Aerosol activation in parcel

[amylu00]

Adding simple aerosol activation in parcel. Takes in one aerosol type at a time (no mixing) and one mode of aerosols only. Should try to replace S_max with the parcel S_l, but will do so in separate PR?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.77%. Comparing base (2e18da2) to head (6a898b1).
Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #429   +/-   ##
=======================================
  Coverage   96.77%   96.77%           
=======================================
  Files          41       41           
  Lines        1425     1425           
=======================================
  Hits         1379     1379           
  Misses         46       46           

Components	Coverage Δ
src	98.72% <ø> (ø)
ext	69.79% <ø> (ø)

[trontrytel]

I'm not sure if thats the way to use the ARG2000 scheme in parcel. - ARG2000 parameterizes the maximum supersaturation ever achieved by an adiabatic parcel and then uses it to compute the number of activated aerosol particles at the cloud base (not throughout the whole cloud). Maybe if we want to stay close to the assumptions of the derivation we would only apply that tendency once when we first reach supersaturation? Not sure how stable numerically it would be, but probably doable in parcel.

I think a much better approach is to use the supersaturation we are solving for in parcel. And follow the ARG2000 math to compute the number of activated aerosol particles at each time step.

[trontrytel]

I left some longer comment. I would vote to either

• keep ARG2000 as a cloud base only activation scheme
• rewrite it to use our supersaturation as input

[amylu00]

I think a much better approach is to use the supersaturation we are solving for in parcel. And follow the ARG2000 math to compute the number of activated aerosol particles at each time step.

I agree. I've already implemented using supersat from the parcel in a local branch and it's stable! I'll update that in this branch this week

[trontrytel]

I think a much better approach is to use the supersaturation we are solving for in parcel. And follow the ARG2000 math to compute the number of activated aerosol particles at each time step.

I agree. I've already implemented using supersat from the parcel in a local branch and it's stable! I'll update that in this branch this week

Thats awesome! Thank you.

[trontrytel]

Should there be a corresponding sink to water vapor as well?

[amylu00]

Because we are neglecting the condensation onto the aerosol, the dqₗ_dt_a2l term only accounts for the aerosol and not any of the water vapor. So technically what it's doing there is converting a prescribed volume of soluble aerosol into a volume of liquid water. A corresponding sink would be a decrease in "q_a" but since mass of aerosols is not tracked, it's not in the parcel anywhere. I could add it though if that's something we want?

[trontrytel]

In that case, shouldn't it be multiplied by the aerosol density and not liquid water?

dqₗ_dt_act = dNₗ_dt_act * 4 * FT(π) / 3 * r_nuc^3 * ρₗ / ρ_air

Maybe it would be better to not mix the aerosol mass and water mass. We could assume some sort of initial cloud droplet size after activation, and compute the corresponding sink of water vapor based on that

[trontrytel]

Left some small comments - mostly unsure about the water budget. I think there is a sink term missing for water vapor. Also, it would be nice to run the example for a little longer, to test if we are getting the expected behavior.

Thank you for adding it!

[trontrytel]

I think you are now counting dqₗ_dt_a2l twice? You already added it to dqₗ_dt_v2l?