wet-bias (high precipitation) over a tropical domain

S

Sarath Guttikunda

New member
I have a wet-bias (high precipitation) in the second domain (urban).

WRF 4.5
d01 = 100x100
d02 = 111x111
15km x 3km
GF,YSU,Thompsom
cu_physics turned off for 3-km domain
sf_surface option = 2, noah lsm

I activated shcu_physics for 15-km domain to control heat/moisture buildup, and it lowered the numbers.

I still have high precip over Oct-Nov months.

One suggestion was to activate
sf_urban_physics for 3-km domain
-- however, simulation stopped on its own

Any suggestions/corrections to get sf_urban_physics option to work.
Most test runs stopped at time step of 6 hours (GDAS input interval)

*************

&physics
mp_physics = 8, 8, 8, 8,
ra_lw_physics = 4, 4, 4, 4,
ra_sw_physics = 4, 4, 4, 4,
radt = 15, 3, 10, 10,
swint_opt = 1,
sf_sfclay_physics = 1, 1, 1, 1,
sf_surface_physics = 2, 2, 2, 2,
sf_urban_physics = 0, 0, 0, 0,
surface_input_source = 1,
num_soil_layers = 4,
num_land_cat = 21,
bl_pbl_physics = 1, 1, 0, 0,
bldt = 0, 0, 0, 0,
topo_wind = 1, 1, 0, 0,
cu_physics = 3, 0, 0, 0,
shcu_physics = 3, 3, 0, 0,
cudt = 0, 0, 0, 0,
cugd_avedx = 1,
cu_rad_feedback = .true.,.false.,.false.,.false.,
prec_acc_dt = 60., 60., 60., 60.,
maxiens = 1,
maxens = 3,
maxens2 = 3,
maxens3 = 16,
ensdim = 144,
isfflx = 1,
ifsnow = 1,
icloud = 1,
sst_update = 1,
slope_rad = 0, 1, 1, 1,
topo_shading = 0, 1, 1, 1,
/

When activating sf_urban, I used this

sf_urban_physics = 0, 1, 0, 0,
mosaic_lu = 1,
mosaic_soil = 1,

Without sf_urban_physics, the three lines above, the simulation works for the set full period.
 
First of all I would suggest to incrase domain sizes. With 5:1 nest ratio and such small domains you might have some unwanted boundary effects within most of your d02. Other ideas, turn on cu_physics for d02 - depending on dynamics and possible orographic effects, without convective parameterization that removes instability when precipitation is produced, you might have significant wet bias as a result.
 
D02 is well inside 30% buffer on all four sides.
I will test a bigger domain.

Suggestion is to not use cu_physics for domains with grid sizes as small as 3-km.
Terrain is hilly. I will test by activating cu_physics = 3,3 for both the domains.
Any case studies which reported success with cu_physics for d02?

Thank you
 
The suggestion to avoid cu_physics on fine grids has its root from the past when there were no scale-aware schemes existing, and KF or BMJ schemes were the best we had. They supress any convection resolved by model and thus are unusable at fine grids. However, with the inclusion of scale-aware schemes like MSKF or Grell Freitas, the situation is different; they can and should be used on finer grids, up to certain point when they completely back off due to small grid sizes, at that point they just add to CPU cycles and do simply nothing. On 3 km they are still very usable. Particularly now as you confirmed sloped terrain which almost certanly contributes to forced lifting and wet bias if instability is not controlled.
 
Please upload your full namelist.inout for me to take a look.

Have you ever checked precip in your D01? Is it overestimated, too?

I agree with (@meteoadriatic) that you should turn on cumulus scheme in both domains with a scale-aware scheme.
 
You must log in or register to reply here.
Back
Top