[Wrf-users] differences in wrf results due to grib2 subsetting

[Eduardo Penabad Ramos]

Hello!
 
in our institution, in order to speed up some operational tasks, we are
starting to work with regional subsets of global 0.5-GFS GRIB2 files as
inital/boundary conditions for WRF (we are using "small_grib" option of
wgrib2 to generate our regional subset) as you can observe in the attached
document
 
We have found some significant differences in WRF results due to use of
those regional files instead of global files (as it can be seen in this
document:
<http://www.meteogalicia.es/galego/modelos/numerico/edu/dif_grib2.pdf>
http://www.meteogalicia.es/galego/modelos/numerico/edu/dif_grib2.pdf).
Differences grow up to 2 degrees in 2m-temperature or 2 m/s in 10m-winds at
the end of the simulation (84/96 hour forecast).
 
We've contacted with people at NCEP related to grib2 processing, and they
said that this is normal, wgrib2 just "generates" some truncation errors,
and that those differences in WRF results are similar to those obtained
using the same initial conditions in differente machines/compilers.
 
Even though, it seems that the differences are really important
(non-negligible)... does anyone has faced with this kind of problem? any
suggestions?
 
Thanks in advance!
 
Eduardo Penabad
MeteoGalicia. NWP Department
 
 
 <http://www.meteogalicia.es/galego/modelos/numerico/edu/dif_grib2.pdf>  
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/wrf-users/attachments/20090401/e94aa97f/attachment.html