Wind speed overestimation for domain over Canary Islands

Questions and other topics related to UEMS 15.
j0nes2k
Posts: 53
Joined: Wed Feb 03, 2010 5:06 pm

Wind speed overestimation for domain over Canary Islands

Post by j0nes2k » Mon Aug 08, 2016 11:19 am

Hello,

I have set up a domain over the Canary islands with 4km resolution and 138x64 points. I am using the following configuration settings:

- GFS 0.25 files
- Adaptive Timestep on
- CU_PHYSICS = 05
- RA_SW_PHYSICS = 02
- BL_PBL_PHYSICS = 1

I am observing a large overestimation of 10m wind speed forecast in my domain. I have the impression that this also may have to do with cloud cover forecasts. I am now testing TOPO_WIND = 1, and I have already tested RA_SW_PHYSICS = 01 and CU_PHYSICS = 0

Do you have any ideas where this might come from or what different configuration settings I should try?


Thank you for your help and best regards,

Jonas

j0nes2k
Posts: 53
Joined: Wed Feb 03, 2010 5:06 pm

Re: Wind speed overestimation for domain over Canary Islands

Post by j0nes2k » Fri Aug 12, 2016 2:49 pm

Here is another update: I have just visualized what is happening in my domain. You can see the animation output at: http://imgur.com/a/8ZGGE

I also ran some tests with changing exactly one physics parameter at a time (CU_PHYSICS, RA_SW_PHYSICS, BL_PBL_PHYSICS) and finally ran another test using the recommended settings from the WRF User's Guide. All of these settings show more or less the same errors.

Do you have any hints on how to debug this or where this might come from?

Best regards,

Jonas

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Wind speed overestimation for domain over Canary Islands

Post by meteoadriatic » Fri Aug 12, 2016 8:58 pm

First of all that domain is too small and as it has not parent one, boundary error are huge and simulation as such is of low quality.

However that should not be the reason of model go crazy on wind like that. Post here namelist.wrfm and namelist.wps files from static directory so that we can see whole configuration...

j0nes2k
Posts: 53
Joined: Wed Feb 03, 2010 5:06 pm

Re: Wind speed overestimation for domain over Canary Islands

Post by j0nes2k » Tue Aug 16, 2016 1:10 pm

Hello,

thank you for your answer! I know the domain is quite small (138x64 points @ 4km resolution). Do you think adding a larger parent domain would help in this case? I also assume that the error is either in the domain setup itself (as it is quite small) or something is injected from the boundary data (GFS 0.25 ptiles). I think I will have to test this with either the "full" GFS 0.25 files or GFS 0.5 files - and if this still leads to the same errors, I will have to enlarge the domain and see if the error persists.

The model is running stable since this issue, the run that caused the error is from 2016-08-08 at 06Z. That said, here is my namelist.wrfm from the current run:

Code: Select all

&time_control
 start_year                 = 2016
 start_month                = 08
 start_day                  = 16
 start_hour                 = 00
 start_minute               = 00
 start_second               = 00
 end_year                   = 2016
 end_month                  = 08
 end_day                    = 19
 end_hour                   = 06
 end_minute                 = 00
 end_second                 = 00
 interval_seconds           = 10800
 input_from_file            = T
 history_interval           = 60
 history_outname            = 'wrfout_d<domain>_<date>'
 frames_per_outfile         = 1
 io_form_history            = 2
 io_form_input              = 2
 io_form_restart            = 2
 io_form_boundary           = 2
 io_form_auxinput2          = 2
 restart                    = F
 output_ready_flag          = T
 auxhist1_interval          = 0
 auxinput4_inname           = "wrflowinp_d<domain>"
 auxinput4_interval         = 360
 io_form_auxinput4          = 2
 fine_input_stream          = 0
 adjust_output_times        = T
 use_netcdf_classic         = T
/

&domains
 time_step                  = 30
 time_step_fract_num        = 0
 time_step_fract_den        = 10
 time_step_dfi              = 60
 max_dom                    = 1
 s_we                       = 1
 e_we                       = 138
 s_sn                       = 1
 e_sn                       = 64
 s_vert                     = 1
 e_vert                     = 45
 dx                         = 4000.0000
 dy                         = 4000.0000
 grid_id                    = 1
 parent_id                  = 1
 i_parent_start             = 1
 j_parent_start             = 1
 parent_grid_ratio          = 1
 parent_time_step_ratio     = 1
 grid_allowed               = T
 max_dz                     = 1000.
 numtiles                   = 16
 nproc_x                    = -1
 nproc_y                    = -1
 hypsometric_opt            = 2
 num_metgrid_soil_levels    = 4
 num_metgrid_levels         = 32
 interp_type                = 2
 extrap_type                = 2
 t_extrap_type              = 2
 use_levels_below_ground    = T
 use_surface                = T
 lagrange_order             = 9
 zap_close_levels           = 500
 force_sfc_in_vinterp       = 2
 sfcp_to_sfcp               = T
 smooth_cg_topo             = T
 rh2qv_wrt_liquid           = T
 rh2qv_method               = 2
 p_top_requested            = 5000
 interp_method_type         = 2
 maxw_above_this_level      = 25000
 trop_horiz_pres_diff       = 7500
 maxw_horiz_pres_diff       = 7500
 use_maxw_level             = T
 use_trop_level             = T
 use_adaptive_time_step     = T
 step_to_output_time        = T
 target_cfl                 = 1.2
 target_hcfl                = 0.84
 max_step_increase_pct      = 5
 starting_time_step         = 20
 max_time_step              = 60
 min_time_step              = 12
/

&dfi_control
 dfi_opt                    = 0
/

&physics
 cu_physics                 = 5
 cugd_avedx                 = 3
 ishallow                   = 1
 cu_diag                    = 1
 convtrans_avglen_m         = 30
 mp_physics                 = 2
 do_radar_ref               = 1
 shcu_physics               = 0
 bl_pbl_physics             = 1
 bldt                       = 0
 grav_settling              = 0
 topo_wind                  = 1
 ysu_topdown_pblmix         = 0
 scalar_pblmix              = 1
 tracer_pblmix              = 1
 sf_sfclay_physics          = 1
 isfflx                     = 1
 sf_surface_physics         = 2
 num_land_cat               = 24
 num_soil_cat               = 16
 num_soil_layers            = 4
 surface_input_source       = 1
 usemonalb                  = T
 rdmaxalb                   = T
 rdlai2d                    = F
 ua_phys                    = T
 sf_surface_mosaic          = 1
 mosaic_cat                 = 3
 sf_urban_physics           = 0
 ra_lw_physics              = 1
 ra_sw_physics              = 2
 radt                       = 4
 swint_opt                  = 1
 ra_call_offset             = 0
 slope_rad                  = 0
 topo_shading               = 0
 cu_rad_feedback            = T
 icloud                     = 1
 sst_skin                   = 0
 seaice_threshold           = 271
 seaice_albedo_opt          = 1
 seaice_albedo_default      = 0.65
/

&noah_mp
/

&dynamics
 non_hydrostatic            = T
 rk_ord                     = 3
 h_mom_adv_order            = 5
 h_sca_adv_order            = 5
 v_mom_adv_order            = 3
 v_sca_adv_order            = 3
 moist_adv_opt              = 1
 moist_adv_dfi_opt          = 0
 scalar_adv_opt             = 1
 momentum_adv_opt           = 1
 tke_adv_opt                = 1
 diff_opt                   = 1
 km_opt                     = 4
 km_opt_dfi                 = 1
 w_damping                  = 1
 diff_6th_opt               = 0
 diff_6th_factor            = 0.25
 damp_opt                   = 0
 zdamp                      = 5000.
 time_step_sound            = 4
 smdiv                      = 0.1
 emdiv                      = 0.01
 epssm                      = 0.1
/

&scm
 scm_force                  = 0
/

&fdda
 grid_fdda                  = 0
/

&tc
 insert_bogus_storm         = F
/

&fire
/

&bdy_control
 spec_bdy_width             = 5
 spec_zone                  = 1
 relax_zone                 = 4
 spec_exp                   = 0
 specified                  = T
 nested                     = F
/

&stoch
 skebs                      = 0
/

&grib2
/

&namelist_quilt
 nio_tasks_per_group        = 0
 nio_groups                 = 1
/

&diags
 p_lev_diags                = 0
/

&afwa
 afwa_diag_opt              = 0
/

&logging
 compute_slaves_silent      = T
 io_servers_silent          = T
 stderr_logging             = 0
/
... and my namelist.wps:

Code: Select all

&share
 wrf_core                   = 'ARW'
 max_dom                    = 1
 start_date                 = '2016-08-16_00:00:00'
 end_date                   = '2016-08-19_06:00:00'
 interval_seconds           = 10800
 io_form_geogrid            = 2
 opt_output_from_geogrid_path = '/home/wrf/uems/runs/wrf_ic4_1/static'
 debug_level                = 0
/

&geogrid
 parent_id                  = 1
 parent_grid_ratio          = 1
 i_parent_start             = 1
 j_parent_start             = 1
 e_we                       = 138
 e_sn                       = 64
 geog_data_res              = '2m'
 dx                         = 4000
 dy                         = 4000
 map_proj                   = 'lambert'
 ref_lat                    = 28.539
 ref_lon                    = -15.772
 truelat1                   = 28.539
 truelat2                   = 28.539
 stand_lon                  = -15.772
 ref_x                      = 69
 ref_y                      = 32
 geog_data_path             = '/home/wrf/uems/data/geog'
 opt_geogrid_tbl_path       = '/home/wrf/uems/runs/wrf_ic4_1/static'
/

&ungrib
 out_format                 = 'WPS'
 prefix                     = 'FILE'
/

&metgrid
 fg_name                    = GFSP25
 io_form_metgrid            = 2
 opt_output_from_metgrid_path = '/home/wrf/uems/runs/wrf_ic4_1/wpsprd'
 opt_metgrid_tbl_path       = '/home/wrf/uems/runs/wrf_ic4_1'
 process_only_bdy           =
/

&mod_levs
 press_pa                   = 201300, 200100, 100000, 95000, 90000, 85000, 80000, 75000, 70000, 65000, 60000, 55000, 50000, 45000, 40000, 35000, 30000, 25000, 20000, 15000, 10000, 5000, 1000
/

&domain_wizard
 grib_data_path             = 'null'
 grib_vtable                = 'Vtable.GFS'
 dwiz_center_over_gmt       = true
 dwiz_desc                  = wrf_can4_1
 dwiz_gridpt_dist_km        = 4.0000000
 dwiz_latlon_linecolor      = -8355712
 dwiz_latlon_space_in_deg   = 10
 dwiz_map_horiz_scrollbar_pos = 7134
 dwiz_map_scale_pct         = 100.0
 dwiz_map_vert_scrollbar_pos = 2593
 dwiz_mpi_command           = null
 dwiz_name                  = wrf_ic4_1
 dwiz_show_political        = true
 dwiz_stand_lon             = -15.772
 dwiz_truelat1              = 28.539
 dwiz_truelat2              = 28.539
 dwiz_user_rect_x1          = 7274
 dwiz_user_rect_x2          = 7508
 dwiz_user_rect_y1          = 2718
 dwiz_user_rect_y2          = 2814
 dwiz_modis                 = false
 dwiz_tcvitals              = null
 dwiz_bigmap                = Y
 dwiz_lakes                 = false
/
Do you have any additional ideas where this might come from?

Best regards,

Jonas

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Wind speed overestimation for domain over Canary Islands

Post by meteoadriatic » Tue Aug 16, 2016 3:11 pm

j0nes2k wrote:The model is running stable since this issue, the run that caused the error is from 2016-08-08 at 06Z.
So, you say that there is no anymore described problem with wind? If that is the case then either there was fault in initial or boundary conditions, or something else... but in that case it would be good if you can rerun that suspicious run again to see what you will get this time, and if you get problem again, then try to fix that exact run first by locating the cause...

j0nes2k
Posts: 53
Joined: Wed Feb 03, 2010 5:06 pm

Re: Wind speed overestimation for domain over Canary Islands

Post by j0nes2k » Wed Aug 17, 2016 9:30 am

Hello,

this error persisted in about 5-6 runs around 2016-08-08. Since then, everything is working smoothly. By any chance, did you already encounter such issues with initial or boundary conditions?


Best regards,

Jonas

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Wind speed overestimation for domain over Canary Islands

Post by meteoadriatic » Wed Aug 17, 2016 9:54 am

There were problems with Robert's ptiles but they are corrected as far as I know. I also use nomads gribfilter which does fail to create proper tile from time to time. If full gribs are used, and downloaded successfully, no problems ever.

However, probably unrelated but, you still will have better forecast if you either: a) make domain larger, that is put domain border further away from your area of interest, or b) create parent domain around.

j0nes2k
Posts: 53
Joined: Wed Feb 03, 2010 5:06 pm

Re: Wind speed overestimation for domain over Canary Islands

Post by j0nes2k » Tue Sep 06, 2016 2:14 pm

Hello,

just getting back from vacation and getting back to this topic. Today I could confirm that this was indeed an error in the GFS 0.25 deg Personal Tile files. I have checked with the GFS 0.5 deg full files and GFS 0.5 deg Personal Tiles, and the simulation was running smoothly.

I would like to verify this also with the GFS 0.25 deg full files, but the available online archive does not date back until my desired date (2016-08-08). Are these grib files available anywhere else?

Best regards and thank you for your help!

Jonas

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Wind speed overestimation for domain over Canary Islands

Post by meteoadriatic » Tue Sep 06, 2016 2:21 pm

Hi,

Is this it? http://rda.ucar.edu/datasets/ds083.3/

Didn't tried it!

j0nes2k
Posts: 53
Joined: Wed Feb 03, 2010 5:06 pm

Re: Wind speed overestimation for domain over Canary Islands

Post by j0nes2k » Tue Sep 06, 2016 2:26 pm

Maybe this could work, thanks. Unfortunately this data is no longer available on NOMADS, so I will have to check if I can download the files manually and put it in the correct places for WRF to use them. Anyway, thank you for your help!

Best regards,

Jonas

Post Reply