Spectrum Processing - Vega

[SteveBz]

Hi Guys,

So I took my embarrassingly out of focus spectrum of Vega from last year, made with one of those plastic gratings off ebay for 3.99 that looked like this:

I put it through GIMP producing this:

And then pushed it throught VSPEC getting this.

Have I done the processing right?  Going from left to right, I think I what I see here is: H-delta (left edge), H-gamma (434), H-beta (481).

What's that big dip just before 580? And why doesn't it look like a Planck's law black-body radiation curve?

Undeterred, I've added a prism and a Barlow collimator to the mix.  It looked great last night, and in focus, but I couldn't get focus on the sensor.  Work still in progress.

Thanks,

Steve

[vlaiv]

33 minutes ago, SteveBz said:

What's that big dip just before 580? And why doesn't it look like a Planck's law black-body radiation curve? 

You need to do instrument response calibration of your data. Guess that you are using your D5000? Take a look at common OSC QE graph:

Every camera will have slightly different shape to it, and peak will have different relative scale (some cameras are most sensitive in green, some in red part of spectrum).

Spectrum that you are recording is in fact modulated by several factors - atmospheric transparency at particular wavelength, scope transitivity / reflectivity curve, sensor QE at particular wavelength.

Each of those can be represented with a graph like above one - wavelength on X, percentage on Y - and spectrum that you record is multiplication of each of those graphs. So you can easily spot dips around 490 and 580 in your data as well - that is transition from blue to green to red filters in bayer matrix (compare to graph I posted).

In order to get "proper" spectrum you need to account for all of that. It is not as hard as it might sound at first - you need a reference star and its spectrum and your recording of reference star - you then divide your graph with reference spectrum to get calibration graph for your other spectra.

Have a look at some tutorials for spectrum processing - that step is always included and explained.

[SteveBz]

17 minutes ago, vlaiv said:

In order to get "proper" spectrum you need to account for all of that. It is not as hard as it might sound at first - you need a reference star and its spectrum and your recording of reference star - you then divide your graph with reference spectrum to get calibration graph for your other spectra. 

Nice. Excellent idea.

[Merlin66]

Steve,

Looks like you're off to a good start. Getting the spectral image into Vspec and producing a 1D profile is a challenge in itself!

What you are seeing, as Vlaiv says is the camera Bayer matrix response superimposed on your spectrum. This needs to be "corrected" by applying (dividing by) an Instrument response curve (IR) . This can be generated by comparing your spectrum with a reference spectrum - In VSpec you'll find a collection of Pickles standard spectra, by selecting an A0 spectrum and dividing you'll get an IR curve; smooth the bumps caused by the Hydrogen absorption line and then use this to correct your original spectrum. See the attached illustration.

This shows the typical Canon camera "dip" around 5500A.

(This is taken from my book "Astronomical Spectroscopy for Amateurs", Chapter 10 - Processing Spectra, p97)

Onwards and Upwards....

[AstroKane]

Guys  - I cannot profess to really understand the above but it was a fascinating read. By coincidence I spent 30 mins on Vega last night gathering some subs...thought I would drop them here to compliment your data. ? Clear Skies

[SteveBz]

1 hour ago, AstroKane said:

Guys  - I cannot profess to really understand the above but it was a fascinating read. By coincidence I spent 30 mins on Vega last night gathering some subs...thought I would drop them here to compliment your data. ? Clear Skies

Nice photos.

Mine aren't as nice as yours, but nearly all the science of astronomy is hidden in the spectra.  So it's worth checking it out.

Again it's not very easy or cheap, but it's a real niche.

Best

Steve.