To reduce the WRF-Chem simulation time

38 views
Skip to first unread message

Aparna C

unread,
May 8, 2025, 5:32:30 AMMay 8
to wrf-chem

Hi All,

I am running a WRF-Chem simulation for a duration of 5 days. However, it is taking nearly 20 hours to simulate each day's output. My current configuration is:

  • Grid spacing (dx) = 10 km

  • Time step = 45 seconds

  • Number of processors = 44

Can anyone suggest how I can speed up the model to generate the output files faster? 

Thanks in advance!


 &time_control
 run_days                            = 0,
 run_hours                           = 0,
 run_minutes                         = 0,
 run_seconds                         = 0,
 start_year                          = 2020, 2024,
 start_month                         = 01,   03,
 start_day                           = 11,   26,
 start_hour                          = 00,   00,
 end_year                            = 2020, 2024,
 end_month                           = 01,   06,
 end_day                             = 16,   05,
 end_hour                            = 00,   00,
 interval_seconds                    = 21600
 input_from_file                     = .true.,.true.,
 history_interval                    = 1440,  60,
 frames_per_outfile                  = 1, 1,
 restart                             = .false.,
 restart_interval                    = 7200,
 io_form_history                     = 2
 io_form_restart                     = 2
 io_form_input                       = 2
 io_form_boundary                    = 2
 debug_level                         = 100,
 history_outname                     = '/nfs3/aparna/MACC/wrf_new/wrf_out_macc_<domain>_<date>'
 rst_inname                          = '/nfs3/aparna/MACC/wrf_new/wrfrst__d<domain>_<date>'
 rst_outname                         = '/nfs3/aparna/MACC/wrf_new/wrfrst__d<domain>_<date>'

 io_form_auxinput5                   = 2,
 auxinput5_inname                    = 'wrfchemi_<hour>z_d<domain>'
 frames_per_auxinput5                = 12,
 auxinput5_interval_m                = 60,

 io_form_auxinput6                   = 2,
 auxinput6_inname                    = 'wrfbiochemi_d<domain>'
 auxinput6_interval_m                = 86400,

 io_form_auxinput7                   = 2,
 auxinput7_inname                    = 'wrffirechemi_d<domain>_<date>',
 auxinput7_interval_m                = 60,
 frames_per_auxinput7                = 1,
 force_use_old_data                  = .true.,
 /

 &domains
 time_step                           = 45,
 time_step_fract_num                 = 0,
 time_step_fract_den                 = 1,
 max_dom                             = 1,
 e_we                                = 420,    220,
 e_sn                                = 400,    214,
 e_vert                              = 34,     45,
 p_top_requested                     = 5000,
 num_metgrid_levels                  = 34,
 num_metgrid_soil_levels             = 4,
 dx                                  = 10000,
 dy                                  = 10000,
 grid_id                             = 1,     2,
 parent_id                           = 1,     1,
 i_parent_start                      = 1,     53,
 j_parent_start                      = 1,     25,
 parent_grid_ratio                   = 1,     3,
 parent_time_step_ratio              = 1,     3,
 feedback                            = 0,
 smooth_option                       = 0
 /

 &physics
 mp_physics                          = 10,    -1,
 cu_physics                          = 3,    -1,
 ra_lw_physics                       = 1,    -1,
 ra_sw_physics                       = 2,    -1,
 bl_pbl_physics                      = 1,    -1,
 sf_sfclay_physics                   = 1,    -1,
 sf_surface_physics                  = 2,    -1,
 radt                                = 30,    15,
 bldt                                = 0,     0,
 cudt                                = 0,     0,
 icloud                              = 1,
 num_land_cat                        = 21,
 sf_urban_physics                    = 1,     0,
 fractional_seaice                   = 1,
 shcu_physics                        = 3,
 progn                               = 1,
 cu_rad_feedback                     = .true.,
 cu_diag                             = 1,
 cugd_avedx                          = 1,
 mp_zero_out                         = 2,
 mp_zero_out_thresh                      = 1.e-8,
 sst_update                          = 0,
 /

 &fdda
 grid_fdda                           = 2,
 gfdda_inname                        = 'wrffdda_d<domain>,
 gfdda_interval_m                    = 360,
 gfdda_end_h                         = 9999,
 io_form_gfdda                       = 2,
 if_no_pbl_nudging_ph                = 1,
 if_no_pbl_nudging_t                 = 1,
 if_no_pbl_nudging_uv                = 1,
 if_zfac_uv                          = 0,
 gph                                 = 0.0003,
 gt                                  = 0.0003,
 guv                                 = 0.0003,
 xwavenum                            = 3,
 ywavenum                            = 3,
 /

 &dynamics
 hybrid_opt                          = 2,
 w_damping                           = 1,
 diff_opt                            = 1,      2,
 km_opt                              = 4,      4,
 diff_6th_opt                        = 0,      0,
 diff_6th_factor                     = 0.12,   0.12,
 base_temp                           = 290.
 damp_opt                            = 3,
 zdamp                               = 5000.,  5000.,
 dampcoef                            = 0.2,    0.2,
 khdif                               = 0,      0,
 kvdif                               = 0,      0,
 non_hydrostatic                     = .true., .true.,
 moist_adv_opt                       = 2,      1,
 scalar_adv_opt                      = 2,      1,
 gwd_opt                             = 1,      0,
 rk_ord                              = 3,
 chem_adv_opt                        = 2,
 tke_adv_opt                         = 2,
 time_step_sound                     = 4,
 h_mom_adv_order                     = 5,
 v_mom_adv_order                     = 3,
 h_sca_adv_order                     = 5,
 v_sca_adv_order                     = 3,
 /

 &bdy_control
 spec_bdy_width                      = 5,
 specified                           = .true.
 /

 &grib2
 /
 
 &chem
 kemit                               = 1,
 chem_opt                            = 112,
 bioemdt                             = 30,
 photdt                              = 30,
 chemdt                              = 5,
 io_style_emissions                  = 1,
 emiss_inpt_opt                      = 111,
 emiss_opt                           = 8,
 phot_opt                            = 1,
 gas_drydep_opt                      = 1,
 aer_drydep_opt                      = 1,
 bio_emiss_opt                       = 3,
 dust_opt                            = 0,
 dmsemis_opt                         = 0,
 seas_opt                            = 0,
 gas_bc_opt                          = 1,
 gas_ic_opt                          = 1,
 aer_bc_opt                          = 1,
 aer_ic_opt                          = 1,
 gaschem_onoff                       = 1,
 aerchem_onoff                       = 1,
 wetscav_onoff                       = 1,
 cldchem_onoff                       = 0,
 vertmix_onoff                       = 1,
 chem_conv_tr                        = 1,
 biomass_burn_opt                    = 2,
 plumerisefire_frq                   = 60,
 scale_fire_emiss                    = .true.,
 aer_ra_feedback                     = 1,
 have_bcs_chem                       = .true.,
 conv_tr_wetscav                     = 0,
 emiss_opt_vol                       = 0,
 aircraft_emiss_opt                  = 0,
 ne_area                             = 210,
 diagnostic_chem                     = 0,
 chemdiag                            = 1,
/


 &namelist_quilt
 nio_tasks_per_group = 0,
 nio_groups = 1,
 /

Mary Barth

unread,
May 8, 2025, 3:08:45 PMMay 8
to aparn...@gmail.com, wrf-chem
Hi Aparna,

Is there any way you can use more processors? With 420x400 grid points, I estimate that each of the 44 processors are assigned about a 62x62 patch of the domain, If you could use enough processors where the patch is 20x20 (or less), then that would be good. A quick estimate is ~350 processors. 

Your namelist shows that you are already taking steps to make the simulation faster with the radt, chem_dt, and photdt coarse compared to the meteorology time step. 

- Mary

^--^--^--^--^--^--^--^--^--^--^--^--^--^--^-^--^--^--^-^--^
Mary Barth 
Senior Scientist   
phone: 303-497-8186    email: bar...@ucar.edu

NSF - National Center for Atmospheric Research
P.O. Box 3000
Boulder, CO 80307

^--^--^--^--^--^--^--^--^--^--^--^--^--^--^-^--^--^--^-^--^


Reply all
Reply to author
Forward
0 new messages