[b:44e995dd1f]From: [/b:44e995dd1f]"Phil" <philbru@earthlink.net ([email]philbru@earthlink.net[/email])>
[b:44e995dd1f]Date: [/b:44e995dd1f]November 2, 2005 12:15:13 PM MST
[b:44e995dd1f]To: [/b:44e995dd1f]"Frank Valdes" <valdes@tucana.tuc.noao.edu ([email]valdes@tucana.tuc.noao.edu[/email])>
[b:44e995dd1f]Subject: [/b:44e995dd1f][b:44e995dd1f]Re: sine wave baseline removal[/b:44e995dd1f]
[quote:44e995dd1f]Q: I have a bunch of spectra (radio data) with baseline ripple problem.
Could you tell me which task is proper to simply take out the sine wave
from the spectra ? x-axis is velocity or freqnucy, and y-axis is
temperature. If you don't have what is your suggestion ?
[/quote:44e995dd1f]
There is a curvefitting procedure (called sinewave removal) written up
to do this at
http://digitalCalculus.com/page/677148. Take a look it
may do what you need.
Phil [quote:44e995dd1f]A: There is no task that will specifically fit a sine wave to a baseline.
The usual task for this type of problem is SFIT in the ONEDSPEC task.
The task fits functions such as chebshev, legendre, and splines, to
baseline regions. The baseline regions can be specified by a cursor or
"sample" string and/or the data regions can be rejected by interative
fitting and rejection. SFIT is both interactive and batch. I don't know
how you want to remove the baseline fit, that is subtract or divide.
In optical work it is quite common to divide by the baseline, called the
continuum in optical work, and so there is a special variant of SFIT
called CONTINUUM. However this is just the same task with the out
fixed as the ratio of the fit to the data.
The axes do not really matter though it is possible the task will not
display the units properly. This should not matter for the purpose of
running SFIT. There could also be some confusion with the dimensionality
of the data, often radio spectral data comes in more than 1D or 2D.
[/quote:44e995dd1f]