Page 1 of 1

2nd-order advection causes boundary artifacts

PostPosted: Fri Jul 17, 2015 7:06 am
by Corey Potvin
Using a modified version of WRF 3.6.1, compiled for the quarter_ss case, with the widely used Del City sounding, I found that setting the horizontal and vertical momentum and scalar advection orders to 2 caused reflectivity to be generated at the northern and southern boundaries immediately once the model begins integrating.

I reproduced this problem using vanilla WRF 3.6.1, compiled for the quarter_ss case, with the default input_sounding and namelist.input. By t=87 min, artifacts appeared on the southern boundary, shortly followed by all model fields being set to zero or ~10e36 (though the model kept running). Using larger grids causes the problem to manifest earlier. Using the default 5th-/3rd-order advection options removes the problem.

Lou Wicker and I are interested in a solution to this problem.

Thanks,

Corey Potvin, CIMMS/NSSL