Differing predictions with increasing numbers of processors

Discuss the nesting capability within the model itself and any problems you might have run into.

Differing predictions with increasing numbers of processors

Postby alanberry » Tue May 03, 2011 10:30 am

Hi all,

I am developing a WRFv3.2.1 one way nested model (9km ->3km ->1km) for the southwest coast of Ireland. Each domain is approx 165 x 165 x 30. I have ran a number of one week duration simulations to determine the optimum settings on our servers here and have encoutnered a problem. I was hoping someone could help me with.

Running the exact same simulation for the nested models results in different model predictions depending on the number of processors employed for the run. I have tested across 160, 320 & 480 processors to date (to get a handle on simulation times) and each simulation generates different predictions on each grid.

When only the parent (9km) model grid is used for the simulations, there are no differences between model predictions for the different numbers of processors.

Is this something that has been seen before? or am I missing something glaringly obvious in my input file? (Note no quilting is applied, and no specific options for parallel processing are set).

I'd appreciate any feedback the community can provide.

Namelist input file attached below.
Code: Select all
&time_control
run_days                 = 7,
run_hours                = 0,
run_minutes              = 0,
run_seconds              = 0,
start_year               = 2010,2010,2010,
start_month              = 09,09,09,
start_day                = 01,01,01,
start_hour               = 00,00,00,
start_minute             = 00,00,00,
start_second             = 00,00,00,
end_year                 = 2010,2010,2010,
end_month                = 09,09,09,
end_day                  = 08,08,08,
end_hour                 = 00,00,00,
end_minute               = 00,00,00,
end_second               = 00,00,00,
interval_seconds         = 10800,
input_from_file          = .true.,.false.,.false.,
history_interval         = 360,180,180,
frames_per_outfile       = 1,1,1,
restart                  = .false.,
restart_interval         = 10080,
io_form_history          = 2,
io_form_restart          = 2,
io_form_input            = 2,
io_form_boundary         = 2,
debug_level              = 0,
/

&domains                 
time_step                = 45,
time_step_fract_num      = 0,
time_step_fract_den      = 1,
max_dom                  = 3,
e_we                     = 162,163,169,
e_sn                     = 162,163,169,
e_vert                   = 30,30,30,
num_metgrid_levels       = 27,
dx                       = 9000,3000,1000,
dy                       = 9000,3000,1000,
grid_id                  = 1,2,3,
parent_id                = 1,1,2,
i_parent_start           = 1,54,55,
j_parent_start           = 1,54,56,
parent_grid_ratio        = 1,3,3,
parent_time_step_ratio   = 1,3,3,
feedback                 = 0,
smooth_option            = 0,
/

&physics                 
mp_physics               = 4,4,8,
ra_lw_physics            = 1,1,1,
ra_sw_physics            = 2,2,2,
radt                     = 10,10,10,
sf_sfclay_physics        = 2,2,2,
sf_surface_physics       = 2,2,2,
bl_pbl_physics           = 2,2,2,
bldt                     = 0,0,0,
cu_physics               = 5,5,0,
cudt                     = 0,0,0,
isfflx                   = 1,
ifsnow                   = 0,
icloud                   = 1,
surface_input_source     = 1,
num_soil_layers          = 4,
mp_zero_out              = 0,
maxiens                  = 1,
maxens                   = 3,
maxens2                  = 3,
maxens3                  = 16,
ensdim                   = 144,
/

&fdda                   
/

&dynamics               
w_damping                = 0,
diff_opt                 = 2,
km_opt                   = 2,
base_temp                = 290.,
damp_opt                 = 1,
zdamp                    = 5000.,5000.,5000.,
dampcoef                 = 0.01,0.01,0.01,
khdif                    = 0,0,0,
kvdif                    = 0,0,0,
smdiv                    = 0.1,0.1,0.1,
emdiv                    = 0.01,0.01,0.01,
epssm                    = 0.1,0.1,0.1,
time_step_sound          = 4,4,4,
h_mom_adv_order          = 5,5,5,
v_mom_adv_order          = 3,3,3,
h_sca_adv_order          = 5,5,5,
v_sca_adv_order          = 3,3,3,
non_hydrostatic          = .true.,.true.,.true.,
moist_adv_opt            = 1,1,1,
scalar_adv_opt           = 1,1,1,
tke_adv_opt              = 1,1,1,
/

&bdy_control             
spec_bdy_width           = 5,
spec_zone                = 1,
relax_zone               = 4,
specified                = .true.,.false.,.false.,
periodic_x               = .false.,.false.,.false.,
symmetric_xs             = .false.,.false.,.false.,
symmetric_xe             = .false.,.false.,.false.,
open_xs                  = .false.,.false.,.false.,
open_xe                  = .false.,.false.,.false.,
periodic_y               = .false.,.false.,.false.,
symmetric_ys             = .false.,.false.,.false.,
symmetric_ye             = .false.,.false.,.false.,
open_ys                  = .false.,.false.,.false.,
open_ye                  = .false.,.false.,.false.,
nested                   = .false.,.true.,.true.,
/


&dfi_control
dfi_opt                  = 0,
/


&grib2                   
/

&namelist_quilt         
nio_tasks_per_group      = 0,
nio_groups               = 1,
/
                     
Alan Berry
Oceanographic Services
Marine Institute
Oranmore
Co. Galway
Ireland
alanberry
 
Posts: 1
Joined: Wed Apr 20, 2011 3:14 am

Return to Nesting

Who is online

Users browsing this forum: No registered users and 2 guests