Hi everyone,
New prerelease executables for ESM1.5 are now available for testing. This update pulls the orbital parameters out from the UM source code, allowing them to be read in from the atmosphere namelist at runtime.
To test the new executables in an ESM1.5 configuration, replace the paths for the executables in the config.yaml
file:
# Modules for loading model executables
modules:
use:
- /g/data/vk83/modules
load:
- access-esm1p5/2024.05.1
with
# Modules for loading model executables
modules:
use:
- /g/data/vk83/prerelease/modules
load:
- access-esm1p5/pr14-3
In the default ESM1.5 configurations, Payu produces an error if it notices that the model executables have changed, as a way to guard against inadvertent changes. To run with the new executables, we’ll have to switch this guard off by commenting out the following lines in the config.yaml
manifest:
reproduce:
exe: True
These lines can be uncommented after the first run.
Once the above executable paths are added, the following variables can be specified in the <control-directory>/atmosphere/namelists
namelist file:
ECCENTRICITY
: The eccentricity of Earth’s orbitOBLIQUITY_RAD
: The obliquity of the Earth, in radiansDAY_PERIHELION
: The time of the perihelion passage, in daysLON_PERIHELION_RAD
: The longitude of the perihelion, in radians.
These variables can be added to the namelist
file in a new namelist group called &ASTRON_PERIODIC
. For example, to specify the values used in a 49ka experiment, add the following:
&ASTRON_PERIODIC
ECCENTRICITY=0.01292,
OBLIQUITY_RAD=0.42647599,
DAY_PERIHELION=323.7698,
LON_PERIHELION_RAD =1.08997557,
/
It doesn’t matter where the &ASTRON_PERIODIC
group appears relative to the other namelist groups. However, note that aside from adding the new group, the order of the existing namelist groups should not be modified.
The &ASTRON_PERIODIC
group can be omitted from the namelist
file, in which case the model will default to the modern values used in previous releases:
&ASTRON_PERIODIC
ECCENTRICITY=1.6710222E-02,
OBLIQUITY_RAD=0.409092804,
DAY_PERIHELION=2.667,
LON_PERIHELION_RAD =1.7967674211761814
/
Tests of the new executables:
The following shows one year of daily zonal mean incoming shortwave from the previous release, and for the new executables with the &ASTRON_PERIODIC
group omitted from the namelist
and the following shows the surface temperature on day 365 using each of the executables.
The following shows zonal mean, monthly mean incoming shortwave for an existing 49ka simulation which used the parameters
ECCENTRICITY=0.01292,
OBLIQUITY_RAD=0.42647599,
DAY_PERIHELION=323.7698,
LON_PERIHELION_RAD =1.08997557,
compared to a simulation using the new executable and the above values specified in the namelist
The following shows zonal mean, monthly mean incoming shortwave for an existing 49ka simulation which used the parameters:
ECCENTRICITY=0.01292,
OBLIQUITY_RAD=0.42647599,
DAY_PERIHELION=323.7698,
LON_PERIHELION_RAD =1.08997557,
compared to a simulation using the new executable and the above values specified in the namelist:
The maximum difference is 3.81469727e-06 W/m2, and it’s unclear whether this arose from different postprocessing, or differences in the executable compilation.
Thanks @HIMADRI_SAINI for sharing the parameters and previous outputs.
Next steps
It would be great to get feedback on using the new executables - do they work with your custom configurations? Do any issues come up trying to use them?
Please feel free to run any tests with the prerelease executables, and let us know how you go! If everything is looking ok, the changes will be incorporated into the next released version of ESM1.5, hence I’d recommend not using this prerelease for any big experiments.