NaNs and iso

From: Derrick Snowden (Derrick.Snowden AT noaa.gov)
Date: Wed Mar 30 2005 - 12:50:54 MST


Hi,

I've been using Silvia's iso.scp and iso.ncl to calculate isotherm depths from
a netcdf file containing temperature. This program seems to work fine but
there is a small quirk I can't figure out. If my input file is called
temperature.atl.nc then the output from iso.scp will be
iso_temperature.atl.nc. Inside iso_temperature.atl.nc is a variable called
ISO with several attributes added. _FillValue among them. The _FillValue
attribute is equal to the _FillValue from my original file. So far so good.

When I use ncdump to view the contents of the new file, I see several fill
values that are indicated by missing data in the ascii listing (e.g. ,_,_,_,
etc) Additionally, I see "not a number" values denoted by "nan,nan,nan" etc.
I'm not sure how to deal with the nans and they've choked up some subsequent
analysis I was doing using NCO utilities. e.g.

ncwa -alatitude,longitude iso_temperature.equ.pac.nc
area_average_iso_z20_ecco.nc

ncwa seems to understand the _FillValue but not the nans. The resulting file
is entirely nans.

Where did the nans come from? Is this a default fill value inside the fortran
program that iso.ncl calls?

Where should I try to fix it? I'd like to mask the nans so that in the file
they are numerically equal to the _FillValue, and I guess iso.ncl is the
place to make this modification.

Ideas?

Thanks much,
Derrick

-- 
Derrick Snowden
PhOD/AOML/NOAA
4301 Rickenbacker Causeway
Miami, FL 33149
Office: 305.361.4322  Fax: 305.361.4392
_______________________________________________
ncl-talk mailing list
ncl-talk@ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk



This archive was generated by hypermail 2b29 : Thu Mar 31 2005 - 09:00:54 MST