[ncl-talk] Question about function - eofunc_varimax

Chi-Hung Chang jack7989 at gmail.com
Thu May 30 12:25:03 MDT 2019


 Dear all,

I have used function *eofunc_varimax *to obtain rotated EOFs of my data,
but there are some questions as follow that I would like to clarify :

(a) There are three options to choose by setting up *optEVX as*
     0 => do nothing; use the normalized eigenvectors directly.
     1 => scale the normalized eigenvectors by the square root of the
associated eigenvalue and return normalized eigenvectors.

    -1 => same as *optEVX*=1 but returns the scaled rotated eigenvectors.

My comprehension is that, by setting up *optEVX *as 1 or -1, it first
*denormalizes* the input EOF calculated by function *eofunc* (since by
default, the output of *eofunc *is normalized EOF). Is my understanding
correct?

(b) By setting *optEVX = 1*, the output rotated EOF is normalized by
dividing with square root of corresponding rotated eigenvalues, while it is
not normalized if setting  *optEVX = -1*

(c) How does the *varimax_variance *and corresponding *percent variance
explained* are calculated? I have read the reference textbook Statistical
Method in Atmospheric Sciences but still can't find the answer.

It would be a great help if anyone who has experience in performing REOF
and this function could guide me :)

Really appreciate your help.

Best,


Jack Chi-Hung Chang, Ph.D. Student

Research Assistant
National Center for Airborne Laser Mapping
University of Houston

cchang21 at uh.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20190530/f08780b5/attachment.html>


More information about the ncl-talk mailing list