<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7652.24">
<TITLE>FW: [Wrf-users] Experiences with high-resolution WRF-ARW</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<BR>
<BR>
<P><FONT SIZE=2>I'd like to thank everyone who responded to my e-mail. There were a lot of<BR>
suggestions made, and I thought it best to respond all at once.<BR>
<BR>
To Nicole Molders: I am using the WSM5 microphysics scheme with this domain.<BR>
I will check to see if heavy precipitation occurs.<BR>
<BR>
To Sen Chiao: I'm using 137 vertical levels. The first 40 levels are in the<BR>
lowest 500 meters (dz ~ 12.5 meters); the next 12 km contain 73 vertical<BR>
levels, with dz varying from ~12.5 meters to ~304 meters using a hyperbolic<BR>
tangent stretching function. The remaining 24 vertical levels fill the<BR>
domain up to 50 mb. (These vertical resolutions assume the U.S. Standard<BR>
Atmosphere and a reference terrain height of 1631 meters.)<BR>
<BR>
To Joe Galewsky: I have also run WRF using dz = 50 m in the lowest 2 km,<BR>
then dz = 125 m for the next 12 km, then 500 m up to 50 mb. Our time step<BR>
for this was 2 sec. I made some attempts at increasing the horizontal<BR>
resolution to 0.5 km but experienced CFL errors.<BR>
<BR>
To Chris Walker: I am using the Mellor-Yamada-Janjic PBL scheme, but<BR>
with the ratio Km/Kh changed to use the equation from Kondo et al.<BR>
(J. Atmos. Sci., 1978) for stable conditions. The 2D Smagorinsky<BR>
scheme is used for horizontal deformation. I originally tried using the<BR>
3D TKE scheme (km_opt = 2), but this does not work with the land surface<BR>
model in WRF 2.2; we may try again with WRF 3.0. I will investigate the<BR>
vertical diffusive CFL number.<BR>
<BR>
To David Ovens: I am using the 30 arcsec global terrain available with<BR>
WRF. The GEOGRID.TBL.ARW file is standard, and specifies:<BR>
* 1 smooth pass<BR>
* the smth-desmth_special smoothing option<BR>
* average_gcell(4.0)+four_pt+average_4pt interpolation<BR>
The documentation for the smoothing and interpolation options is poor, so<BR>
I need to look at the code to see what exactly is going on.<BR>
<BR>
Thanks again!<BR>
<BR>
-Eric<BR>
<BR>
Eric M. Kemp<BR>
Meteorologist<BR>
Northrop Grumman Information Technology<BR>
Intelligence Group (TASC)<BR>
4801 Stonecroft Boulevard<BR>
Chantilly, VA 20151<BR>
(703) 633-8300 x7078 (lab)<BR>
(703) 633-8300 x8278 (office)<BR>
(703) 449-3400 (fax)<BR>
eric.kemp@ngc.com<BR>
<BR>
<BR>
<BR>
-----Original Message-----<BR>
From: wrf-users-bounces@ucar.edu on behalf of Kemp, Eric M.<BR>
Sent: Wed 2/27/2008 4:01 PM<BR>
To: wrf-users@ucar.edu<BR>
Subject: [Wrf-users] Experiences with high-resolution WRF-ARW<BR>
<BR>
<BR>
Dear wrf-users:<BR>
<BR>
My group is running WRF-ARW at high resolution (dx = 1 km) to simulate turbulence. <BR>
Recently we tried running at very high vertical resolution (dz ~= 12.5 m in the lowest<BR>
500 m of the domain), and encountered CFL errors with time steps as<BR>
low as 1/8 of a second. These runs used the 6th order diffusion<BR>
option from Knievel et al (2007), along with real terrain (area around<BR>
Albuquerque, NM) and physics (MYJ TKE, Noah land surface,<BR>
RRTM/Dudhia radiation, and WSM5 microphysics).<BR>
<BR>
Has anyone here run WRF at such fine resolution, and if so, what were<BR>
your experiences with the time step? The ARW documentation recommends<BR>
6 seconds for each kilometer of dx, ignoring the effect of vertical resolution. <BR>
Are there other settings that should be tuned?<BR>
<BR>
-Eric<BR>
<BR>
Eric M. Kemp<BR>
Meteorologist<BR>
Northrop Grumman Information Technology<BR>
Intelligence Group (TASC)<BR>
4801 Stonecroft Boulevard<BR>
Chantilly, VA 20151<BR>
(703) 633-8300 x7078 (lab)<BR>
(703) 633-8300 x8278 (office)<BR>
(703) 449-3400 (fax)<BR>
eric.kemp@ngc.com<BR>
<BR>
<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>