Difference between revisions of "Priorities"
Line 13: | Line 13: | ||
!style="text-align:top;rowspan:2";|Description |
!style="text-align:top;rowspan:2";|Description |
||
!style="text-align:top;rowspan:2";|Owner |
!style="text-align:top;rowspan:2";|Owner |
||
− | !style="text-align:top;rowspan:2";|Desirable for UKESM2? |
||
!style="text-align:top;rowspan:2";|Status |
!style="text-align:top;rowspan:2";|Status |
||
!style="text-align:top;rowspan:2";|Comments from this meeting |
!style="text-align:top;rowspan:2";|Comments from this meeting |
||
!style="text-align:top;rowspan:2";|Complete |
!style="text-align:top;rowspan:2";|Complete |
||
|- |
|- |
||
− | |1 || Review UKCA tests to cover all requirements || Luke Abraham / Mohit Dalvi |
+ | |1 || Review UKCA tests to cover all requirements || Luke Abraham / Mohit Dalvi || Will cover these in a table in future for review. <br/> Updated testing coverage on Linux and can run 1x1 decomposition || || |
|- |
|- |
||
− | |2 || Updates/fixes to GLOMAP routines || Dan Grosvenor |
+ | |2 || Updates/fixes to GLOMAP routines || Dan Grosvenor || Graham and Sandip developed updates for stratospheric aerosol on old branch <br/> It is desirable to include this to the trunk at some point <br/> November 23 - we clarified what this priority was Graham and Sandip developed updates for stratospheric aerosol on old branch <br/> It is desirable to include this to the trunk at some point <br/> (originally owned by Adrian Hill) || || |
|- |
|- |
||
− | |3 || Standalone UKCA configurations || Luke Abraham/Scott Archer-Nicholls |
+ | |3 || Standalone UKCA configurations || Luke Abraham/Scott Archer-Nicholls || ExCALIBUR work progressing.<br/> Aim to remove UKCA from UM trunk at vn12.3. <br/> See [[Box_model|UKCA Box Model]] page || || |
|- |
|- |
||
− | |4 || Emissions flexibility || probably Mohit Dalvi??? |
+ | |4 || Emissions flexibility || probably Mohit Dalvi??? || (a) <br/> Modify ‘ukca_set_emissions’ to handle data for 1D domains (for consistency with |
ukca_set_environment). <br/> (b) <br/> Improve interface for registration of aerosol emissions by avoiding the need for parent to know |
ukca_set_environment). <br/> (b) <br/> Improve interface for registration of aerosol emissions by avoiding the need for parent to know |
||
the number of entries required for each emission in the internal data structure. <br/> (c) <br/> Improve handling of vertical scaling factors to reduce redundancy. <br/> [Desirable on technical grounds: (a), (b) and potentially (c) will remove unnecessary complexity for parents. (b) will require linked tickets for compatibility so becomes harder to do later if used by more parents. Vertical scaling factors are 3D fields, often all 0 apart from surface - cannot persist internally between time steps in LFRic/NAME and must be regenerated or handled by parent, so scope for efficiency savings with (c).] |
the number of entries required for each emission in the internal data structure. <br/> (c) <br/> Improve handling of vertical scaling factors to reduce redundancy. <br/> [Desirable on technical grounds: (a), (b) and potentially (c) will remove unnecessary complexity for parents. (b) will require linked tickets for compatibility so becomes harder to do later if used by more parents. Vertical scaling factors are 3D fields, often all 0 apart from surface - cannot persist internally between time steps in LFRic/NAME and must be regenerated or handled by parent, so scope for efficiency savings with (c).] |
||
|| || |
|| || |
||
|- |
|- |
||
− | |5 || RADAER API || Alan Hewitt (main) |
+ | |5 || RADAER API || Alan Hewitt (main) || Radaer API on hold until after GA7 port to LFRic. <br/> Functionality of RADAER in LFRic will work without an API although it will be preferable to update this in future. <br/> Not clear if RADAER will work independently of UM in single column mode. <br/> AJH and AJ and JH and Ben Johnson and NB need to work out if desirable and practical to run independent of UM. || || |
|- |
|- |
||
− | |6 || RADAER optimisation <br/> emulation || Alan Hewitt (main) <br/> Ben Johnson |
+ | |6 || RADAER optimisation <br/> emulation || Alan Hewitt (main) <br/> Ben Johnson || AJH highlighted that there have been some discussions about this - a few people interested in this (James Manners, AJ, Ben Johnson, AJH). <br/> AJH and others to look at emulation of RADAER || || |
|- |
|- |
||
− | |7 || RADAER optimisation <br/> integration points || Alan Hewitt <br/> Ben Johnson (main) |
+ | |7 || RADAER optimisation <br/> integration points || Alan Hewitt <br/> Ben Johnson (main) || Have had someone looking to see if something could be changed to make it faster to make a trimmed-down version with less calculations for e.g. NWP. <br/> AJH and BJ to look at making number of integration points a user variable <br/> AJH has functionality to do 1 integration point now. BJ is going to lead this one and see what happens when run with this for use in a weather context. Would give a ~4x speed-up for RADAER. || || |
|- |
|- |
||
− | |8 || FastJX-GLOMAP coupling || TBC <br/> Ticket owned by Fiona |
+ | |8 || FastJX-GLOMAP coupling || TBC <br/> Ticket owned by Fiona || UMvn10.8 branches exist - see ticket [https://code.metoffice.gov.uk/trac/um/ticket/2689 UM:#2689]<br/>'''Note:''' FastJX rates also need updating - potential for code as well as input-file changes || || |
|- |
|- |
||
− | |9 || Cloud-J || TBC |
+ | |9 || Cloud-J || TBC || || || |
|- |
|- |
||
− | |10 || UKCA STASHmaster PP-codes || Luke Abraham |
+ | |10 || UKCA STASHmaster PP-codes || Luke Abraham || Verification system makes some use of existing PP-codes. '''New items should have a PP-code of 0''' and verification team will use STASH items for these if required. || || |
|- |
|- |
||
− | |11 || Removal of redundant code || ALL |
+ | |11 || Removal of redundant code || ALL || Parked whilst GA7.0 is a protected suite. || || |
|- |
|- |
||
− | |12 || GLOMAP-CLIM API refresh || Alan J Hewitt |
+ | |12 || GLOMAP-CLIM API refresh || Alan J Hewitt || Cuurently parked. <br/> Extend API to include GLOMAP-clim within the core UKCA code (or alternatively provide UKCA routines required by GLOMAP-clim in the UKCA API). Required to avoid the need to by-pass the API.|| || |
|- |
|- |
||
− | |13 || N48L85 configuration || TBC |
+ | |13 || N48L85 configuration || TBC || || |
|- |
|- |
||
− | |14 || ACTIVATE optimisation || TBC |
+ | |14 || ACTIVATE optimisation || TBC || A quick win here is to reduce nwbins in namelist ( see um:#4740 ). <br/> Parked for now. <br/> There does not seem to be resource to look at this at present.<br/> This could be tested for UKESM-fast. Another potential change would be to alter the call to only when there are clouds || || |
|- |
|- |
||
− | |16 || Glomap aerosol and GPU || Alan J Hewitt (main) <br/> Joe Abram |
+ | |16 || Glomap aerosol and GPU || Alan J Hewitt (main) <br/> Joe Abram || Started at vn13.3 || || |
|- |
|- |
||
− | |17 || Photolosis and GPU || Joe Wallwork <br/> Joe Abram || |
+ | |17 || Photolosis and GPU || Joe Wallwork <br/> Joe Abram || || || |
|- |
|- |
||
− | |18 || ASAD Chemistry solver and GPU || Joe Wallwork |
+ | |18 || ASAD Chemistry solver and GPU || Joe Wallwork || Nov 23 - Working with PsyClone for this, as itdoes most of the directives. || || |
|- |
|- |
||
− | |19 || Support for UM hybrid resolution model and provision of internal variables via the API || Mark Stringer (main) |
+ | |19 || Support for UM hybrid resolution model and provision of internal variables via the API || Mark Stringer (main) || Required to avoid the need to by-pass the API. <br/> Currently a low priority task. <br/> Code is protected by rose-stem so do not need to fix this urgently. <br/> Nov 23 - MS has little time to work on this at present. An edge case for API usage. || || |
|- |
|- |
||
− | |20 || Removal of unnecessary UM module dependencies in LFRic builds || Richard Gilham (main) |
+ | |20 || Removal of unnecessary UM module dependencies in LFRic builds || Richard Gilham (main) || LFRic currently uses the UM version of ukca_um_legacy_mod which is dependent on a large number of modules (e.g. STASH system modules) that are not required. <br/> In principle, this could be fixed by an LFRic version of ukca_um_legacy_mo. <br/> Rich has been looking at an alternative method using compiler directives (see Ticket ukca:#123 Purge usage of copydiag and friends for non-UM use cases). <br/> [Essential: Remaining dependencies on a large number of unused modules (e.g. STASH etc.) are impacting on efficiency of LFRic build.] || || |
|- |
|- |
||
− | |21 || Removal of remaining UM module dependencies in non-UM/LFRic builds || John Hemmings |
+ | |21 || Removal of remaining UM module dependencies in non-UM/LFRic builds || John Hemmings || This will allow UKCA to be built with no UM repository modules (7 are currently needed). Unassigned ticket UKCA:#28 covers most of the required work. <br/> [Essential: It must be possible to use UKCA without access to the UM repository.] || || |
|- |
|- |
||
− | |22 || Reworking of internal data flow for API-level multi-thread capability || TBC |
+ | |22 || Reworking of internal data flow for API-level multi-thread capability || TBC || Covered by unassigned ticket UKCA:#30. <br/> [High priority for both LFRic and NAME: Needed a.s.a.p. to achieve the required performance but expected to be a big job]. || || |
|- |
|- |
||
− | |23 || Removal of photolysis support code || TBC |
+ | |23 || Removal of photolysis support code || TBC || Photolysis still relies on UKCA to determine its environmental driver requirements. Covered by unassigned ticket UKCA:#79. Needs to be coordinated with photolysis API development. <br/> [Essential: Until this is done, photolysis cannot be run independently of UKCA and the need to retain support for photolysis adds unnecessary complexity to UKCA.] || || |
|- |
|- |
||
− | |24 || Handling of solar calculations and time inputs || TBC |
+ | |24 || Handling of solar calculations and time inputs || TBC || Covered by unassigned tickets UKCA:#27 and UKCA:#92. Ticket #27 addresses inconsistencies between calculations for offline oxidants and isoprene and the potential for inconsistency with photolysis as well as avoiding the need for a 'previous_time' input. <br/> [Desirable on scientific and technical grounds: Current redundancy in required inputs adds unnecessary complexity for parent models and there is significant scope for efficiency improvements and internal simplification.] || || |
|- |
|- |
||
− | |25 || New error handling method || John Hemmings |
+ | |25 || New error handling method || John Hemmings || (a) <br/> Replace error handling in API routines with new method using 'error_report' (see Ticket UKCA:#29). <br/> (b) <br/> Replace all 'ereport' calls with 'error_report' and add support for optional return of control to parent. <br/> [Desirable on technical grounds: Current inconsistencies are confusing and make code over complex. (a) is therefore relatively high priority. Return-control-to-parent functionality to be completed by (b) is lower priority but is desirable for anticipated large ensemble runs (to avoid having to abandon the whole ensemble on failure).] || || |
|- |
|- |
||
− | |26 || Replace/remove UM legacy code in UKCA (where possible) || TBC |
+ | |26 || Replace/remove UM legacy code in UKCA (where possible) || TBC || Includes UM humidity calculations, grid-specific aspects of volcanic SO2 and mass calculations, surface area calculation for heterogeneous chemistry on CLASSIC aerosols. May include ozone column calculation which is also used by photolysis (can be provided as a parent callback routine to act on post chemistry ozone). <br/> [Desirable on technical grounds: Current code is over complex, by-passes the API (thus restricting UKCA development) and is difficult to maintain.] || || |
|- |
|- |
||
− | |27 || UKCA API user guide || John Hemmings (main) |
+ | |27 || UKCA API user guide || John Hemmings (main) || This needs to include comprehensive instructions for setting up UKCA from scratch, independently from the UM, in a new parent application. <br/> [Essential.] || || |
|- |
|- |
||
|} |
|} |
Revision as of 11:00, 17 January 2024
The aim of this page is to record progress with UKCA priority tasks.
Progress with priorities table
ACTION REQUEST FROM UKCA SCIENCE & MANAGEMENT BOARD: CMG to provide list of priorities & their status, with particular thought to UKESM2 changes that will be required.
We may not need Desirable for UKESM2 for much longer.
The current priorities are:
# | Description | Owner | Status | Comments from this meeting | Complete |
---|---|---|---|---|---|
1 | Review UKCA tests to cover all requirements | Luke Abraham / Mohit Dalvi | Will cover these in a table in future for review. Updated testing coverage on Linux and can run 1x1 decomposition |
||
2 | Updates/fixes to GLOMAP routines | Dan Grosvenor | Graham and Sandip developed updates for stratospheric aerosol on old branch It is desirable to include this to the trunk at some point November 23 - we clarified what this priority was Graham and Sandip developed updates for stratospheric aerosol on old branch It is desirable to include this to the trunk at some point (originally owned by Adrian Hill) |
||
3 | Standalone UKCA configurations | Luke Abraham/Scott Archer-Nicholls | ExCALIBUR work progressing. Aim to remove UKCA from UM trunk at vn12.3. See UKCA Box Model page |
||
4 | Emissions flexibility | probably Mohit Dalvi??? | (a) Modify ‘ukca_set_emissions’ to handle data for 1D domains (for consistency with ukca_set_environment). |
||
5 | RADAER API | Alan Hewitt (main) | Radaer API on hold until after GA7 port to LFRic. Functionality of RADAER in LFRic will work without an API although it will be preferable to update this in future. Not clear if RADAER will work independently of UM in single column mode. AJH and AJ and JH and Ben Johnson and NB need to work out if desirable and practical to run independent of UM. |
||
6 | RADAER optimisation emulation |
Alan Hewitt (main) Ben Johnson |
AJH highlighted that there have been some discussions about this - a few people interested in this (James Manners, AJ, Ben Johnson, AJH). AJH and others to look at emulation of RADAER |
||
7 | RADAER optimisation integration points |
Alan Hewitt Ben Johnson (main) |
Have had someone looking to see if something could be changed to make it faster to make a trimmed-down version with less calculations for e.g. NWP. AJH and BJ to look at making number of integration points a user variable AJH has functionality to do 1 integration point now. BJ is going to lead this one and see what happens when run with this for use in a weather context. Would give a ~4x speed-up for RADAER. |
||
8 | FastJX-GLOMAP coupling | TBC Ticket owned by Fiona |
UMvn10.8 branches exist - see ticket UM:#2689 Note: FastJX rates also need updating - potential for code as well as input-file changes |
||
9 | Cloud-J | TBC | |||
10 | UKCA STASHmaster PP-codes | Luke Abraham | Verification system makes some use of existing PP-codes. New items should have a PP-code of 0 and verification team will use STASH items for these if required. | ||
11 | Removal of redundant code | ALL | Parked whilst GA7.0 is a protected suite. | ||
12 | GLOMAP-CLIM API refresh | Alan J Hewitt | Cuurently parked. Extend API to include GLOMAP-clim within the core UKCA code (or alternatively provide UKCA routines required by GLOMAP-clim in the UKCA API). Required to avoid the need to by-pass the API. |
||
13 | N48L85 configuration | TBC | |||
14 | ACTIVATE optimisation | TBC | A quick win here is to reduce nwbins in namelist ( see um:#4740 ). Parked for now. There does not seem to be resource to look at this at present. This could be tested for UKESM-fast. Another potential change would be to alter the call to only when there are clouds |
||
16 | Glomap aerosol and GPU | Alan J Hewitt (main) Joe Abram |
Started at vn13.3 | ||
17 | Photolosis and GPU | Joe Wallwork Joe Abram |
|||
18 | ASAD Chemistry solver and GPU | Joe Wallwork | Nov 23 - Working with PsyClone for this, as itdoes most of the directives. | ||
19 | Support for UM hybrid resolution model and provision of internal variables via the API | Mark Stringer (main) | Required to avoid the need to by-pass the API. Currently a low priority task. Code is protected by rose-stem so do not need to fix this urgently. Nov 23 - MS has little time to work on this at present. An edge case for API usage. |
||
20 | Removal of unnecessary UM module dependencies in LFRic builds | Richard Gilham (main) | LFRic currently uses the UM version of ukca_um_legacy_mod which is dependent on a large number of modules (e.g. STASH system modules) that are not required. In principle, this could be fixed by an LFRic version of ukca_um_legacy_mo. Rich has been looking at an alternative method using compiler directives (see Ticket ukca:#123 Purge usage of copydiag and friends for non-UM use cases). [Essential: Remaining dependencies on a large number of unused modules (e.g. STASH etc.) are impacting on efficiency of LFRic build.] |
||
21 | Removal of remaining UM module dependencies in non-UM/LFRic builds | John Hemmings | This will allow UKCA to be built with no UM repository modules (7 are currently needed). Unassigned ticket UKCA:#28 covers most of the required work. [Essential: It must be possible to use UKCA without access to the UM repository.] |
||
22 | Reworking of internal data flow for API-level multi-thread capability | TBC | Covered by unassigned ticket UKCA:#30. [High priority for both LFRic and NAME: Needed a.s.a.p. to achieve the required performance but expected to be a big job]. |
||
23 | Removal of photolysis support code | TBC | Photolysis still relies on UKCA to determine its environmental driver requirements. Covered by unassigned ticket UKCA:#79. Needs to be coordinated with photolysis API development. [Essential: Until this is done, photolysis cannot be run independently of UKCA and the need to retain support for photolysis adds unnecessary complexity to UKCA.] |
||
24 | Handling of solar calculations and time inputs | TBC | Covered by unassigned tickets UKCA:#27 and UKCA:#92. Ticket #27 addresses inconsistencies between calculations for offline oxidants and isoprene and the potential for inconsistency with photolysis as well as avoiding the need for a 'previous_time' input. [Desirable on scientific and technical grounds: Current redundancy in required inputs adds unnecessary complexity for parent models and there is significant scope for efficiency improvements and internal simplification.] |
||
25 | New error handling method | John Hemmings | (a) Replace error handling in API routines with new method using 'error_report' (see Ticket UKCA:#29). (b) Replace all 'ereport' calls with 'error_report' and add support for optional return of control to parent. [Desirable on technical grounds: Current inconsistencies are confusing and make code over complex. (a) is therefore relatively high priority. Return-control-to-parent functionality to be completed by (b) is lower priority but is desirable for anticipated large ensemble runs (to avoid having to abandon the whole ensemble on failure).] |
||
26 | Replace/remove UM legacy code in UKCA (where possible) | TBC | Includes UM humidity calculations, grid-specific aspects of volcanic SO2 and mass calculations, surface area calculation for heterogeneous chemistry on CLASSIC aerosols. May include ozone column calculation which is also used by photolysis (can be provided as a parent callback routine to act on post chemistry ozone). [Desirable on technical grounds: Current code is over complex, by-passes the API (thus restricting UKCA development) and is difficult to maintain.] |
||
27 | UKCA API user guide | John Hemmings (main) | This needs to include comprehensive instructions for setting up UKCA from scratch, independently from the UM, in a new parent application. [Essential.] |
Table of previously completed priority tasks
Description | Owner | Status | Comments |
---|---|---|---|
UKCA API | John Hemmings (main) | UKCA API work is progressing (AJH). Radaer API on hold until after GA7 port to LFRic. Functionality of RADAER in LFRic currently works without an API although it will be preferable to update this in future. Main task complete |
Main task complete Some follow up tasks have been created |
UKCA and NAME coupling | John Hemmings (main) | ukca:#25 was lodged for UM13.2 Now working with RAQ. PMJ has been able to build UKCA as a shared library for NAME use. |
Work complete on UKCA side Science testing started with NAME |
Separare UKCA repository | Luke Abraham | Separate UKCA trac environment set-up on MOSRS here, will hold UKCA code on completion of current ExCALIBUR project. Aiming for vn12.3. UKCA removed from UM at UMvn13.0 - see UKCA repo Closed |
|
Non-API LFRic work, e.g. Porting aerosol science functionality etc. | Alan J Hewitt | Porting calculation of CDNC from GLOMAP Climatology aerosols complete. I will pass required elements of ukca_radaer to ukca_radaer_band_average (um:#5957) Porting calculation of fields in radaer_band_average to be passed to Socrates underway(lfric:#2488) Includes RADAER, need to use imode_setup=8 RADAER now callable from LFRic, can do GA-type comparisons from LFRic to UM. JH to put UKCA into LFRic soon - working in an SCM configuration (GLOMAP with offox) - needs more testing. A lot more work to do, possibly until end of 2021. AJH has tickets going in for 12.1 that would cover all of the basic GA7 but with climatological aerosol. JH tickets will do the prognostic alternative. JH currently has an LFRic-SCM configuration with prognostic aerosol. JH now has UKCA working in proto-GAL7 (LFRic) config. Needs to couple with convection & plume scavenging. Convection & plume scavenging has been done. AJH - some of this has been done or is in progress. |
|
Dry deposition and Jules tiles | Garry | If Jules adds extra tiles for UKESM2 we may need to hard code options in dry deposition (like in um:#1761) Aerosol Dry Deposition will be considered in the JULES context once the gas-phase changes have been committed. Completed for UM13.1 |
|
Fixes for chemistry rates and ASAD (bugs #101 & #103) | Luke Abraham | #101 still open, #103 completed (necessary for GC5 ticket GMED:#551) Issues on EX1A. Closed - see also UKESM:#841 for UKCA package ticket |
|
Rc fixes for water (bug #104) | UKESM Core Team | Lodged at UMvn11.7 in ticket UM:#5167 | |
CH4 deposition fixes (bug #109) | Gerd Folberth | Closed | |
Solar cycle in photolysis | Luke Abraham | Some code lodged at UMvn10.9 (UM:#2587) but needs bugfixing. This work has stalled in ticket UM:#4828 (bug #110). UM:#4828 lodged for UMvn12.2. See also UKESM:#841 for UKCA package ticket |
|
Deposition updates from CSIRO | Code lodged in UM:#4020 (UMvn11.4, oceanic dry deposition) and UM:#5713 (UMvn11.8, Lightning NOx updates). See also UKESM:#841 for UKCA package ticket |
||
Sol/Insol | Anthony Jones (main) | ukca:#14 was lodged for UM13.2 | |
Stratospheric Ozone improvements | James Keeble/Luke Abraham | Work lodged and in package testing for UKESM2 |