Transcript Presentation - Spring School of Spectroscopic Data Analyses

Spring School of Spectroscopic Data Analyses
8-12 April 2013
Astronomical Institute of the University of Wroclaw
Wroclaw, Poland
Echelle
spectra
reduction
The
art of
cooking
spectra
with IRAF*
Giovanni Catanzaro
INAF – Osservatorio Astrofisico di Catania
Warning! This is not the “theory” (if any…) of spectra
*IRAF (Image Reduction and Analysis Facility) is distributed by the National Optical
reduction.
I show you just the main steps for the
Astronomy Observatories, which is operated by the Association of the Universities for
reduction
of echelle
spectra
acquiredagreement
with with
a fiber-fed
Research in Astronomy,
inc. (AURA)
under cooperative
the National
Science
Foundation
spectrograph and I provide you with a “recipe” for
“cooking” (extracting) your spectra
1
Spectroscopic School of Data Analysis
08/04/2013
What does reduction mean?
We acquired images
like this one
2
Spectroscopic School of Data Analysis
We want to extract normalized spectra
like this one
08/04/2013
Typical images acquired
during a night
 bias
 flat field lamp
 Calibration lamp
 objects
3
Spectroscopic School of Data Analysis
08/04/2013
The reduction process
The reduction process consists of a series of operations aimed at
removing and/or taking into account the defects and the problems that
affect the star signal, before the extraction of the stellar spectrum. These
are due both to the optics and the detector.
Echelle orders
Scattered light
4
Spectroscopic School of Data Analysis
08/04/2013
OVERSCAN SUBTRACTION AND IMAGE TRIMMING
BIAS SUBTRACTION
SCATTERED LIGHT SUBTRACTION
SPECTRA EXTRACTION
Basic steps
DIVISION BY FLAT SPECTRUM
WAVELENGTH CALIBRATION
NORMALIZATION TO THE CONTINUUM
5
Spectroscopic School of Data Analysis
08/04/2013
Bias
6
Exposures with texp=0 sec and closed shutter
We produce a “master” bias by averaging the individual
bias frames in order to remove cosmic rays.
zerocombine task
Spectroscopic School of Data Analysis
08/04/2013
Overscan removal
The overscan level in ADUs is only an
“offset” related to the electronics which
reads out the CCD.
Its value could
slightly change from one line to the
other due to very small variations in the
reading conditions. We can account for
this effect even if it is normally
negligible.
During
this
operation,
performed with the task ccdproc, we
can also trim the image leaving only
the true pixels in the final image. The
r.m.s of the overscan values is a good
measure of the read-out noise in ADUs.
In this example we
are performing in the
same time the
overscan and bias
removal and the
image trimming
7
Spectroscopic School of Data Analysis
08/04/2013
Preparing master Flat
We average flat frames – which are indeed images of a
continuum, featureless spectrum (tungstene or quartz
lamp) - after the bias subtraction with the imcombine or
flatcombine task
8
Spectroscopic School of Data Analysis
08/04/2013
Scattered light subtraction
Scattered light is clearly seen between the spectral orders
where the star signal is higher and the orders are closer; it
can be due to several causes: dust grains, defects in the
optics, spurious orders (ghosts), etc. that bring light away
from its path. It can be removed to a very large extent.
scattered light
contribution to the
background
Background
after
subtraction
9
Spectroscopic School of Data Analysis
08/04/2013
Aperture finding and tracing
We use the apscatter task inside the echelle package
We must tell IRAF
where the spectral
orders are, for
evaluating the
scattered light in the
inter-order regions.
The apscatter task
allows us both to
define the apertures
(echelle orders) and to
evaluate and subtract
the scattered light
10
Spectroscopic School of Data Analysis
08/04/2013
11
Spectroscopic School of Data Analysis
08/04/2013
Each aperture is traced by fitting the traced points with a Legendre polynomial
For HERMES spectra, I used
an order n=7
This fitting has been
done for all orders
12
Spectroscopic School of Data Analysis
08/04/2013
The level of scattered light is
evaluated line by line by fitting
the x-cuts where the echelle
orders have been removed.
A new image, with the fits in
each line is temporarily
created
13
Spectroscopic School of Data Analysis
The vertical cuts of the new
image containing the x-cut fits
are taken and fitted with a spline.
Thus, a 2-dimensional fit of the
scattered light is performed and
this “smooth” scatter image is
subtracted to the original one.
08/04/2013
Since the fiber is in a fixed position, for all the other
images we can do automatically the subtraction of
scattered light taking the previous image (refstar) as a
reference for the aperture parameters
Whenever the
position of the
star along the
entrance slit
changes, one
must define the
apertures for
each individual
image
14
Spectroscopic School of Data Analysis
08/04/2013
Spectra extraction
As for apscatter, for all the other objects we use an input list
and extract automatically by choosing an image as aperture
and a “profile” reference-frame (for cosmic ray rejection)
For object frames: clean  yes (Optimal extraction)
For flat field and Th-Ar frames: clean  no
15
Spectroscopic School of Data Analysis
08/04/2013
16
Spectroscopic School of Data Analysis
The echelle blazing has been largely
removed. It is much more easy and
safe to define a continuum in this
spectrum
08/04/2013
Wavelength calibration: ThAr (Ne) lamp
17
Spectroscopic School of Data Analysis
08/04/2013
We identify for each order some
line and then we type “l” to
automatically find additional
lines in the spectrum, whose
wavelength is contained into a
file inside IRAF
linelists$thar.dat
18
Spectroscopic School of Data Analysis
We type “f” to perform a fit
of wavelength as a function
of pixel number. In the plot
the residuals (in Ǻ) are
plotted
08/04/2013
We assign reference Th-Ar spectra with refspectra
task
Dispcor is the task that corrects the dispersion and
resample the spectra with a linear dispersion
continuum task normalizes the spectra
19
Spectroscopic School of Data Analysis
08/04/2013