Re: [ferret_users] EOF: area weighting?

["William S. Kessler" <william.s.kessler@xxxxxxxx>]

Hi Andreas -

Good question. The area represented by each gridpoint should indeed be  
taken into account in computing an EOF. However, in the present EOF  
functions, it is not.

You should therefore do the weighting explicitly before computing EOFs.

But is your expression for the weighting (cos(y/(180/pi))^(0.5))  
correct?

The distance along latitude y1 from x1 to x2 is (x2-x1)*cos(y1).
The area of a thin strip dy along y1 is (x2-x1)*cos(y1)dy.
To find the area between y1 and y2, integrate this in y:

A = (x2-x1)INT_y1^y2(cos(y))dy = (x1-x2)(sin(y2)-sin(y1))

Regards .... Billy K



On Dec 7, 2009, at 3:45 AM, Andreas Baumgaertner wrote:

> Hi ferret users,
> I am using the functions EOF_SPACE, EOF_TFUNC and EOF_STAT to  
> calculate
> EOF of geopotential height. This calculation requires equal-area grid
> box sizes, i.e. a latitude weighting with cos(y/(180/pi))^(0.5). Is  
> this
> already accounted for in these functions, or do I need to do that  
> before
> using them?
> Thanks,
> Andreas Baumgaertner