Re: EOF and correlation

Partha

I am not entirely clear from your post what it is that you're trying to
do, but I *think* you want to correlate the time-varying amplitude of each
EOF with SST in time and space.

If that is the case, then the EOF amplitude (more correctly the time
varying eigenvalue of each EOF) can be calculated using the eofunc_ts()
function
(http://www.ncl.ucar.edu/Document/Functions/Built-in/eofunc_ts.shtml),
which is then easy to correlate with SST using 'escorc()'

Will

On 21/03/13 1:09 PM, "Partha S Bhattacharjee" <pbhatta2@jhu.edu> wrote:

>Hi,
>
>I am a new NCL user and have some problem using the EOF function.
>I have annual mean of precipitation (lat X lon X years) over a region and
>took first 3 eofs (using eofunc_Wrap) to plot the spatial pattern map and
>time series. However, I want to find the linear cross-correlation using
>escorc of those eof's with global annual sst (lat X lon X year) for the
>same time duration. But escorc needs time in the rightmost axis for both
>x and y. Time is absent in the eof components (they have eof X lat X
>lon).
>
> I believe, the other way round is to reshape the gridded files in to
>column (using ndtooned to have lat X lon in one dimension) and do the
>eof (years X latlon) which will return eof components X year and can be
>correlated with SST using escorc. Is this a correct way of doing this?
>
>Please suggest if any other way of doing the correlation.
>
>
>
>
>
>
>
>
>_______________________________________________
>ncl-talk mailing list
>List instructions, subscriber options, unsubscribe:
>http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Mar 20 21:10:12 2013