[Thread Prev][Thread Next][Index]

Re: trouble with netcdf (?) data



Hi everybody,
I omitted to tell the list the resolutions of my troubles.
Here they are very briefly since it's nothing spectacular:
- the note from ferret:
 *** NOTE: too many values in attribute "missing_value" in netCDF file
variable: rh
resulted from the fact that the attribute "missing_value" was declared
as type character. Not my data - I swear! It did not occur to me that people
could do something like this and didn't pay enough attention to the output
of ncdump. However, Emilio had more sharp eyes and pointed out the quotes
around the 99999 like so: 
>         float rh(time, platform) ;
>                 rh:long_name = "relative humidity at 1.5m" ;
>                 rh:units = "percent" ;
>                 rh:missing_value = "-99999." ;
His comment was that he could imagine that this would cause problems. I go
with that.
- The real trouble was that the dimension "platfrom" was also defined as a 
  variable, but very unfortunately as a character variable to hold the 
  name of an instrument - not my data, really!

netcdf test {
dimensions:
        time = UNLIMITED ; // (144 currently)
        platform = 111 ;
        fldlen = 20 ;

Thus the designer of the data structure mixed up the number of platforms
with their actual names. ferret didn't like that because it expects in
description of the axis in the variable with the same name as the 
dimension - as I understand it.

Thus, the whole thing turned out to result from very bad data design
- imho - and didn't have anything to do with ferret itself.
Cheers,
Joerg
--
Jörg Kaduk                           phone:  +1 (650) 325 1521 ext 416
Carnegie Institution of Washington   fax:    +1 (650) 325 6857
Department of Plant Biology
260 Panama Street                    e-mail: joerg@jasper.stanford.edu   
USA - Stanford, CA 94305             


[Thread Prev][Thread Next][Index]

Dept of Commerce / NOAA / OAR / PMEL / TMAP

Contact Us | Privacy Policy | Disclaimer | Accessibility Statement