Tuesday, April 10, 2012


cc: Keith Briffa <k.briffaatXYZxyz.ac.uk>
date: Sat, 13 Jul 2002 14:14:55 -0400
from: Ed Cook <drdendroatXYZxyzo.columbia.edu>
subject: Evolutive spectral analysis.
to: "Ian Harris (Harry)" <I.HarrisatXYZxyz.ac.uk>

Hi Harry (and Keith),

Keith has asked me to send you some stuff on programs to do evolutive
spectral analysis. First, here is the web site for doing it the
wavelet way.


If your data are less then 2000 years long, you can paste your series
columnwise into a window and the evolutive spectral analysis is done
for you. If you have more then that number of years, you can download
Fortran, IDL, or MATLAB versions of the program. Presumably, you can
expand the dimensions to handle whatever length time series you want
to run. Using this program is probably the easiest way to go. It is
also widely accepted, although I think that there is an argument for
do it in the more classical sliding window way using MTM.

If you do want to go the MTM way, I am sending you the Fortran source
code, originally from Mike Mann and slightly modified by me, along
with four small data files that contains stats for computing
confidence levels. Those four files could easily be incorporated into
data statements in the program, but neither Mike nor I have done
that. I must say that Mike's program is not that easy to figure out
how to use. It does work okay, but the learning curve may be a bit
steep and frustrating.

I am off to China on Monday, July 15 at 9:30am and do not return
until July 31. So, if you have any problems or questions, you must
either contact me before 9:30am (possible due to the time difference)
or wait until after my return.




Attachment Converted: "c:\eudora\attach\mtmcohere1.f"

Attachment Converted: "c:\eudora\attach\fcoher.dat"

Attachment Converted: "c:\eudora\attach\fcoher-sec.dat"

Attachment Converted: "c:\eudora\attach\chisquare.dat"

Attachment Converted: "c:\eudora\attach\ftest.dat"

No comments:

Post a Comment