Skip to content

CTSM 5.3: New surface datasets, git-fleximod intro, Leung dust, explicit AC, Matrix-CN, FATES transient LU, time/history metadata changes

Latest
Latest
Compare
Choose a tag to compare
@ekluzek ekluzek released this 30 Jan 00:37
· 2 commits to master since this release
ctsm5.3.021
fcc3e81
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Purpose and description of changes since CTSM 5.2 (tag ctsm5.2.005)

New features

Answer changes

Changes to defaults for clm6_0 physics:

  • Urban explicit A/C turned on (links above).
  • Snow thermal conductivity method is now Sturm1997. (PR #2348; see also discussion #1960)
  • New initial conditions files for f09 ("1-degree" 1850, 2000), f19 (“2-degree” 1850), and ne30 (1850, 1979, 2000) resolutions.
  • New crop calendars. (PR #2664; informed by Rabin et al., 2023)
  • Dust emissions method is now Leung_2023 (links above).
  • Excess ice is turned on. (PR #1787)
  • Updates to MEGAN for BVOCs. (PR #2588)
  • New BGC fire method li2024crujra: Avoid crop fires during growing season; allow lightning ignitions in tropical closed forests; add effect of landscape fragmentation on ignitions and duration; recalibrate against GFED5 burned area and with CRU-JRA climate. (PR #2684, PR #2711, PR #2715)

Changes for all physics versions:

  • Parameters updated for CRU-JRA forcing: PPE-based modifications were made to the parameters leafcn, slatop, froot_leaf, medlynslope, and kmax. Lowers LAI and biomass in boreal and tropical forests, without reducing latent heat in the tropics. Affected PFTs: NET temperate, NET boreal, BET tropical, BDS boreal, C3 arctic grass. (PR #2500)
  • FATES parameter file updated (see section below).
  • Pass active glacier (CISM) runoff directly to river model (MOSART) (MOSART PR #94)
  • New surface datasets and landuse timeseries files (see section below).
  • CNMatrix is new default spinup method (links above).

Heads up

  • Small glacier changes mean that you can’t use a 5.3 surface dataset with pre-5.3 code and vice versa anymore. (Merged with PR #2500)
  • Updates the definition of history variable “time” from end of time_bounds to middle of time_bounds. (PR #2838; see section below)
  • Standardizes history variable attributes and a history dimension name. (PR #2052; see section below)

Additional detail

Changes related to time and history files

(Note that the same information in this section applies to MOSART and RTM.)

Startup and hybrid runs no longer run the 0th time step, consistent with the same change in CAM. (Branch and continue runs never had this 0th time step.) This means you will not get an extraneous initial history file anymore. In some circumstances this may also affect the names of history files.

In most cases, the history time variable is now defined as the middle of a history file’s time_bounds instead of the end, for consistency with the same change in CAM. The exception is if you specify hist_avgflag_pertape = 'I' for that file, in which case it will be treated as an “instantaneous” file. Instantaneous history files (a) have their time coordinate set to the end of the last timestep (as did all history files before this tag) and (b) do not include time_bounds.

The history dimension name hist_interval (of output variable time_bounds) is standardized to be nbnd. History variables time_bounds, mcdate, mcsec, mdcur, and mscur are standardized to include the calendar attribute.

New surface datasets and landuse timeseries files (PR #2500)

  • Transient landuse timeseries files going back to 1700 now possible (and made for f09).
  • Fix an important bug on soil fields that was there since ctsm5.2.0. This has the side effect of mksurfdata_esmf now giving identical answers with a change in number of processors, as it should. (Issue #2744)
  • Surface datasets now provided for the ne0np4.POLARCAP.ne30x4 grid. (PR #2716, issue #2720)
  • Surface datasets now have their version number embedded to prevent mismatch of surface dataset and CTSM version. (Issue #2723)
  • Remove outdated hydrology VIC (Variable Infiltration Capacity Hydrology model) fields from surface datasets.
  • Updates to input datasets:
    • PFT/LAI/soil-color raw datasets; now from the TRENDY2024 timeseries that ends in 2023. (Issues #2570 and #2452)
    • Two fire datasets: crop fire peak month and peatland fraction. (Issue #2618)
    • Glacier behavior dataset (related to how non-Greenland glaciers are handled). (Issue #423)

Changes to FATES parameter file

Changes to rpointer files

The rpointer files are simple text files that CESM uses to keep track of how far simulations have progressed, pointing to the filename of the latest restart file for that component. There is one such file for each component, so for CTSM I cases that's lnd, cpl, and atm (and rof if it's active). Normally, when the user is just extending the length of simulations, there’s no need to worry about these files.

However, if there was a problem when a simulation shut down, it's possible that different components will have mismatched restarts and rpointer files. In the past, this meant figuring out what restart file should be pointed to in each component rpointer file and correcting it by hand in an editor. There was only the final set of rpointer files that was kept for a case.

Now, with this update, the lnd, cpl, and atm rpointer files have the simulation date in the filenames, so it's easy to spot if the restarts are mismatched for one of the components. Also, since there are matching rpointer files for each time restarts are created, it's now easier to (a) make sure restarts and rpointer files are all correctly matched and (b) for a user to take a set of restarts and matching rpointer files to start up from for any part of an existing simulation. This means you don't have to hand-edit the rpointer files, making sure you don't make a mistake when you do.

Old rpointer filenames:

  • rpointer.atm
  • rpointer.cpl
  • rpointer.lnd

New names:

  • rpointer.atm.YYYY-MM-DD-SSSSS
  • rpointer.cpl.YYYY-MM-DD-SSSSS
  • rpointer.lnd.YYYY-MM-DD-SSSSS

Where YYYY-MM-DD-SSSSS is the year, month, day, and seconds into the day for the simulation timestamp. For example, rpointer.lnd.2000-01-01-00000 for a rpointer file for starting at midnight (beginning of the day) on January 1st, 2000.

Note that this is backwards-compatible, so for all the components you can use either the new or old format for the rpointer filenames. Thus, if you are restarting from an existing case before ctsm5.3.016, you CAN use the rpointer filenames that don't have the timestamps in the name.

Simulations supporting this release

Note: Dust emissions in CTSM 5.3 will be different from the above simulations because of a tuning update that came in after those.

Assets 2
Loading