Adding Cloudy tendencies #159
Conversation
sajjadazimi
force-pushed
the
ej/add_cloudy
branch
from
5e6e36c
to
916c50a
Compare
Add to-do-list updates Update CloudyMoisture Add cloudy params + activation WIP on advection Fixing advection with dolla-dolla-i final touches now it's unstable with activation fix the activation issue making the results work... Setup the col-sed only case final touches on col_sed + plots working on making it stable fix the activation source term updates to plotting
sajjadazimi
force-pushed
the
ej/add_cloudy
branch
from
916c50a
to
7ce8cee
Compare
| arg_type = String | ||
| default = "NonEquilibriumMoisture" | ||
| "--prognostic_vars" |
There was a problem hiding this comment.
I'm very happy to commit to one prognostic variable choice. Do you have a preference for theta or T? We should delete the other from CI and tendency options too
There was a problem hiding this comment.
I prefer T, because in PySDM KiD and also in Shipway and Hill T is fixed.
edejong-caltech
changed the title
test/create_parameters.jl
Outdated
| @@ -85,4 +86,36 @@ function create_kid_parameters(toml_dict) | |||
| return kid_params | |||
| end | |||
| Base.broadcastable(x::KID.Parameters.KinematicDriverParameters) = Ref(x) | |||
|
|
|||
| function create_cloudy_parameters(NM, FT) | |||
There was a problem hiding this comment.
Could you add some docstring explaining what is happening here? Or link to Cloudy documentation if it's explained there? I'm asking about NM / 3, 1e6 and 1e-9 in norms, 5e0 in kernel_func, 5e-10 1e-6 in KernelTensors and CoalescenceData and vel, etc.
src/Common/tendency.jl
Outdated
| @. q_rai = q_(Y.moments.:($$rain_mass_ind), ρ) | ||
| @. moments = Y.moments | ||
|
|
||
| tmp = @. cloudy_precompute_aux_precip_helper(cloudy_params, moments, pdists) |
There was a problem hiding this comment.
I'm not very good at this, but I think that the way the Cloudy tendencies are added, both here and in 1D model, is allocating. Maybe instead we could create additional variables that we need for Cloudy and put them in scratch for temporary things that need to be stored during computations (like supersaturation, maybe pdists?) and in aux for the ones that need to be precomputed for tendencies.
I was trying to write all the other tendencies like
@inline function precompute_aux_important_sources!
source_term = aux.scratch.tmp
(; stored_variable) = aux.microph_variables
@. source_term = some_micro_foo(stored_variable)
@. aux.important_sources = to_sources(-source_term, source_term)
end
@inline function important_sources_tendency!
precompute_aux_important_sources!
@. dY.ρq_tot += aux.thermo_variables.ρ * aux.important_sources.q_tot
end
It is more complicated for Cloudy than the other microphysics models. There are software office hours on Wednesday. If we can't figure it out by then, maybe we could submit it as a request to help then?
Thank you for adding Cloudy to KiD! |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #159 +/- ##
==========================================
- Coverage 81.13% 72.34% -8.79%
==========================================
Files 28 28
Lines 1834 1884 +50
==========================================
- Hits 1488 1363 -125
- Misses 346 521 +175 ☔ View full report in Codecov by Sentry. |
src/K1DModel/tendency.jl
Outdated
|
|
||
| aerosol_distribution = | ||
| CMAM.AerosolDistribution((CMAM.Mode_κ(r_dry, std_dry, N_aer, (FT(1),), (FT(1),), (FT(0),), (κ,)),)) | ||
| N_act = CMAA.N_activated_per_mode(activation_params, aerosol_distribution, air_params, thermo_params, T, p, w, q)[1] |
There was a problem hiding this comment.
This is out of date with the aerosol activation PR that I merged, apologies!
There was a problem hiding this comment.
Can we use this one: CMAA.total_N_activated ? Or does it have to be only one aerosol mode for Cloudy right now?
Purpose
To-do
Content