Updates fluxes to apply scaled metrics in deep-atmosphere configurations #577
Conversation
dc1eead to
d2cacb2
Compare
a20f92d to
0d94a18
Compare
|
While it might be better to change the integration rather than the flux directly in the future, I think this is easier for now and is fine. The only problem I can see is that |
|
Is there an easy way to make this backward compatible by e.g. making the default of the metric_scaling one if it's not provided? If not, please open a corresponding PR in ClimaAtmos, so that other people who are running AMIP know what to change if their workflow is affected, thanks! |
Yes - my current approach is to handle the backward compatibility through |
0237b3b to
5409fb6
Compare
e1dc057 to
440d3da
Compare
4726767 to
c014707
Compare
|
@Sbozzolo your 3 comments have been resolved in the latest commit. |
modified: src/optics/Fluxes.jl modified: test/all_sky_with_aerosols_utils.jl modified: test/all_sky_with_aerosols_dyamond_gpu_benchmark.jl modified: test/clear_sky_utils.jl modified: test/clear_sky_dyamond_gpu_benchmark.jl modified: test/cloudy_sky_utils.jl modified: test/cloudy_sky_dyamond_gpu_benchmark.jl modified: test/Project.toml modified: test/aqua.jl
83438fe to
50a191c
Compare
Pass metric scaling term to RRTMGP solver functions for use within the flux calculations to account for column expansion with altitude in deep-atmosphere configurations.
Also fixes broken OS unit tests due to missing package in test env (
Random)Performance comparison:
(main) : https://buildkite.com/clima/rrtmgp-clima-a100-pipeline/builds/53
(PR) : https://buildkite.com/clima/rrtmgp-clima-a100-pipeline/builds/61
Degradation is approx 0.2%
e.g. build log: https://buildkite.com/clima/climacoupler-coarse-nightly-amip/builds/317/steps