Add cloud top height diagnostic #4023
Conversation
oalcabes
force-pushed
the
oa/cltz_diag
branch
5 times, most recently
from
5b0819b to
afbf27f
Compare
src/diagnostics/core_diagnostics.jl
Outdated
|
|
||
| ct_constants = CAP.microphysics_cloud_params(cache.params).ct_constants | ||
|
|
||
| q_cond = @. lazy(clw + cli) |
There was a problem hiding this comment.
| q_cond = @. lazy(clw + cli) | |
| # Condensate density (kg/m^3) | |
| q_cond = @. lazy(state.c.ρ * (q_liq + q_ice)) |
There was a problem hiding this comment.
@tapios why do you want this to be per volume? (Doing it this way would require changing the threshold and k constants to also be per volume, which I wasn't initially doing)
There was a problem hiding this comment.
It's the same you had before, isn't it? There were two multiplications by density in l. 1133/1134 before, and I replaced by the one multiplication here. I was wondering why it was there but thought you had reasons for it. It does make more sense to me to have the thresholds be on specific humidity (i.e., by mass).
There was a problem hiding this comment.
Oops you're right! I'll just remove it.
oalcabes
force-pushed
the
oa/cltz_diag
branch
4 times, most recently
from
05d339b to
28fbf53
Compare
oalcabes
force-pushed
the
oa/cltz_diag
branch
from
28fbf53 to
2ab1cf9
Compare
| ct_constants = CAP.microphysics_cloud_params(cache.params).ct_constants | ||
|
|
||
| # Condensate density (kg/m^3) | ||
| q_cond = @. lazy(state.c.ρ * (q_liq + q_ice)) |
There was a problem hiding this comment.
It's better to show densities (per volume) with something else maybe L or ρq
| # 2. Create a numerically stabilized exponential weight to favor higher altitudes | ||
| az = @. lazy(ct_constants.a * z) | ||
| max_az = maximum(az) # This prevents overflow in the exp() call | ||
| exp_az_stabilized = @. lazy(exp(az - max_az)) |
There was a problem hiding this comment.
The parameter a is dimensional (1/m) unless the variable exp_az_stabilized is given by exp(a(z/zmax - 1))
| q_cond = @. lazy(state.c.ρ * (q_liq + q_ice)) | ||
|
|
||
| # 1. Create the "cloudiness" mask using the sigmoid function | ||
| w = @. lazy(1 / (1 + exp(-(state.c.ρ * ct_constants.k) * (q_cond - (state.state.c.ρ * ct_constants.thresh))))) |
There was a problem hiding this comment.
You can also try multiplying the mask by condensate mass signal itself to get rid of noise.
There was a problem hiding this comment.
Also, here k is dimensional (m^6) too.
3 participants
This diagnostic calculates a resolution invariant cloud top height. It is sensitive to three parameters that have been added to ClimaParams -- "cloud_top_threshold," "cloud_top_k," and "cloud_top_a." Results are sensitive to parameter choices. I hand tuned to get somewhat decent default parameters with these results:
