Minutes of the code management group meeting 15/5/2012

From UKCA
Revision as of 09:18, 7 September 2012 by NHSavage (talk | contribs) (add to category)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Location: Cambridge

Present (for all or part of the meeting): Nick Savage (Met Office), Mohit Dalvi (Met Office), Colin Johnson (Met Office), Luke Abraham (Cambridge University), Alex Archibald (Cambridge University), Graham Mann (Leeds University)

  1. Code management. The code management guidelines were discussed. The following amendments to the guidelines were agreed.
    1. The purpose section should be amended to make it friendlier and more focussed on UKCA as a community model implying common working practices. Following the guidelines is required for code to be lodged in the main UKCA code base. All developers are encouraged to follow this advice.
    2. It was agreed that the provision of jobs (both for scientific and testing purposes) was a responsibility of the whole CMG and not just the Met Office.
    3. There was a long discussion about what we mean by "standard jobs" and it became clear that NS and the NCAS scientists were using the term to mean something different! Two type of jobs are really needed and we have to help provide these. First of all well validated, supported, scientific configurations which can be used as 'control' models to test new science. These will be called Supported science configurations. Secondly, jobs used to test the impact of changes being made. These jobs would typically be run for very short integrations to demonstrate that the change has not altered configurations it should not be changing (i.e. if you turn this new change off are the results identical to a job without the branch). These may include some jobs with UKCA off and will be called Standard testing jobs. They are also the basis for any automated testing through the UTF. The details of the jobs to be provided was discussed further later in the meeting.
    4. Tickets. It was decided to attempt to have the ticketing system for UKCA changes on the Collaboration Wiki not on PUMA.
    NS to make requested modifications and publish the new guidelines
    To be proposed at next UKCA exec meeting that a living science plan is needed including input from management on future plans
  2. code ownership.
    1. Formal code owner at Met Office. It was agreed to propose to the next UKCA executive meeting that Nick Savage takes over formal responsibility as code owner. He is already representing UKCA at UM Project Board Meetings.
    2. Subsection experts - names and responsibilities. As the code owner cannot be an expert in every part of the code, it was agreed that we would nominate experts for each subsection of the code. These people would act as first points of contact for people considering modifying the code and the code owner will check if they are happy with the code as part of the lodging process. Note that those names below in italics were not at the meeting and so are currently only suggested as subsection experts, pending discussion. There will be an expert for each of the chemistry schemes.
      • GLOMAP-mode - Graham Mann
      • Core code (ukca_chemistry_ctl , ukca_d1_defs etc) - Colin Johnson
      • RADAER and greenhouse gas data flow to radiation code - Nicolas Bellouin
      • Wet deposition - Zak Kipling
      • Dry deposition - Gerd Folberth
      • Lower BC and emissions - Mohit Dalvi
      • Photolysis - Paul Telford
      • ASAD - Luke Abraham
      • RAQ chemistry - Nick Savage
      • ExtTC - Gerd Folberth
      • Top boundary conditions - Alex Archibald
      • TropIsop (aka CheT) - Alex Archibald
      • Standard tropospheric chemistry - Alex Archibald
      • Chemistry: Stratospheric-Tropospheric (CheST) - Alex Archibald
      • Stratospheric chemistry (includes transform_halogen) - Luke Abraham
      • Heterogeneous chemistry - Colin Johnson
      • Aerosol chemistry - Colin Johnson
    NS to check that all people above are willing to be subsection experts and then publish a list so users know who to contact
  3. Plans for the next year
    1. vn8.3 - minor technical changes only. It has been agreed by the UM project board that this should be a closed release. No science changes will be allowed. At the moment 4 UKCA tickets are open: putting the namelist for UKCA into its own module; making UKCA ENDGAME compatible; splitting up the stratospheric photolysis routines into separate files and removing the final UKCA specific include files (replacing them with modules). Code review submission closing date 14/09/12.
    2. Met Office science changes. The two main Met Office priorities for next year are HadGEM3-ES and moving AQUM to using GLOMAP-mode aerosols. Ticket supporting HadGEM3-ES include: update to lightning emissions, RADAER part 2, nitrate aerosols for GLOMAP-mode, Modal dust. More minor changes which did not make 8.2 but are needed include: the fix to transform halogen, improving the process split for emissions of aerosols, plumbing the ACTIVATE results into the cloud schemes and any changes which come from the work Colin and Graham are doing on aerosol wet deposition. Clearly the addition of the two extra aerosol types will be very challenging. For making AQUM work with GLOMAP-mode the two main task is adding sulphur chemistry to RAQ. There is also a ticket open to add interactive biogenic emissions and to add the ExtTC chemsitry scheme. Colin wishes to extend the work on wet deposition of aerosols to also include soluble gases. Finally a planned rewrite of the emissions code is underway, which will make the code more flexible and also use NetCDF files for input.
    3. NCAS science changes. Although many of the above changes came from NCAS initially, as they have already been adopted by the Met Office, no external review prior to adoption will be needed. The code will still have to be reviewed by the usual Met Office code reviews. Pieces of code which will undergo a preliminary review prior to adoption by MO scientists are described in this section. One of the first pieces of code which will be reviewed is likely to be the wet deposition updates by Zak Kipling. Other possible updates include a version of the unlumped CFC scheme (LA), the flight-track code (Paul Telford with input from NS especially on rotated grids), stratospheric aerosol code updates (GM), rate update for StratTrop/CHEST chemistry (AA to make 8.2 branch and validate). We also need to discuss further issues to do with rate coefficients, such as the policy on IUPAC vs JPL. AA to co-ordindate discussion of wider issues on rates.
  4. Setting up new UKCA trac system. The first question on this was where do we host it. It was agreed that we would see if it could reside on the Met Office Collaboration wiki like the GA projects. If this does not happen we will use the trac system on PUMA. The tickets on this site will be opened in additon to any ticket on the UM trac servers (Met Office and PUMA). Unfortunately this does mean there will be two tickets for each code, but this has the advantage than anyone will be able to see the tickets. To get a community of developers for a community model, we need to be able to see what other developers are doing and what progress they are making. We also need to have a forum for recording the code reviews. NS to ask if we can have a Trac system on the Collaboration website.
    1. templates for code review and ticket summary. NS to set up templates for code review and ticket summary based on typical UM templates.
    2. tickets. Once the trac system is set up, NS will populate it with Met Office tickets which already exist inside the Met Office and invite NCAS scientists to add their tickets.
  5. Configuration management. Colin presented a list of 7 configurations we will need to try support in the medium term. These are:
    1. The AQ model, with RAQ chemistry at 12 km. A currently operational model.
    2. Global forecast models with GLOMAP-mode aerosol in various configurations. Not yet existing
    3. HadGEM3-ES. A version with STratTrop+aerosol chemistry+GLOMAP-mode+ FastJ-X. Under development at the Met Office.
    4. HadGEM3+S chemistry+offline oxidants+GLOMAP-mode. Not yet in existence.
    5. HadGEM3-AO+Stratospheric chemistry.
    6. HadGEM3-AO+ ExtTc. ExtTC only exists on branches at present.
    7. Regional Climate model + ExtTC. Again there is currently no RCM at 8.2 (only at 7.7)
    Given the immature state of some of the jobs above it was agreed that for 8.2 we would initially only provide the following two jobs:
    1. HadGEM3A plus StratTrop chemistry+ACHEM+GLOMAP-mode+Fastj-X using GA4. NS with Fiona O'Connor to provide. Also provide the base climate job from which this job is built to allow easier diffing.
    2. AQUM. NS to provide
    We need both the jobs and input data to be available and a scientific validation. LA to discuss with Fiona.
  6. Code testing prior to adoption by Met Office
    1. Configurations to undergo routine testing. The new code management guidelines mean testing of code must occur before the code is accepted by the Met Office for lodging. We need to clarify what configurations we are to test to see if the code has unexpected impacts (in addition to scientific evaluation of the new code, it should ideally not change results if the change is turned off) . It was agreed that this would involve as a minimum the two 'UKCA On' configurations above plus two 'UKCA off': a small version of the UKV model and an N48 test job, plus a Linux compile test.
    2. UTF on postproc and alternatives. At present the UTF is not available on the Monsoon postproc as it needs an upgrade to RHEL6 first. If this is not available in time for testing of external code, we will need to use more basic methods to test the code.
    3. External reviewers for NCAS changes. To review the code, we at present need Met Office scientists to do it. Possible reviewers for key changes this year (italics those who have not yet agreed):
      1. Convective washout - CJ
      2. Flight-track - NS
      3. Strat aerosol code update - NS
      4. Aerosol primary emissions changes - MD
      5. Rate update - Fiona O'Connor
      6. Sea spray module - Andy Jones
  7. Location for minutes of these meetings and access policy. Agree to put then on the UKCA website and make freely available. NS to write minutes and upload to web.
  8. Date of next meeting. A teleconference will be held in July. Zak Kipling from Oxford to be invited to participate plus any other suitable people from Leeds and Cambridge (Sandip). NS to set up Doodle poll to co-ordinate.

Also discussed more informally were:

  1. Optimisation. LA to co-ordinate any work he does with Met Office optimisation team's work.
  2. Validation suite. Need more data and validation code to be submitted to Mohit for inclusion in suite. GM to send aerosols validation IDL code and data. LA to provide Cambridge data/code. NS to facilitate (i.e. nag).
  3. IRIS. LA expressed an interest in using IRIS but some potential issues came up. LA, MD and NS to investigate further.