[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 19990127: grib to netcdf

On Thu, 11 Feb 1999, Melissa Richey wrote:

> 
> Hi Robb-
> 
> I have been looking at gribtonc using the debugger, so I thought I
> would let you know what I have found. I can't figure out why it is
> doing what it is doing, but I thought it might mean something to
> you... 
> 
> The function 'get_prod' controls the reading of one grib product into
> a buffer. In reading a normal grib product, 'look_for_mark'
> successfully finds the start of the product on the first
> try. (FOUND_START)
> It then successfully fills the rest of the buffer and finds the end of
> the product. (FOUND_END)
> For the bad mixing ratio product, it first finds the end of the
> product, then finds the start, then fills the data buffer before ever
> finding the end again, (NOT_FOUND) resulting in "oversize grib
> product."

Missy,

My first impression is that the grid for the water vapor mixing ratio is
somehow corrupted or the the grid immediately before is corrupted. Does
this happen on all your raw files, I would test it on more than one. 
After looking at the code, I wonder if the FOUND header state machine code
is parsing correctly wvmixr grid. We don't have a raw ruc2 presently, I
should have one tomorrow.  Then I can take a first hand look.

Robb...

ps Have you tried od (octal dump) on the ruc2 raw file? It could be a
sanity check.
> 
> Just my two cents worth.
> 
> Missy
> 
> 
> 
> + Missy,
> + 
> + That's what I started to do before I found out about the conflict of 185. 
> + The only suggestion I have is to use the brute force method, that is get a
> + raw grib file including param 185 and use a debugger on gribtonc. My
> + feeling gribtonc is trying to process 185 according to the NWS manner
> + instead of the FSL manner.  I'm not the originator of gribtonc and I have
> + not added/changed params before, so I would have to research this. I'll
> + look at the code today to see if there is something else. Also, as a guide
> + line I would look at how the other hybrid ruc2 variables are processed.
> + Have you changed the cdl wvmixr variable to match your modified gribtonc?
> + 
> + Just some thoughts,
> + Robb...
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================