wave breaking not clear
clear wave break
clear wave break
Test 1: L120, change the tropopause from 20 km to 10 km:
crashes due to negative moisture layer depth near the model top
Test 2: change tau of the Rayleigh fraction layer from 5 to 2.5:
crashes due to the same reason
Test 3: change tau from 5 into 10:
success
wave breaks @ ~20 km
wave breaks @ ~25 km
interesting features:
why doesn't the gravity wave break right above the tropopause @ 10 km?
might be an issue of not enough vertical resolution, can test by adding the vertical layers
then why does the gravity wave break @ 25 km?
might dut to the vertical wind shear, can test by adjusting the wind filed settings
Test 1: L120, double the lapse rate in troposphere and stratosphere
crashes due to negative moisture layer depth near the surface
Test 2: double se_nsplit:
success
wave persists after 5 days
wave dissipates soon
Test 1: L120, double the lapse rate in only stratosphere
crashes due to negative mositure layer depth near the model top
Test 2: change tau from 5 to 10:
success
no obvious difference
CTL vs. blocky mountain (Jishi Zhang)
default ne30: PHIS, U, OMEGA
blocky mountain: PHIS, U, OMEGA
mlon_width = 14.0 deg -> 8.0 deg
nord = 2 -> 20 in (lon_i-mlon/mlon_scale)**nord
U of default ne30
U of blocky mountain
u vs. omega cross section of default ne30
u vs. omega cross section of blocky mountain
def
ne30
Ri, Fr, N2, du_dz**2
blocky
Ri, Fr, N2, du_dz**2
the flow behavior changed a lot in this case, like the zonal and vertical winds. The stronger insteability is also reflected in the Froude number near the foot of the mountain, and the stronger u wind shear in the lower layers. If we futher double the height of this blocky mountain, which basically lowers the Froude number further, we got crash, which is great
blocky + higher mountain:
mlon_width = 14.0 deg -> 8.0 deg
nord = 2 -> 20 in (lon_i-mlon/mlon_scale)**nord
mountain_amplitude = 4000 m -> 8000 m
We hit the instability crash 😄
negative moist layer thickness at nstep = 351
doesn't crash, but numerical noise in the stratosphere
Test 2: Removal of Laplacian layer and Strengthening of Rayleigh layer (changes: tau = 2 → 5 days, layer depth = 15. → 20. km):
crashes due to negative moisture layer depth near the model top
Test 3: Strengthening of Rayleigh layer and upgrade physics coupling from 30 to 10 mins (ATM_NCPL=144):
success
Is it convenient to use just one sponge layer (Rayleigh Friction)?
similar results to the base case, but
takes more than 3x the runtime
does not reduce (probably slightly increases) the noise at the top of the model
Diffusion is 8x on the right, default on the left
By adding noise to the initial condition, it's possible to create an ensemble mean.
Does the power spectra degradation from diffusion match the power spectra of the ensemble mean?
Left: Ensemble Mean and Spread from 1percent perturbation