Release notes

From UKCA
Jump to: navigation, search

These UKCA specific notes should be read in conjunction with the general release notes which are available on the collaboration Twiki. These notes cover changes which have a significant impact on UKCA developers and cover these changes in more detail than the standard release notes. Note that this does not include any purely technical changes made to make the code cleaner.


UM 8.3

Bugs

  1. Fix minor bugs in UKCA. Met Office ticket #4382. Large collection of bugs important to include if running at 8.2.
  2. Bugs found using Linux AQUM job. #4641. Format statements and segmentation fault in Fast-J. Repeated USE statements in Main.

Enhancements

  1. Make UKCA ENDGAME compatible. Met Office ticket #4194. First set of changes to make UKCA work with ENDGAME. Other changes were later found to be needed.

UM 8.4

Bugs

  1. Bugfix to set the mass mixing ratio of COS. Met Office ticket #4956

Enhancements

  1. Updates to lightning emissions. Met Office ticket #4151
  2. Update reaction rate co-efficients in the UKCA Newton-Raphson StratTrop and StratTrop&Achem chemistries. Met Office ticket #4399
  3. Plumbing for 2nd aerosol indirect effect using MODE aerosols. Met Office ticket #4793
  4. Change process split for emissions of aerosols in UKCA. Met Office ticket #4298
  5. Add Vehkamakki nucleation routine to UKCA aerosols. Met Office ticket #4397

UM 8.5

The first of a series of changes to support the transition from the UMUI to Rose were made at this release. At present, most users should continue to use the UMUI at vn8.5 and vn8.6. Version 9.0 will retire the UMUI and all users will have to use Rose from then on. The changes discussed below do not have a significant impact on most users but some are important for those developing new code, especially for chemistry schemes which have not yet been lodged. This is an overview of the changes, if you need more information or have any problems at 8.5+ please contact Nick Savage who made the changes.

  1. All ukca related input logicals are now read in from the run_ukca namelist including l_ukca, l_ukca_radaer, l_ukca_aie1, l_ukca_aie2 which were previously in nlstcatm.
  2. The run_ukca namelist has moved from the CNTLATM file to SHARED.
  3. All namelist variables are now given an initial value when declared. See ukca_option_mod.F90. Any new namelist variables being added must have an initial value and this should normally be false for logicals and the Missing Data value for integers and reals.

Chemistry namelist changes

  1. The namelists no longer contain a separate logical for each chemistry scheme e.g. l_ukca_trop, l_strat etc. These are all replaced by a single integer i_ukca_chem. The meaning of each value of i_ukca_chem is set up in the new module ukca_chem_schemes_mod.F90
  2. The original logicals still exist inside the code and are set depending on the value of i_ukca_chem in the new subroutine ukca_setup_chem_mod.F90. It is planned to completely remove the chemistry scheme logicals and replace them with i_ukca_chem at a later date.
  3. The number of reactions, number of tracers etc (jpctr, jpspec, jpbk, jptk, jppj, jphk, jpnr, jpdd, jpdw) have also been removed from the namelists and are set up in ukca_setup_chem_mod.F90
  4. If adding a new chemistry scheme, simpler hand edits can now be used as there are fewer items to set, but developers must add a magic number for the new chemistry scheme in ukca_chem_schemes_mod.F90 and set the number of reactions etc in ukca_setup_chem_mod.F90. However, the active tracers still have to be set via a handedit to the tc_ukca 150 element array. This will change at 8.6 or 9.0.

Photolysis namelist change

The separate logicals for the photolysis schemes have been retired and replaced with a single logical i_ukca_photol with magic numbers defined in ukca_photo_scheme_mod.F90

GLOMAP-mode namelist changes

A number of redundant items have been removed from the namelist which either could not be set from the UMUI or had no effect inside the code.

New Rose metadata

As well as the namelist changes described above, please also note that Rose metadata (used to control the GUI which will replace the UMUI) have been set up. These are held in rose-meta/um-atmos/HEAD/rose-meta.conf and any new namelist items to be lodged will also require this metadata. Please consult Nick Savage for help.

Diagnostics changes

The other major change at this version was the moving of all UKCA chemistry diagnostics into section 50. This means that section 34 only contains prognostic variables and helps to create space to increase the number of available tracers at a later date. Any new diagnostics should be added to section 50. Also note that if you plan to lodge a new diagnostic, you should request a STASH code for it to prevent future conflicts.

IFDEF retirement

As part of wider changes in the UM, the IFDEFs at the top of each routine have been removed from the code and must not be included in any new subroutines. i.e. #if defined(A34_1A) at the top and #endif at the bottom must be removed.

UM 8.6

This was a release open to science changes. Some important non-UKCA items to note:

  • To provide a UM message handler for scaling to 10000's cores, writing output to the "stdout" stream, commonly unit 6 in Fortran is no longer permitted. You must use the newly provided API, which is accessible by including USE umPrintMgr in the calling code. Met Office ticket #4630
  • The UM and UMUI now support up to 999 STASHitems per section. #5345
  • Generation of LBCs within the Unified model is no longer possible. MakeBC must be used instead.

Bug fixes

  • Fix array indexing/ dimensioning issues in UKCA. These problems are ENDGAME specific and are needed for any EG configuration with UKCA on. Met Office ticket #5420
  • Logicals/ Condition block mismatch for UKCA StratTrop and Het/PSC. Met Office ticket #5465
  • Fix for missing activation of Section 38. Met Office ticket #5754
  • Enable UKCA to run in an "Age of Air only" configuration. Fix missing logicals/ conditions to run this configuration. Met Office ticket #5175

Enhancements

  • Introduce gas phase emissions from NetCDF files. Met Office ticket #4270
  • Improve convective washout of aerosols - "in plume scavenging". Met Office ticket #4301
  • RADAER Part 2. Package of changes to GLOMAP-mode direct radiation interaction. Includes nitrate aerosol, stratospheric optical properties, and changes to the width of size distribution of aerosols. Met Office ticket #4302
  • RCP scenario code. Adds code to the UM to read in files with time-series for trace gases including CO2, CH4, N2O and CFCs. The format of these files is the same as those containing the data for the Representative Concentration Pathways (RCP) used in IPCC AR5.Met Office ticket #5220
  • UKCA serial optimisations. Met Office ticket #5409

UM 9.0

UM9.0 was closed to all science functionality changes; it was dedicated to retiring the UMUI and making further Rose changes. It is not intended for general use, rather it is an interim release towards a Rose GUI UM functional release. The umui is no longer supported for this and future UM releases.

The major UKCA change was the retirement of the tc_ukca array used to control which tracers are on and its replacement with a system based on option codes. See Controlling UKCA tracers from vn9.0.

A set of changes to UKCA diagnostics were also introduced. MEt Office ticket #5922 bug in the 3D UKCA emissions diagnostics that mean that the code divides these diagnostic by the timestep twice. Also as there is no code to support it, 50.086 was removed from the STASHmaster file and there was also a comments change to make them clearer.

Finally there were two STASH related changes.

  1. New pseudo levels for Aerosol Optical Depth were introduced, so that it is clear that these are 6 monochromatic wavelengths used for the AOD calculations which are not related to the wavelengths bands used for the radiative transfer code. Met Office ticket #5722
  2. A new time code was introduced so that when a chemistry solver is not called every model timestep, the output is only available on the valid timesteps and the meaning works correctly. Met Office ticket #5434

UM 9.1

UM vn9.1 released 27/06/14

Some further significant changes have been introduced at 9.1 relating to Rose and optimisation for very large numbers of cores anticipated on the next machine. Important non-UKCA items to note are:

  • environment variables are no longer used for file paths - all file paths are now explicitly written in namelists. Met Office ticket #5903. This only changes the way that the RADAER input files are handled. These are now in the namelist nlcfiles although they may be moved to run_ukca at a later date.
  • the handling of many UM namelists has been updated. Details are on the collab Twiki [1]. The model is now only opening the file once and then sending the data in one message to all other processors. This means additional coding is needed to add additional namelist items. If a developer wishes to add a new variable to run_ukca the new read_nml_run_ukca subroutine located in ukca_option_mod.F90 will need changing. The changes required are
    • increment the relevant type parameter by the variable size (for a real scalar increase n_real by 1)
    • add a new line to the list in the my_namelist type declaration in the relevant variable type.
    • add a new line to the my_nml population section in the relevant variable type
    • add a new line to the namelist population section in the relevant variable type.
  • Finally, user STASHmaster files have been retired as part of a rewrite of the handling of STASHmaster reading which was also hugely inefficient for massively parallel machines. In future, if you need to use non standard STASH records (for adding new diagnostics, using user ancillary files or changing which transported tracers are turned on) you will need to make a branch to the STASHmaster_A file, extract it from the repository and point the UM at this file instead.

The major UKCA specific technical change at this release was to change the way that the non transported prognostic variables (chem_diags) are controlled. Previously user STASHmaster files were used to turn these on for UKCA configurations. This is now controlled by the same option codes used to control tracers. If you are using a standard chemistry scheme without any modifications, this means that these stash items are controlled automatically based on the chemistry scheme you have chosen. See Controlling UKCA tracers from vn9.0 for more information on this and instructions on how to change which stash items are on for a specific chemistry scheme.

This release also saw a large number of UKCA tickets lodged to support the release of GA7 using GLOMAP-mode aerosols including a package of bug fixes.

Bug fixes

  • Fast-JX error in UKCA. Needed for any jobs with Fast-JX, correct an indexing error. Met Office ticket #5404
  • UKCA Bugfix Package. Large package of bug fixes. Recommended for all jobs. Met Office ticket #6105
  • missing call to dr hook in tstmsk_ukca. Minor issue, only relevant if using dr_hook timing. Met Office ticket #6011

Enhancements

  • Make "Chemical Diagnostics" work from option codes. This is a key update at this version. It means that the non transported prognostics (items 34.151 onward) now use the same option code for control as the tracers. No STASHmaster changes are needed. Met Office ticket #6129
  • Offline oxidants for MODE. Allows GLOMAP-mode to be run with a very small chemistry scheme using climatological oxidant fields. Met Office ticket #4300
  • Clean up UKCA interface - part 2 (Remove link to CLASSIC). Met Office ticket #4471
  • RADAER and radiation spectral bands. Ensure that the spectral bands used for RADAER are consistent with those in the radiation scheme. Met Office ticket #5723
  • Include deposition diagnostics in UKCA. Additional dry dep diagnostics for the BE RAQ scheme. Met Office ticket #5784
  • UKCA Rose clean ups part 2. Retire some unused logicals, better metadata, additional error checking. Met Office ticket #5882
  • Control new emissions diagnostics with option codes. Met Office ticket #5973
  • Investigate alternative (eg. Priestley) Conservation scheme for UKCA and other tracers under ENDGame. These revisions to tracer transport are required for ENDGAME runs but are not needed for New Dynamics. Met Office ticket #6085
  • tidy up chemistry time codes metadata. Met Office ticket #6042

Changes to Rose stem tests

  • Upgrade the Science options in the HadGEM-UKCA app under Rose-stem. Met Office ticket #5934
  • update AQUM job in Rose Stem to use netCDF input files. Met Office ticket #6178
  • Add dump comparison to HadGEM_UKCA ana task under Rose stem. Met Office ticket #6276

Optimisations

  • UKCA vector optimisations. Met Office ticket #5407
  • Speedup first UKCA timestep. Met Office ticket #5901