[bug] Perturbation factors not applied to Jacobian simulations in `feature/global_inversion` branch

[eastjames]

Perturbation factors aren't applied to global Jacobian simulations. In src/components/jacobian_component/jacobian.sh, the Jacobian geoschem_config.yml files are edited with sed to apply the perturbation factors, but the command refers to emission_perturbation while geoschem_config.yml has emission_perturbation_factor. As a result, perturbation factors aren't applied.

Here are the lines to change:

https://github.com/geoschem/integrated_methane_inversion/blob/c3aa27366bad9fe55efc7c2f2e0d94e9c29ed1d2/src/components/jacobian_component/jacobian.sh#L95-L99

They can be replaced with

 # Update settings in geoschem_config.yml except for the base run 
    if [ $x -ne 0 ]; then 
     sed -i -e "s|emission_perturbation_factor: 1.0|emission_perturbation_factor: ${PerturbValue}|g" \ 
            -e "s|state_vector_element_number: 0|state_vector_element_number: ${xUSE}|g" geoschem_config.yml 
    fi 

[laestrada]

Hi @eastjames,

I believe that this is just because the feature/global_inversion branch hasn't been synced with dev in quite a while. I suspect running git merge origin/dev in that branch would update the sed command.

also tagging @msulprizio

[eastjames]

Ok, thanks Lucas!

[ltmurray]

I just ran into this problem in the main branch as well.

[laestrada]

Hi @ltmurray,

The underlying issue is likely due to your GCClassic version. The main branch of the IMI (version 1.2.1) assumes that you are using GCClassic v14.0.2. v14.0.2 has the perturbation variable in geoschem_config.yml set by emission_perturbation. This variable got renamed in GCClassic v14.2.1 to emission_perturbation_factor.

There are a few other changes related to the GCClassic version. I recommend either using the main branch with v14.0.2 or switching to our dev branch for IMI 2.0 (the dev branch code is relatively mature at this point). In dev the IMI automatically downloads the correct GCClassic version/ checks that the correct version is installed (Now using v14.2.3).

Let me know if you run into additional issues!

-- Lucas

[msulprizio]

The feature/global_inversion branch has now been merged into dev and this issue should now be resolved. Please let us know if you continue to have problems.