Office: Jane Mulcahy (JM)

Summary of actions

• MD and LA to ensure that all fixes lodged at 9.0 are in 8.4 release job prototype. FOC to tell Luke location of 8.6 branch on Met Office system
• CJ to circulate plots of radical concentrations in stratosphere showing odd banded structure, including version with different photolysis scheme options. Carried forward.
• CJ to email LA re problems with SO2 diagnostics for emissions
• NS Ensure all bug fixes at 9.0 are available at earlier UM versions where needed both for NCAS and Met Office. Not done. See later.
• GM check 8.4 versus 7.3 GLOMAP-mode code and report any bug fixes to Mohit. Carried forward.
• CJ to open ticket and start work on resolving inconsistencies in aerosol chemistry schemes. Need to get rid of unused tracers like CS2 and add in missing chemistry in CHEST.
• NS to speak to JM status of checks for incorrect order of fields in 3D diagnostics and report any relevant information to all.
• NS to ensure bug reporting strategy circulated to all by end of July (in consultation with LA, FOC and MD)
• NS Get GM access to the ukca_bugs email box.
• NS add 9.1 release notes to UKCA website

Actions from last meeting

• FOC Sensitivity test on importance of fixed CH4. Still having stability problems. Not due to the top boundary condition code. Ongoing.
• FOC Create bug fix branch for lodging at 9.1. Lodged.
• CJ and FOC circulate plots of radical concentrations in stratosphere showing odd banded structure, including version with different photolysis scheme options. Carried forward.
• NS to add to this wiki a short explanation of the changes to activation of UKCA tracers at 9.0. Done - link from front page.
• NS Ensure all bug fixes at 9.0 are available at earlier UM versions where needed both for NCAS and Met Office. Not done. See later.
• GM check 8.4 versus 7.3 GLOMAP-mode code and report any bug fixes to Mohit. Carried forward.
• LA Carried forward from last meeting. prepare talk on philosophy of diagnostic code for meeting in Exeter on 2nd April. Done.
• CJ and GM Carried forward from last meeting. discuss way forward on aerosol chemistry and how to handle inconsistencies between chemistry schemes (CHET vs CHEST). Way forward agreed, need to make changes to implement.
• JM keep up dialog on sedimentation between Leeds and Met Office. Ongoing
• JM Carried forward from last meeting.check up on status of checks for incorrect order of fields in 3D diagnostics and report any relevant information to all
• NS Circulate updated version of proposal on how to manage reporting, fixing and lodging of bug fixes including milestones. New timeline agreed to get this up and running by end of July.
• NS Get GM and LA access to the ukca_bugs email box. LA added. NS to also add GM
• NS Produce draft page on release jobs (CHEST+aerosol, CHET+aerosol) with sufficient caveats and circulate. Jobs still not ready. See later item,
• ZK to circulate more details of failures with AEROCOM jobs. Issues resolved - was due to running 2010 without setting trace gas concentrations for 2011.

Fiona noted that volcanic erruptions (e.g. Pinatubo) of SO2 are automatically switched on when stratospheric aerosol is on and might be better on a switch

UM9.1 update

• 9.1 will be released soon. NS to add UKCA specific release notes to the UKCA website (as for earlier releases). There are three key things to be aware of at this release.
  • The non transported prognostics are now controlled via option codes as for the tracers and so do not need STASH changes to turn them on
  • From this release there will be no more User STASHmaster files. Instead users will need to edit the STASHmaster_A in their branch (rose-meta/um-atmos/vn9.0/etc/stash/STASHmaster) and ensure that their suite installs this and uses it correctly.
  • Namelist reads also need some additional work as now only PE0 reads the namelists and broadcasts the names. This means some additional code is needed beyond just adding new items to the relevant names.