iraf.net

From jacoby Thu Feb 4 14:55:14 1988
Received: by lyra.noao (5.51/SAG.7)
id AA09389; Thu, 4 Feb 88 14:55:10 MST
Date: Thu, 4 Feb 88 14:55:10 MST
From: jacoby (George Jacoby X292)
Message-Id: <8802042155.AA09389@lyra.noao>
To: valdes
Subject: Deblending in SPLOT
Status: RO


Frank,

Tony Keyes at UCLA asked me for some ideas when having
trouble with deblending in SPLOT (for Calvin Cliff who
sent the IRAF mail message). Here is my response.

They have already tried your suggestion to multiply
by a large number, and while that helped for some
lines, close blends end up as a single fitted line.

George


The only suggestion I can think of for deblending closely blended
lines is to use the fixed separation option. This will force
a second line at the appropriate wavelength; however, the program
is pretty stubborn when it thinks it knows better, so it may
set the flux of the second line to zero. What this means is
that statistically it can't tell the difference between the blend
and a single line.

The only suggestion I can think of (if that fails) is to:

(1) fit only the unblended side of the strong line
by marking with the cursor up to the peak or so

(2) subtract the model profile with the subtract
command (a "-" I think???) over the entire
blended region.

(3) now fit the residual which will contain the
weak line plus noise.

This method allows the two fits to have different sigmas,
and so there is some extra freedom in the fits that may or
may not help.