[gembud] GEMPAK going away in 3-5 years? (IDV discussion)

[Lloyd A Treinish]

I can't help but comment that these sorts of issues have been around for a 
long time as well as a solution.  A host of artifacts are likely with such 
grids from WRF and other NWP codes for a variety of reasons, which I can 
outline.  Granted, the impact will depend on the specific model 
configuration and data (e.g., wind field near the surface in a 
high-resolution nest vs. a moisture field in an upper layer at low 
resolution with radiation processes dominating).  Some are introduced by 
the conversion process while others by the choice of contouring or other 
realization algorithm.  They are exacerbated by nested grids.  We 
developed a solution to this about 17 years ago in Data Explorer, which 
has been available in open source since 1999 (www.opendx.org).  You simply 
operate directly on the native grids and bypass these issues (i.e., 
because they are directly supported by the internal data model and 
processing algorithms).  I've attached a very simple example, which is 
screen dump from a protoype WRF-ARW viewer I built a few years ago with 
DX.  The 3d viewing coordinate system is also native to WRF.  In addition 
to avoiding the various artifacts, the intermediate processing step is 
unnecessary.



--------------------------
Lloyd A. Treinish
Project Scientist, Big Green Innovations
IBM Systems & Technology Group
1101 Kitchawan Road, Yorktown Heights, NY 10598
914-945-2770, lloydt at us.ibm.com
http://www.research.ibm.com/people/l/lloydt/
http://www.research.ibm.com/weather/DT.html
http://www.ibm.com/technology/greeninnovations



Don Murray <dmurray at unidata.ucar.edu> 
Sent by: gembud-bounces at unidata.ucar.edu
01/11/2008 04:06 PM
Please respond to
dmurray at unidata.ucar.edu


To
Dr Robert Pasken <rpasken at eas.slu.edu>
cc
gembud at unidata.ucar.edu
Subject
Re: [gembud] GEMPAK going away in 3-5 years?






Dr Pasken-

Dr Robert Pasken wrote:

> On the other hand I have not been impressed with IDV. The conversion of 
> the compiled version of vis5d to the JAVA based IDV has resulted in a 
huge 
> performance hit. 

FYI, the IDV is not a conversion of the Vis5D code to Java although
some of the Vis5D code (e.g. the contour algorithm) was moved to
Java in the VisAD library.  We borrowed ideas from Vis5D in
the IDV development.

Regardless of the hardware or OS release the performance
> of IDV is really bad. As an example I have a run from WRF with 
dimensions 
> of 90x90x41 vis5d will have a streamline field up before IDV has 
finished 
> loading. There are some interesting features in IDV but ther performance 

> hit isn't worth it.

Are you loading the Vis5D file into the IDV or the WRF staggered
netCDF grid?  There are known issues with using the native
WRF netCDF grids in the IDV which is why we suggest converting
to pressure coordinates through WRF-Post or converting to
Vis5D files.  Could you share a sample file?  Also, which
version of the IDV are you using?  You can answer off the list
if you wish since this isn't GEMPAK related.

Don
*************************************************************
Don Murray                               UCAR Unidata Program
dmurray at unidata.ucar.edu                        P.O. Box 3000
(303) 497-8628                              Boulder, CO 80307
http://www.unidata.ucar.edu/staff/donm
*************************************************************

_______________________________________________
gembud mailing list
gembud at unidata.ucar.edu
For list information or to unsubscribe,  visit: 
http://www.unidata.ucar.edu/mailing_lists/ 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.unidata.ucar.edu/mailing_lists/archives/gembud/attachments/20080118/6f23b6d9/attachment-0001.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: WRF_viewer.png
Type: application/octet-stream
Size: 132755 bytes
Desc: not available
Url : http://mailman.unidata.ucar.edu/mailing_lists/archives/gembud/attachments/20080118/6f23b6d9/attachment-0001.png