>I was wondering if anyone knew how to perform a rotated eof for SSTs.
>The first issue is how to treat the missing values over the land, since
>the varimax function in NCL doesn't allow for missing values.
>Second, how do you obtain the % variance of each eof mode after
>rotation? Thanks.
>
The public domain code used by NCL is directly from:
c Orthogonal Rotation according to varimax criteria.
c D.J. Veldman
c Fortran Programming for the Behavioral Sciences
c Holt, Rinehart, Winston, 1967
c pp 214-215
I used this for some personal stuff. I was asked to include it
NCL. I had some reservations because it was not consistent with
other EOF routines in NCL ...ie: it did not allow missing values
and it did not recompute the %variance. The original requestor
stated that all he wanted was to see if the patterns were
altered in any significant fashion. After, including
it there have been several requests for (1) calculating
the % variance explained and (2) for it to allow missing
values. **I will send this fortran code (which has many f66-isms)
in it if anybody wants to look at it.** Maybe, an eager
user can alter the code to address both short comings. :-)
Alternatively, if someone has a well-written code
[fortran or C] that rotation, allows missing values and computes
percent variance, I'll work with you to include
it in NCL.
Dennis Shea
_______________________________________________
ncl-talk mailing list
ncl-talk@ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
This archive was generated by hypermail 2b29 : Thu Aug 14 2003 - 17:05:39 MDT