[ncl-talk] Problem with reconstructed plot

[Dennis Shea]

'eof2data' will construct fields based on the rotated components. The code
is just 'do loops'

BUT...

--
Classic EOFs are orthogonal (independent). The contributions of each
component can be added. The associated principle components are not
correlated.

However ..

If you rotate a subset via varimax, the subset will be orthogonal to one
another within the subset BUT **not** to the entire field.
You are 'mixing' the subset eofs. The principle components associated with
the patterns **are** correlated

---




On Fri, Feb 20, 2015 at 4:05 AM, Ipsita Putatunda <
ipsita.putatunda at gmail.com> wrote:

> Hi all,
>    Am doing varimax roatation of first three EOFs with my data using
> "eofunc_varimax_Wrap" function, and then trying to reconstruct the plot
> using "eof2data" function. Though my first 3 REOFs is contributing more
> than 98% of variance but still am not getting similarity in  my
> reconstructed plot with the original one. I doubt if am making any mistake.
> Part of my script has been provided herewith. Any help will be appreciated.
>
>
> .........[snip]...
>   xo=LHo(lev|:,time|:)
>   optEOF =True
>   eofo=eofunc_Wrap(xo,neof,optEOF)
>   eof_tso=eofunc_ts(xo,eofo,False)
>   printVarSummary(eofo)
>   eof_roto=eofunc_varimax_Wrap(eofo,1)
>   eofunc_varimax_reorder(eof_roto)
>   printVarSummary(eof_roto)
>   do n=0,neof-1
>       eof_tso(n,:)=eof_tso(n,:)+eof_tso at ts_mean(n)
>   end do
>   printVarSummary(eof_tso)
> ;=====================================================
> ;    Reconstructed Mean
> ;===================================================
>     oo=eof2data(eof_roto(0:2,:),eof_tso(0:2,:))
>  ;  oo=eof2data(eof_roto,eof_tso)
>
>     [snip]
>
>
> Thanks,
> Ipsita
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20150226/0d2da545/attachment.html