Page 1 of 1

Innermost (horizontal) gridlength 22 m: divide by zero

Posted: Fri Jan 05, 2018 7:03 pm
by j1d1w1
I'm hoping to use UEMS to diagnose the near-ground wind field over a domain of about a square kilometer with a horizontal resolution of the order of 20 m; this innermost domain (no. 3) is nested within a domain 2 that (in turn, in the usual way) is nested within domain 1, where the outermost domain (domain 1) covers about 20 by 20 km with gridlength 3 x (3 x 20) = 180 m. I'm using the ARW core, and the lat-lon grid. The goal is to simply spin up the fields over 6 hours onto the fine mesh, using an analysis or reanalysis field for initial and boundary conditions. Eventually, i.e. if I can get this to "go", I will provide high resolution (lidar-derived) topography for domain 3.

In my experience dwiz has proven worse than useless to create such a domain, or even to view such a domain that one had created via the command line.

But sidestepping that point, I have used ems_domain.pl to create my domain, adjusted the namelist.wpl file to reflect what I want, then localized successfully. But when I run (ems_run), I immediately get the error: "Illegal division by zero at .../NLphysics.pm line 4412" and that line reads

Code: Select all

my@opts=();
Any suggestions would be appreciated. I'm not trying LES, and have the YSU boundary-layer scheme enabled. No Cu parameterization. Thanks, John

Re: Innermost (horizontal) gridlength 22 m: divide by zero

Posted: Sat Jan 06, 2018 7:46 pm
by Opsasy
Try to change time step for 60 (seconds). It worked in my case. If I used time step equal to more than 60 (e. g. 120) or Auto_R or Auto_S the I had the same error :)

Re: Innermost (horizontal) gridlength 22 m: divide by zero

Posted: Sat Jan 06, 2018 8:45 pm
by Opsasy
You should also change radt value for more or equal 1.

Re: Innermost (horizontal) gridlength 22 m: divide by zero

Posted: Mon Jan 08, 2018 3:06 am
by j1d1w1
Thanks -- I'll play with that. I had set TIME_STEP = AUTO_S and UEMS had chosen "a large timestep of 0.5 seconds". Maybe I'll prescribe an even smaller value. John