Tue, 01/06/2015 - 16:42
For those doing low-resolution spectrocopy, the Orion Project web site has some low-resolution wavelength calibrated line profiles that can be refereneced to help calibrate other profiles. These profiles have just the first order spectrum profile shown and have the hydrogen Balmer line wavelengths positions shown as blue lines. These are mainly for the bright Orion Project stars, but also includes Vega.
Since most stars exhibit at least some hydrogen Balmer Lines and if you can identiify at least two of those lines you can get a good calibration.
See: http://www.hposoft.com/Orion/OrionSpectra.html
Jeff Hopkins (187283)
Hopkins Phoenix Observatory
I see these were taken using an ALPY 600 slit spectrograph. This instrument has a significantly non linear dispersion. A wavelength calibration of this instrument using just 2 points has significant errors. If using Balmer lines to calibrate ALPY 600 spectra the recommendation of the instrument designer (Christian Buil) is to use as many Balmer lines as posssible, suplemented if possible by neon/argon lamp lines in the red region. I can confirm that this technique works well and is built into ISIS sofware
A large proportion of stars do not in fact show clear Balmer lines but this is not a problem since a calibration performed on one that does can be transfered to any object provided one has a fixed reference point. In the case of a slit spectrograph the slit position forms that reference point. In the case of a slitless spectrograph eg Star Analyser the zero order is normally used as the reference point as this is of course present for every object.
It is worth noting also that except for Vega which is a main sequence A star none of the other stars shown are a particularly good choice of reference star for the Star Analyser.
For calibration of low resolution spectra taken with the ALPY 600, a much better tried and tested set of reference stars are stars from the database of MILES standard spectra, chosing one which is close in elevation to the target star to minimise differential atmospheric extinction.
Robin
Here is Christian Buil's recomended procedure for wavelength calibrating the ALPY 600 low resolution slit spectrograph.
http://www.astrosurf.com/buil/isis/guide_alpy/resume_calibration.htm
It can also be used for other low resolution slit spectrographs.(It is not directly applicable to slitless spectrographs like the Star Analyser) The example uses ISIS software but the procedure can be used with other programs. Three different procedures are described
1. Using Balmer lines in an A or B star
2. Using a Ne-Ar calibration lamp
3. Using a combination of 1 and 2 which gives the best results, particularly at the extremes of the wavelength range.
I can confirm that these (particularly 3) give very good results with residual errors well below 1A across the calibrated range.
Robin
Part of the trick for calibrating a star spectrum’s profile requires a bit of detective work. As mentioned, most stars display fairly distinct hydrogen Balmer lines and after a while they become very easy to recognize and used as calibration points. No need to use the zero order reference. This is applicable for low-resolution work where there is no high Doppler shift. For high-Doppler shift and high-resolution work naturally you want to use a fixed Earth-based set of references to determine any Doppler shift.
Reference Wavelengths:
Here are some reference wavelengths. Note for low-resolution, just drop the digits after the decimal.
Hydrogen Bakmer Lines
alpha - actually a doublet 6562.72Å & 6562,85Å, use 6582.81Å
beta - 4861.35Å
gamma - 4340.472Å
delta - 4101.734Å
epsilon - 3970.075Å
Sodium D lines (note for low-resolution these are combined and will show a wavelength between those listed)
D1 - 5889.950Å
D2 - 5895.950Å
Helium I - 6678.15Å
All the reference profiles on the Orion Project web site Reference page have the wavelengths calibrated precisely, by using the actual star spectrum spectral lines. Again this is for reference. Naturally no wavelength shift information is available since the wavelengths were set by the calibration. So no Doppler can be determined for these reference stars, but then for both the Star Analyser work and ALPY 600 the resolution is too low for that anyways. I hope these references will be helpful to the beginner and show how easy the hydrogen Balmer lines can be identified.
While there are several free spectrum processing software programs, it is highly recommended that for beginners and those who like the ease of learning and use of RSpec, that it be invested in. Too many people have lost interest in spectroscopy due to the steep learning curves of the freeware.
Spectroscopy is actually pretty easy and can be exciting and rewarding.
Jeff Hopkins (187283)
Hopkins Phoenix Observatory
The only recommended procedure to initially calibrate low resolution spectra (regardless of what software is being used) is to use the zero order and one known point on a spectra of a star that has easily identifiable features (like Type A stars, which have very prominent Hydrogen Balmer lines.)
After this initial calibration, you can calibrate other spectra of any type of object by using the "one-point" calibration method.
The zero order must be visible for this procedure. This is why it's important to install and use the grating properly. There's a discussion and useful calculator here: http://rspec-astro.com/calculator.
Tom
Note: Videos 3 and 24 at http://www.rspec-astro.com/more-videos demonstrate the above calibration procedures.
Be aware that a diffraction grating can be considerably non-linear in it's dispersion, particularly with targets at low altitude (high airmass). So while you may measure 12 A/px between your zero order and H alpha in the red, you may get a different value between your zero order and H beta in the blue. That difference is often negligable, but can be quite noticable at low altitude and will lead to large errors if you assume 12 A/px across the whole spectrum.
So I'm not sure that I would agree that the only recommended procedure is to use the zero order and a single feature for calibration. Far better to use the zero order and multiple features to create the dispersion function with a non-linear fit. Happy to stand corrected as usual though.
Malc
Good luck to anyone who follows that advice for Star Analyser spectra. I have lost count of the number of people who have contacted me direct or via the forums asking for help calibrating their Star Analyser spectra after they missed the zero order off the image. (For example how would you calibrate as Star Analyser spectrum of Betelgeuse you show there without the zero order) If you have the zero order and the dispersion you can calibrate a spectrum of any object with certainty without knowing anything about it and you dont need to be Sherlock Holmes or a clairvoyant ;-)
Robin
We need to be careful to define what sort of spectrograph we are talking about here. Tom's comment is not correct for low resolution spectrographs in general. For example there are many low resolution spectrographs which have a slit so do not need the zero order but definitely require a non linear calibration. Spectra taken using The ALPY 600 which was referenced at the top of this thread is an example. If we confine ourselves however to the slitless Star Analyser in the converging beam arrangement, then,
The wavelength calibration of the Star Analyser used on its own in the converging beam is approximately linear (Theoretically very close to linear because of the small diffraction angle used, but in practise in certain circumstances various effects can make the non linearity noticeable). A simple linear (A/pixel) calibration will get you close and is what I would recommend for the beginner but if you are looking for better accuracy then a 2nd order calibration using 3 or more points ie the zero order and a selection of Balmer lines in the reference star (H alpha, H beta, H delta for example) will get you closer. The recommended procedure is the same though.
1. Determine your dispersion equation using the zero order and Balmer lines in a reference star
2. Apply that equation to your target spectrum using the zero order as your fixed reference to define the offset (The constant in the dispersion equation) this is all handled by the various software programs.
The key point to remember though is that to wavelength calibrate any slitless system you need a fixed reference point which is present in all spectra and that the zero order location can be used for this purpose.
Robin
(edited to correct typos)
I want to emphasize what Robin said. The wavelength solution can be non-linear when:
1) The resolution is very high or under sampled. If your resolution is several 10s of angstroms per pixel, then you lack the resolution on most systems to detect any non-linearity.
2) If there is a significant field distortion in your system. But in most cases if you have this the spectrum will still be well approximated as linear, it is just that the linear term will change depending on where your source is in the field.
3) When there is significant atmospheric refraction (at high air mass). But to be a real issue the unfiltered image of the star would have to be visibly distorted along the paralactic angle in order for this to be a factor. In fact depending on the orientation of your grating, this effect is just as likely to cause the spectrum to curve on the cross dispersion axes as it is to cause a wavelength distortion. I think it will be relatively uncommon for most of you to detect this circumstance.
Really when it comes down to it the wavelength calibration isn't something you need to spend hours worrying about. It will be difficult at this resolution to measure the wavelength of something to any accuracy. So really the calibration is a guide to help you locate and identify the features in the spectrum. You should not be putting a lot of stock in shifts you may observe in the wavelength of those features. So my advise is: don't knock yourself out worrying about all the secondary effects and errors.
Hi John,
I think you are mixing resolution and dispersion. Angstroms/pixel is dispersion. The higher the dispersion number the lower the resolution. Resolution has no units and is just a ratio. I believe what you are talking about is dispersion, not resolution. Star Analyser spectrum line profiles have high dispersion, but are low-resolution.
Also, I see the term grating spectroscopy when I think what is meant is slitless spectroscopy as with the Star Analyser. I know of noone doing prism spectroscopy.
Jeff Hopkins (187283)
Hopkins Phoenix Observatory