Thanks a lot for the response. I had the problem in presenting data for my LAS server. I am think to write an external ferret function to deal with the problem. It is very helpful to know that the FFT doesn't like the missing/fill values.
Bob
"E.D. (Ned) Cokelet" wrote:
I've used this FFT before. As a matter of fact I wrote documentation notes for
it because I did not understand what it did when I first tried it. It cannot
take missing values; you have to give it a complete set of valid data points.
If your data has missing values, then you must fill those with interpolates. To
do that is an art. If you've got a very few adjacent missing values, you could
try linear interpolation, but be advised that the "ramp" this makes in your data
has spectral consequences. You might try to randomize around the "ramp" which
puts in high frequency noise that also has a spectral signal. Better look up a
reference text on FFTs and spectra.--
Edward D. (Ned) Cokelet, Ph.D. Edward.D.Cokelet@noaa.gov
NOAA/PMEL ph: (206) 526-6820
7600 Sand Point Way NE fax: (206) 526-6485
Seattle, WA 98115-6439
-- (Bob) Yunyue YU Code 971, Goddard Space Flight Center Greenbelt, MD 20771 (301)614-6850 yuy@nemo.gsfc.nasa.gov