remove grid-scale thermo state #4231
Conversation
szy21
force-pushed
the
zs/thermo_state
branch
27 times, most recently
from
c98b81f to
e95e753
Compare
szy21
force-pushed
the
zs/thermo_state
branch
2 times, most recently
from
4dfb4ea to
934de93
Compare
szy21
force-pushed
the
zs/thermo_state
branch
from
4fd4cd6 to
953e904
Compare
|
working atmos longrun: https://buildkite.com/clima/climaatmos-gpulongruns/builds/712 |
szy21
force-pushed
the
zs/thermo_state
branch
from
141522a to
e4892be
Compare
tapios
left a comment
tapios
left a comment
There was a problem hiding this comment.
LGTM. Amazing work, and huge lift. Thank you!
|
|
||
| # Saturation state at current thermodynamic state (via θ, p, q_tot) | ||
| ts_current = TD.PhaseEquil_pθq(thermo_params, p_c, θli, q_tot) | ||
| q_sat = TD.q_vap_saturation(thermo_params, ts_current) |
There was a problem hiding this comment.
It might be more straightforward to pass in temperature rather than pressure, and then compute q_sat and other quantities we need from that. Wouldn't that avoid use of PhaseEquil... here? (Same for the other cloud_fraction function above.)
Perhaps your intention was to do that, or something like it, in a separate PR, and that's fine too.
There was a problem hiding this comment.
Yes, I want to remove the PhaseEquil in the next PR, partially because I didn't know how to deal with this field of named tuple. But now I think I know:)
| end | ||
|
|
||
| function saturation_distance(q_tot, q_sat, ᶜts, θli, thermo_params, Δθli_fd) | ||
| function saturation_distance(q_tot, q_sat, p_c, θli, thermo_params, Δθli_fd) |
There was a problem hiding this comment.
Similar to above: I think this gets easier to change, and to reduce reliance on equilibrium constructors, by passing in temperature rather than pressure.
src/cache/precomputed_quantities.jl
Outdated
| @. ᶜts = ts_gs(thermo_args..., Y.c, ᶜK, ᶜΦ, Y.c.ρ) | ||
| @. ᶜp = TD.air_pressure(thermo_params, ᶜts) | ||
| # Compute thermodynamic state variables using individual getter functions. | ||
| # Note: For EquilMoistModel, this calls saturation_adjustment 3 times per grid point |
There was a problem hiding this comment.
See comment above: why not combining getters so that we call SA only once?
src/parameterized_tendencies/gravity_wave_drag/orographic_gravity_wave.jl
Show resolved
Hide resolved
szy21
force-pushed
the
zs/thermo_state
branch
6 times, most recently
from
fdd4b9a to
86cce9c
Compare
szy21
force-pushed
the
zs/thermo_state
branch
from
86cce9c to
fd8bef0
Compare
Purpose
Removes grid-scale thermo state and uses new Thermodynamics API.
Removes ts and adds ᶜT, ᶜq_tot_safe, ᶜq_liq_rai, ᶜq_ice_sno to the cache. Also adds h_tot back to the cache because using
@lazyresults in gpu error. The difference between ᶜq_tot_safe and ᶜq_tot (a lazy broadcasting) is ᶜq_tot_safe is clipped to 0. The difference between ᶜq_liq_rai, ᶜq_liq and ᶜq_rai is ᶜq_liq_rai = ᶜq_liq + ᶜq_rai and clipped to 0. Same for ᶜq_ice_sno. I am open to different names. ᶜT, ᶜq_tot_safe, ᶜq_liq_rai, ᶜq_ice_sno essentially create the previous thermo state. Clipping the humidities to 0 recovers the previous behavior of thermo state.I will remove subgrid thermo state and remaining TD.PhaseEquil in separate PRs.
To-do
Content