Re: How do I can standardize a dataset?

[Andrew Wittenberg <atw@xxxxxxxxxxxxx>]

On Mon, 28 Oct 2002, antonio rodríguez wrote:

> I have a dataset of this size:
> 
> I=1:4
> J=1:4
> L=1:428
> 
> and I want to standardize the region in order to calculate EOF's and I don't
> know exactly how to perform this within Ferret. I have performed the
> calculus of the anomalies but don't know how to divide the region by its
> standard deviation.

This will get very close:

   let data_new = data / data[i=1:4@var,j=1:4@var,l=1:428@var]^.5

although for a truly unbiased result you'd have to multiply data_new above
by an additional factor ((N-1)/N)^.5 where N=4*4*428.

If your grid has uneven spacing, you may want to use x,y,t (instead of
i,j,l) and their associated limits, so that Ferret can weight the variance
properly using the grid box sizes.  This may only be appropriate, however,
if your EOFs are themselves based on area-weighted data.

Cheers,

Andrew

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