Relax zone importance for BC

[lslrsgis]

Dear WRF community,

As a beginning user of WRF, I would like to know:

(1) If spec_bdy_width = 0, what happens to the BC and final simulation? How important the relaxation zone is to improve BC?

(2) If spec_bdy_width = 5, does it mean all 5 pixels in the boundary (left right up bottom) should be removed for analysis? If so, is that the case for all nested outputs (d01,d02,d03...)?

Thanks.

LSL

 

[davegill]

Here is some code from share/module_check_a_mundo.F:

 261 !-----------------------------------------------------------------------
 262 ! LBC: spec_bdy_width = spec_zone + relax_zone
 263 !-----------------------------------------------------------------------
 264    IF ( model_config_rec%specified(1) ) THEN
 265       model_config_rec%spec_zone = 1
 266       model_config_rec%relax_zone = model_config_rec%spec_bdy_width - model_config_rec%spec_zone
 267    END IF

In WRF (for real data cases), the spec_zone is ALWAYS = 1.

The relaxation zone is the difference of the total width (spec_bdy_width) and the hard-coded outer row and column (spec_zone).

There are certainly arguments for excluding some boundary regions from statistical analyses of model output.

In most setups, the WRF model uses a regional domain, which implies that lateral boundary conditions are required. Theoretically, the model will fail with no boundary conditions. In the WRF implementation, this would probably be some sort of divide by zero (the width of the lateral boundary).

The nested domains utilize the same boundary nudging as on the most coarse domain.