Sarath Guttikunda
New member
We are using GFS/GDAS at 0.25 degree resolution as primary input for WRF simulations in forecast mode for the next 3-4 days.
While the regional simulations at 20 and 25 km resolutions are looking good, the precipitation patterns for coastal regions when downscaled to 3km are consistently showing under predictions (we are using 2-domain configuration at 15km and 3km with grid and time step ratio of 5). For precipitation, if we use the coarser grid resolution results, the numbers and patterns are looking reasonable.
Area is west coast India - with heavy monsoonal rains this season (2023-July).
WRF version is 4.5
Namelist components from 15-3km simulation
This namelist is using WSM6 for mp_phsyics (=6), 3dTKE for PBL (=0), and KF for cu_phsyics (=1)
We have tried a couple of other combinations. The numbers did increase with mp_physics = 8 (Thompson), but not as much as expected and simulation time increased 20-30%.
Wondering if any of the users have experienced similar situations or tried anything different for coastal applications to make these results better.
Thank you.
While the regional simulations at 20 and 25 km resolutions are looking good, the precipitation patterns for coastal regions when downscaled to 3km are consistently showing under predictions (we are using 2-domain configuration at 15km and 3km with grid and time step ratio of 5). For precipitation, if we use the coarser grid resolution results, the numbers and patterns are looking reasonable.
Area is west coast India - with heavy monsoonal rains this season (2023-July).
WRF version is 4.5
Namelist components from 15-3km simulation
&physics
mp_physics = 6, 6, 6, 6,
ra_lw_physics = 4, 4, 4, 4,
ra_sw_physics = 4, 4, 4, 4,
radt = 5, 5, 5, 10,
swint_opt = 1,
sf_sfclay_physics =91, 91, 91, 91,
sf_surface_physics = 2, 2, 2, 2,
sf_urban_physics = 0, 0, 0, 0,
bl_pbl_physics = 0, 0, 0, 0,
bldt = 0, 0, 0, 0,
surface_input_source = 1,
num_soil_layers = 4,
num_land_cat = 21,
cu_physics = 1, 0, 0, 0,
cudt = 5, 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 = 0,
icloud = 1,
sst_update = 1,
slope_rad = 0, 0, 0, 1,
topo_shading = 0, 0, 0, 1,
/
&dynamics
w_damping = 0,
hybrid_opt = 0,
diff_opt = 2, 2, 2, 2,
km_opt = 5, 5, 5, 5,
diff_6th_opt = 0, 0, 0, 0,
diff_6th_factor = 0.12, 0.12, 0.12, 0.12,
base_temp = 290.
damp_opt = 3,
zdamp = 5000., 5000., 5000., 5000.,
dampcoef = 0.2, 0.2, 0.2, 0.2,
khdif = 0, 0, 0, 0,
kvdif = 0, 0, 0, 0,
non_hydrostatic = .true., .true., .true., .true.,
tke_adv_opt = 1, 1, 1, 1,
moist_adv_opt = 1, 1, 1, 1,
scalar_adv_opt = 1, 1, 1, 1,
/
This namelist is using WSM6 for mp_phsyics (=6), 3dTKE for PBL (=0), and KF for cu_phsyics (=1)
We have tried a couple of other combinations. The numbers did increase with mp_physics = 8 (Thompson), but not as much as expected and simulation time increased 20-30%.
Wondering if any of the users have experienced similar situations or tried anything different for coastal applications to make these results better.
Thank you.