WAVELENGTH CALIBRATION
QUALITY STUDY

Page updated the 2012-02-27


-------

Context:

One of the main difficulty of amateur astronomers in spectroscopy is wavelength calibration. However, wavelength calibration depends on geometric corrections' quality that is done before on long slit spectrograph.

We will see if large CCD can be managed for high resolution spectroscopy (R=14000 to 17000) and how wavelength calibration can be improved even without any expensive Th-Ar lamp.


-------

Standard wavelength calibration on wide CCD:

Here are screenshots of a del Sco spectrum taken with a LHIRES3 spectrograph+KAF1600 CCD. It has been calibrated with 3 neon lines (6532, 6598 and 6678 A) and then shifted using telluric lines with SpcAudace 1.6.2 software. I choose this spectrum because it's not one of the best obtained, but not one of the worst too.

del Sco 1 del Sco 2
del Sco spectrum global view Zoom on telluric lines near Ha's blue side
del Sco 3 del Sco 4
Zoom on telluric lines just on red Ha's side Zoom on telluric lines on further red Ha's side
del Sco 5 del Sco 6
Telluric lines near 6642 A Zoom on telluric lines near 6642 A


During the pipeline, the following information about wavelength calibration were printed:

  1. Calibration with Neon lines:

    ...
    Loi de calibration : 6513.51899242+0.119943882391*(x-1)+-3.33855160728e-06*(x-1)^2+0.0*(x-1)^3
    avec RMS=1.55885024527e-07
    Loi linéarisée : 6513.51899242+0.114839236985*(x-1)
    ...

  2. Shifting with telluric lines:

    ...
    # Les limites non nulles sont 45 ; 1527
    # Spectre nettoyé des bords (45;1527) sauvé sous
    ...
    # spc_calibredecal Spectre décalé 0.230879432 CRVAL1=6518.91763751
    ...
    Spectre de calibration avec (2) de meilleure qualité (dec de Meanshift).
    Loi de calibration finale linéarisée : 6518.917637511+0.114839236985*(x-1)
    Qualité de la calibration :
    RMS=0.0347094997329 A
    Ecart moyen=0.00181969599989 A

As telluric lines' position in spectrum (blue plot) coincide with theoretical position (green plot) from left side to right side of the CCD sensor, wavelength calibration is then welldone all over the whole sensor. 3 Neon lines aren't sufficient, in general, for a usuable wavelength calibration on large CCD as shifting in this example is around 2 pixels and other times it can be smaller.

1Shifting found with telluric lines isn't 5.4 A. Pixels at spectrum's edges that had intensity equal to 0 are removed.


-------

How to improve wavelength calibration:

1. Tips and trics:

2. Using more Neon lines ?:

Actually, LHIRES3's users work around Halpha line almost because they study Be stars. With KAF400 CCD sensors, only two strong Neon lines are usuable and three with KAF1600 CCD sensors.

Is there any other lines usuable in Neon spectrum around Ha? Let's see on longer exposure lamp spectra.

4 Neon 4 Neon

Between 6598 and 6678 Neon lines, there are two lines. Great!

Wavelength calibration with the 4 strongest Neon lines (6532, 6598, 6552 and 6678) and without any use of telluric lines gives the following result:

4 Neon 4 Neon
Zoom on Ha's line blue side. Zoom on Ha's line red side.

Wavelength calibration with 4 Neon lines on KAF1600 CCD sensor is so accurate that shifting with telluric lines isn't necessary anymore.


-------

The most powerfull way:

If you have enough money to by a Thorium-Argon lamp, and feed Lhires3 with this light, you will then have the benefit to have a lot of calibration lines even in high resolution spectra!

Some peaple have allready done such setup on Lhires3:

-------

Conclusions:

Large CCD (KAF3200 sensors and so) has enough Neon lines to do an accurate. But KAF1600 CCD sensors has only 3 strong Neon lines. We can improve wavelength calibration on such sensors using telluric lines but a better way is to have initial wavelength calibration allready accurate: this can be done using more Neon lines. However, on lower resolution spectra (id. with 150 g/mm grating), there is a lack of Neon lines in spectrum's blue side.

An other way would be using an other calibration lamp with more lines and with better distribution over the whole visible spectrum.