[Wrf-users] Near Surface Temperatures

[Sam Drinkard]

List:

    I've got WRF v.2.1 up and running real data, but I am not impressed 
with near surface or surface temperatures.  I typically see in my 
domain(s) anywhere from 5, to as much as 10 degrees difference between 
observed and forecast.  I'm curious if this is something everyone has to 
deal with, or if perhaps something in my setup is causing these 
discrepancies.  The largest temp difference usually occurs in the 
afternoons, and WRF is running on average, 7-9 degrees too cool.  In the 
mornings, the model is not picking up on the low temps either, but 
errors on the order of 2-3 degrees are common.  I can live with that.  
The current setup for the inner domain is 12km, 34 levels, the lower 
ones being 1.000, 0.999, 0.997, and 0.994.  The rest above that are 
defaults from SI.  The model is initialized via the NAM. 

As for physics, the run is set with::

mp_physics=5 (Ferrier (new ETA))
ra_lw = 1, (rrtm scheme), radt = 33
sf_sfclay_physics = 2 (Monin-Obukhov)
sf_surface_physics = 2 (Noah LSM)
bl_pbl_physics = 2 (Mellor_Yamada_Janjic)
cu_physics = 1 (Kain-Fritsch)

I've played around with various schemes trying to find a mixture of 
things that would help bring surface temps in line with where they 
should be, but no combinations I've tried have made any significant 
difference.  About the only thing I can see that does appear to be 
mostly correct is the surface temp of the Clark Hill lake, approx 15nm 
north of me.  I do see as much as 5-7 degrees difference at times over 
water, vs land. 

I'd appreciate very much if someone has some suggestions to try or 
anything that might help get surface temps in line.

Graciously thanking you,


Sam