July 11th
Apology from Juergen
Previous actions status
For discussion
@RachelLaw Single site comparison status - brief mention of ‘lake’ issue and snmin (trigger for changing to multi-level snow)
Notes
@inh599 - mapping tiles to POP/POPLUC discussion - technical and science issue. How to restructure to make most of existing code base. List of requirements e.g. LUC happens at start of year. Flowcharts of what we do now and what we could do for adding POP.
Actions
July 18th
Previous actions status
For discussion
CABLE4 deliverables for next 12 months
ESM1.6 scope
Notes
Idea of a CABLE4 hackathon to start on coding work - funding not available from ACCESS-NRI except if combined with ACCESS-NRI workshop (first week Sep). Also land working group meeting in October as possible to tag onto.
Discussed possible deliverables and limited resourcing. Option to use time at end of ACCESS-NRI workshop to design code restructure to facilitate extra tiles/vegetation mapping needs etc.
Discussed ESM1.6 scope - noting what may not be achievable in time available. Also need to get proto-ESM1.6 build with CABLE3 available soon for testing.
Actions
July 25th
Previous actions status
For discussion
Need to decide whether to add extra meeting onto NRI workshop.
@RachelLaw Single site testing update
Notes
Extra suggestions on single site testing - impact of diffuse fraction, radiation parameter changed by factor of 2 rhocdf/refl_dif. Also note a soil evaporation parameter change.
Discussion on running CABLE-POP (with/without POP) at ACCESS resolution. Worth continuing and single-site work useful for interpreting.
Soil ancillary - latest UM ancillaries look quite different.
Re meeting - check whether Jhan can attend. Ben/Juergen OK for Sep, also Ian. Start planning for Sep date.
Actions
Rachel to contact Jhan re meeting in Sep.
Claire to deal with sum_rad_gradis issue
August 1st
Previous actions status
For discussion
@RachelLaw - single site testing, impact of radiation
Notes
@Juergen in middle of trendy runs for this year. Using input routine from Lachlan. Made some changes to land use processing - will present in a few weeks. Crop and grass transitions. - Crop fraction wasn’t changing from 1700 and better harvest fraction. Also accounting for cropland being abandoned. LUH processing scripts (from PB) have been rewritten.
Note discussion from Julia Pongratz re how LUH2 data has been interpreted for CMIP models. Scientific uncertainty/interpretation as well as technical interpretation. Coming from PhD student papers that Tilo/Ian involved with. Forest cover layer seems to be one part of confusion and how to use it. Savannah challenging.
@RachelLaw showed latest site results - impact of spitter bug and sensitivity to VIS/NIR split. Tried to get VIS (PAR) information out of UM, also direct component of PAR - but output seems to have some potential issues.
@inh599 showed spitter bug impact for Australian test case and benchcab flux sites.
Actions
@inh599 may look further at radiation diagnostics in UM.
@RachelLaw to move onto ESM1.5-CABLE3 case.
August 8th
Apologies: Ben
Previous actions status
For discussion
@Juergen Progress on crops and prim_only file
@lachlanswhyborn restart requirements
Notes
Need further clarity on what fields from restart and what from ancillaries for the CABLE with ACCESS forcing.
Trendy branch running with ACCESS met forcing. What is useful to test with this? Spin up to give vegetation types, then run with and without land use. Relatively cheap to run.
@Juergen happy to see crop working in current trendy runs. Will show some slides another time. Checked out prim_only file - calculation in code causing problems if grid-cells free of primary forest at start of run. Now done as an external input, only done once and consistent with other LUH2 inputs. primn to rangeland conversion - checked impact but decided best not to consider - ends up converting large parts of Australia to grassland. Checking allometry parameters - biomass looks good but trees too short. Two parameter sets in code but don’t know history of why changes were made. Look into wood harvest - but probably for next year.
Actions
August 15th
Previous actions status
For discussion
Code design session after ACCESS-NRI workshop
CABLE4 work deliverables for FY24-25 CMIP7 work plan
@Juergen can show presentation on code development and LUH2 processing (crop fraction, prim_only file). Also two discussion slides on 1) grazing in CABLE, and 2) future model runs with varying POP parameters.
Notes
Code design session - room booked. Need to check video facilities.
Aim for session - what order to tackle code development, which can be done in parallel, add details to the early ones so can be started when people ready. Also whether BLAZE can be included.
Current version now capturing crop reasonably well - still some difference from LUH2 - maybe because CABLE-POP can’t put crops in desert. Capturing declining crop area as well as increasing crop area. Added extra transitions (all within CABLE grass type).
@inh599 spin-up/starting date implications - question around consistency between states/transitions.
prim_only - is this needed. Just in to speed up code? Now done as an input produced when other LUC input files dealt with. Adds to complexity in dealing with output so may be better to remove. Convoluted in code so also a time investment to remove. Discuss for CABLE4.
Representation of grazing based harvest discussed. Note impact for nutrients as well as carbon.
Impact of tree allometry - short - but has big impact on LAI, GPP etc. so left unchanged.
@Juergen Post-Trendy plans. 1000 grid-cell testing of POP on/off and number of cohorts/patches.
Actions
August 22nd
For discussion
@lachlanswhyborn further discussion of CABLE-POP with ACCESS forcing
@RachelLaw latest single site tests - now comparing ESM1.5-CABLE3 with CABLE offline
Notes
@lachlanswhyborn showing slides. Looking at results from individual points, challenge of understanding how biome is mapped to vegetation type - and why different between ACCESS-based run and CABLE-POP run. CABLE-POP seems to be changing points to bare ground.
Reinforces the need to work out how we are going to do biome/pft distribution for CABLE4 e.g. not generated from meteorology.
CABLE-POP needs to have capability to read in a biome/pft map - currently set as a vector.
@RachelLaw - latest comparisons between ESM1.5-CABLE3 online and CABLE3 offline. Latent/sensible heat partition very different for tundra location, also wetland. Some sites showing sensitivity to root fraction, sometimes improving comparison and sometimes making it worse. Running an online case with the rootbeta based froot. Plant respiration finally fixed. GPP still generally lower for online than offline (and new online lower than ESM1.5). c4 grass and crop have larger differences between online/offline. Need to check parameters being used.
Actions
Claire to send out calendar invitations for Thursday/Friday meeting.
Ian, Claire, Juergen to work on agenda.
CABLE4 design workshop
Thursday 2-5pm
Talk through Summary post - have we thought of everything?
First priority to work on ‘CABLE4 offline design / mapping 27-3 (general rather than detail)’
Ian to run through / refresh slides
Timelines / dependencies / resourcing / gaps / what can happen in parallel / what happens in each branch and when can/should they merge
Use forum to capture ‘fleeting thoughts’ and/or hard-copy in room
Time for people to show some slides where relevant
Would be good to talk about vegetation distribution / LUH processing if time.
POP options - what are useful to maintain and implications for what will be needed in ACCESS. Also what will be needed for BLAZE (climate variables). Simpler options than running means etc. POP/CASA interaction re allocation also could be thought about.
Testing environment and test cases
Friday 9am-1pm
Issue-writing - identify which are mostly technical and which are science. Get to the point where some of these can be assigned.
Refine timeline
August 28th
Apologies: Jhan
For discussion
Current plans for next week’s meeting - any pre-work needed?
@RachelLaw Tundra grid-cell: soil evaporation??
Notes
Claire working on work plan summary topic.
Any code writing when people in the room together? Depends on how far we progress with design work.
Code organisation needs to be on the agenda - what the starting point is. Keeping track of issues.
CABLE as ACCESS - getting stuck in spin-up.
Grid-cell testing - tundra point - sensitivity to different meteorology for latent heat / soil evap. Potential problem with initialisation of frozen fraction but would this impact summer evap?
Trendy runs done. Crop harvest large fraction of land use flux. 2023 El Nino showing interesting results. Tried running with ACCESS-NRI ilamb but dependency issues.
Actions
@clairecarouge to prepare agenda for meeting based on notes
@inh599 to prepare / tidy up intro slides
August 18th
Rachel may be late - please start without me.
For discussion
Follow-up from Canberra meeting. Suggest @clairecarouge talks us through the Zenhub roadmap.
Notes
Vegetation distribution - needs discussion soon. JULES has a method for creating pft distribution from other datasets. Is this worth looking at for us? Need some method for AM3, also decision on what ESM1.6 will use.
BIOS3 merge - ancillary as netcdf almost done. Checking reproducibility. Met files in progress. BLAZE capability working in BIOS3 - need to make sure not lost in merge with Trendy branch - MPI issues and BLAZE I/O.
@inh599 has added information to issue about splitting up POP/CASA types.
@RachelLaw - latest on GPP comparisons. @inh599 queried differences in photosynthesis code - noted that formulated differently to accommodate Medlyn option. @Jhan suggested writing out only for a single iveg type.
Actions
Ensure review requests go to tech team not directly to @clairecarouge