Re: request

>To: address@hidden
>From: "M.R.SRIDHAR" <address@hidden>
>Subject: Re: 20011108: Re: request
>Keywords: missing values

Hi,

> i am sridhar working on oceandata modelling..i have one ascii file
> which consists of latitude,longitude,sealevel,julianday,date,month
> ,year. The latitude and longitude is of .25 gridded...in sealevel
> there are few missing data (9999) available in the ascii file...when
> i am putting the values into netcdf format for the sealevel which is
> the function of lat,long and time. I want to know whether in the
> place of missing value whether i should put 9999 or blank...i could
> able to put 9999 but when the .nc file visualised through ferret its
> not reading properly and giving junk plots.So i am requesting you to
> give solution for this.

If you create an attribute for the sealevel variable named
"_FillValue" of the same type as its variable, then you can set that
to any value you want, including 9999.  However, it is best if the
value is outside the valid range for values of sealevel (I expect 9999
is out of the valid range so is OK).

In terms of the CDL (where I'm just guessing at variable types and
dimensions), you should see something like this from ncdump of your
file:

netcdf ... {

 ...
variables:
 ...
 float sealevel(time, lat, lon);
    sealevel:_FillValue = 9999;
    sealevel:units = "m"; // or whatever
 ...

data:
 ...
sealevel = ..., 9999, ... ;

}

For more information on conventions for missing values, see

 http://www.cgd.ucar.edu/cms/eaton/netcdf/CF-current.htm

Also, it's better if you send netcdf support questions to
"address@hidden", since then they will be handled by someone
who is here, in case I am away from my email ...

--Russ

_____________________________________________________________________

Russ Rew                                         UCAR Unidata Program
address@hidden                     http://www.unidata.ucar.edu