Re: Thinking like a Ferret

[Andrew Wittenberg <atw@xxxxxxxxxxxxx>]

On Fri, 24 Jan 2003, Bruce or Jennifer Ford wrote:
>          I have some simple calculations that I can't seem to get scalar
> values for.  For instance, using NetCDF global files, how can I get a scalar
> value (to use in curve-fitting coefficient calculations to place a best-fit
> line on a scatter plot) for:
> 
> sum of x*y  (x*y for each gridpoint, then sum them.  It's the summing for a
> scalar that is problematic)
> 
> n  (count the number of gridpoints in a given region return a scalar)

Hi Bruce,

I'd echo Lynn deWitt's earlier suggestion -- check out
   $FER_DIR/go/regresst.jnl
and the corresponding x,y,z versions.  The sum of products you describe is
contained in their variable PQAVE.  Although these scripts compute
regressions along only a single dimension, they can easily be generalized
to multiple dimensions -- just change the [t=@ave] stuff to
[x=@ave,y=@ave,t=@ave] etc.

The number of valid values (gridpoints) in the current SET region for a
variable V is, for example,
   v[x=@ngd,y=@ngd]
and the number of "missing" values is
   v[x=@nbd,y=@nbd]
with the total number of points the sum of these.

Andrew

+--------------------------------------------------------+
|   Dr. Andrew T. Wittenberg   |        GFDL/NOAA        |
|      atw@gfdl.noaa.gov       |      Princeton, NJ      |
+--------------------------------------------------------+