[RegCNET] =?x-gbk?Q?Re=A3=BARe=3A__problem_of_GrADSN?= =?x-gbk?Q?cPlot?=

[Graziano Giuliani]

On 06/15/2011 06:59 AM, suxingtao at sina.com wrote:
> Dear Dr. Graziano,
> I am sorry to disturb you. I have download latest GrADS version from IGES,
 > but it still does't work.
> The version of GrADS is grads-2.0a8. Also I compiled GrADS from source with
 > netCDF I/O format support, but it failed too. Why the format of output results
 > is not same to RegCM4.0? If I want to get the same format results to RegCM4.0,
 > what I need to modify?
[...]

If the answer from GrADS is "dtype netcdf not supported" I will trust it. One 
possible source for this error is that You have multiple versions of GrADS with 
different options enabled. The plot helper for GrADS does call just the first 
grads in the path, does not know about the others if they are not in the path.
Type in at grads prompt the

q config

and see if in the answer You see these lines:

   o NetCDF interface ENABLED
       http://www.unidata.ucar.edu/software/netcdf

if not the case (DISABLED appears) then You need to double check where the 
correct grads executable is on Your system, either replacing the faulty one or 
hardcoding the correct path in the helper program fortran code. Note that the 
latest GrADS version is 2.0a9. Identify the GrADS program You are running with 
command:

which grads

and decide a strategy (replace or change order in the PATH variable or hardcode 
correct path in code).

About the format choice itself, we have switched to netCDF for a series of reasons:

1) it is MANDATORY in the CORDEX experiment exchange format specifications.
2) the old binary format was missing capability to store metadata needed to 
correctly define the model, the dataset and the configuration, was not portable, 
was not indexed, cannot be transparently compressed, did support in GrADS just 2 
projections, cannot be directly used for distribution tools like THREDDS.
3) we did provide some help just for GrADS (now You can use multiple 
plot/analysis tools)
4) it was really cumbersome to maintain both I/O format, we decided to support 
just the one we NEED to implement and by any reasonable metric the best one.

This said, You nevertheless can obtain from netCDF the binary format:

ncks -B -b MYEXP_ATM.YYYYMMDD00 MYEXP_ATM.YYYYMMDD00.nc -o foo.nc

but this leaves You the burden to write the ctl file.

No option in the model itself to support the old binary format is left.

Ciao!

Graziano.

-- 
Graziano Giuliani - Earth System Physics Section
The Abdus Salam International Centre for Theoretical Physics
Strada Costiera, 11 - I - 34151 Trieste Italy