I am very new to spectroscopy, but thanks to Robin's informative posts have been learning alot. I am particulary interested in using the zero order star image of the Star Analyzer for differential photometry of spectroscopy targets. This allows an efficient use of time and equipment (both of which are always limited!). With SS Cyg well placed for observation, I have been observing it with a 12-inch LX-200 nightly since September 24, 2019 using a SA200 for spectroscopy and creating light curves using the zero order image. I have over 5000 spectra so far! Its fascinating seeing the changes in the spectrum and being able to correlate them with exact places on the light curve! One concern I had though was how precise was the curve obtained using the SA200. So on the night of October 18- 19, 2019 I observed SS Cyg with two telescopes 100 feet apart, using the same type of CMOS camera (QHY 174) for 5 hours.One telescope was the 12-inch LX-200 using a SA200. The other was an Explore Scientific 102mm FCD100 APO refractor. I focussed on the spectra in the 12-inch, and used the defocus method (7.6pix FWHM) with the 102mm APO. SS Cyg was near minimum and flickering like crazy. I am happy to report both light curves are nearly identical. Attached to this message is an image of the results and a sample of the spectra resolution I am getting.
The zeroth order image through a star analyzer may be good for relative magnitude flux or magnitude measurements. However, I would think it suffers at least the same limitations as imaging through a clear filter (or UV-IR blocking filter) for transformation onto a standard system for comparison with observations of other observers and perhaps moreso since it may not even be comparable with observations through a clear filter or UV-IR blocking filter such as a a BG40. Therefore while the shapes of the light curves may be very similar, which is useful for many applications, I expect that there would be a significant offset between the zeroth order light curve through the SA vs. a standard filter that may not be possible to reliably eliminate by transformation. Have you looked at that?
Brad Walter
Aha, I see Robin addressed my concerns about using the Sar Analyzer zeroth order for photometry on the
"Begginer Object?" thread. His comment is copied below to include it in this thread:
"Using the zero order for differential photometry is a cool idea. Be aware though that the spectral content is rather odd (the inverse of the grating spectral response so predominantly light from the blue and red ends with the middle wavelengths missing) so it would be very difficult to transform the results to a standard photometric system (best to colour match your comparison stars as close as possible)"
That doesn't mean that it can't be useful particularly if the compstar has a very similar spectrum or at least very similar color index.
Brad Walter, WBY
Using a low-resolution grating spectrum for photometry is something I'm going to try out by myself in coming months. But my plan is slightly more complex - namely synthetic photometry on low resolution spectra. Using (pseudo)flux-calibrated spectra and convolving them with e.g. Bessell filter curves would give you flux in that particular filter with constant accuracy. For that synphot or even better, pysynphot could be used. When measuring some constant stars too, the result would be effectively differential multi-colour photometry. If those compstars happen to have standard magnitudes, a reasonably good and simultaneous extinction estimate could be found as well.
When instrumental profile (taking into account atmospheric extinction!) is determined with high accuracy (at the moment I consider that as one of the most demanding tasks here) and properly applied to target(s), probably even photometric system transformation is not needed anymore.
Most probably, life is not exactly THAT bright and cloudless, but I'm going to give a try!
Why bother? E.g. when one wants to measure really bright stars.
I wish you good luck!
Best wishes,
Tõnis
Have you looked at this technique?
http://www.astrosurf.com/buil/calibration2/absolute_calibration_en.htm
This is for a slit spectrograph but might give some ideas.
Terry
Terry, thank you for refreshing my memory! I have read that long time ago. Seems that I can use that for starter. I plan to build final reduction software in python as a pipeline.
Tõnis
Absolute flux calibration as described there using a slit spectrograph is tough as you need to use both a wide slit to collect all the flux and a narrow slit to get the resolution. You also need stable photometric skies as you need to take spectra for both target and reference. In practise it is generally easier to take coincident V mag brightness measurements and use these to convert the spectrum calibrated in relative flux to absolute flux.
Returning to slitless systems, it is indeed possible in principle to do "differential spectrophotometry" using a slitless spectrograph and then calculate brightness values from the flux calibrated spectrum. I outlined the technique at the joint BAA/AAVSO meeting last year. It was developed from the technique I used here to study fast transients in T Tauri stars.
http://www.threehillsobservatory.co.uk/astro/spectra_42a.htm
Slitless systems have the advantage over slit systems for this work as unlike a narrow slit, no flux is lost from the spectrum and provided a reference star is included in the same frame, both can be measured simultaneously giving the same advantages that differential photometry gives. The elephant in the room though is flat field correction which is impractical for wide field slitless systems due to the interraction between the spacial and spectral components in the flat. In practise. The applications are therefore somewhat limited, useful for studing fast changes in spectra for example. Andrew Smith for example is currently developing such a system to detect and study flares. He is using the zero order to detect the flare and the spectrum to investigate the evolution.
https://stargazerslounge.com/topic/345624-first-flare/
Cheers
Robin
Dear Robin,
My apologies for being away from this thread for awhile to all. I reviewed your study on the fast T Tauri transient as well as Andrew Smith's work on capturing the flare! Both are so interesting to me and congrats on great work!. I too monitor flare stars in hopes of catching a flare using my SA200 and so far no definate success. I think I might have caught a faint 0.1 mag flare on EV Lacerta which seemed to show a faint enhancement of the H beta and gamma emmision lines, but with all the complexities of flat fielding I am reading about I am not sure. I did catch a dramatic brightening of the continuum of SS Cygni over a period of about 90 minutes which looks similiar to your animation of the T Tauri star in your study. I can post these files if desired, but definately have alot more learning to do and look forward to further discussions on this thread!
Here is another interesting example of extracting photometric brightnesses from Star Analyser spectra (By Christophe Pellier on Neptune)
https://www.cloudynights.com/topic/679171-spectro-photometry-of-neptune-201490822/
Cheers
Robin
Absolute flux calibrating spectra using simultateous photometry (ie the converse of deriving photometric brightnesses from spectra) is described here in David Boyd's SAS workshop presentation "scientific analysis of amateur spectra" top of the page.
https://britastro.org/node/19378
The technique described there arose from discussions in the ARAS forum and was formalised by David.
Cheers
Robin
Robin, thank you for pointing to that link. David's presentation is indeed an extremely good overview of how spectroscopy can be done, with good details etc. I will point my students to that ;-)
Best wishes,
Tõnis
Hi all,
This is a very interesting discussion to me because I'm in this work (photometry with a Star Analyzer) since a few months. Although my main targets are planets, I have been making experiences on variable stars just as a training (and of course it became a point of interest as itself!)
I have a question on the method. Making photometry from spectroscopy asks for a preliminar step that is processing the spectra in absolute flux. Ok on this. But this supposes to find a true standard when you observe, and it seems that there are very few.
So far I have been using a more simple method which is a mere adaptation of the method used with glass filters and digital imaging, where we count ADU data and then calculate color coefficients, etc., such as described in the 6th chapter of the AAVSO CCD guide (on this website).
I am assuming that a spectrum processed in ADU as well can be used for photometry without the need to pass by the absolute flux step: I calculate the integrated uncorrected flux in each one of the color band and calculate coefficients just like if I was using glass filters. Is this method flawed at some point or is it just too unprecise ? Or complicated ?
Here are a few experiences on variable stars. Although my method is still experimental, I do find coherent or even correct values.
Betelgeuse
R Leonis
P Cygni
Regards,
Christophe
Hi Christophe,
Yes your method is essentially the same as that I described at the BAA/AAVSO meeting. The problem is you either need a comparison star in the same field (like differential photometry) in which case you potentially run into the flat field problem or you record the comparison star in a separate observation which, provided you place the target and comparison at the same position, solves the flat field problem but means you need stable sky conditions (equivalent to all sky photometry) This is not nomally a big problem with spectra calibrated in arbitrary flux units as any atmospheric effects are only second order. (eg clouds are ~grey so have no effect on the spectrum) but when comparing absolute flux measurements as here, any change in atmospheric extinction will be a first order effect, eg if cloud cover reduces the flux by say 10% this will produce a 10% error in the measured brightness
Cheers
Robin
Hi Robin, thanks for your answer. Differential photometry is indeed the way I'm doing it, and take much care of choosing comparison stars located at the same elevation, and of course only during nights that I suppose to be photometric (it can be hard to be sure, sometimes, in western France winters...)
The color coefficients are hard to stabilize though, so I surely want to try improvements in some ways.
Robin, have you estimated which is more problematic - slit losses or transparency changes? When one is doing standard spectrophotometry, transparency changes affect always the result as well. It seems to me, that the issue of slit losses (that can be very significant in blue, specially when slit is not perpendicular to the horizon) is much more unpredictable and difficult (if not impossible) to correct.
About flats - please correct me if I'm doing a mistake. If one is using twilight flats, every point in the sky is producing spectrum. The result is a convolution of all of those spectra, and that should be a flat surface. So at least it could be used to correct for PRNU, hopefully also for vignetting - providing that filter wheel is positioning very well.
Of course, flat in blue can be rather different of that in red (at least EEV CCD-s show dramatically different patterns) and that probably can't be corrected (unless positioning target and compstar(s) exactly in the same way). I'm not sure how problematic it can be with typical front-illuminated CCD or CMOS sensor.
With best wishes,
Tõnis
I find that except perhaps for the uv end of the range, a reasonable continuum calibration in relative flux can be produced ie the wavelength dependent slit effects (eg due to chromatism in the optics or atmospheric dispersion) are generally manageable provided conditions are kept the same for both target and reference. (Ideally telescope optics with low chromatism should be used and the slit should be aligned with the paralactic angle but in practise my setup fulfills neither of these criteria) For example here are some tests I ran for both slit and slitless spectra using standard stars.
http://www.threehillsobservatory.co.uk/astro/spectroscopy_21.htm
I dont believe absolute flux calibration can be done using a slit spectrograph with any degree of accuracy however unless the slit is wide compared with the star psf (Shelyak for example supply spectrophotometric slits which have both wide and narrow regions.) and sky conditions are photometric. If absolutef flux is required I think it is easier to measure photometric brightness conventionally eg using a V filter and use that to absolute flux calibrate the spectrum.
The problem with flat correcting sltless spectra is that because the illumination at a particular location in the field depends on both the position of the source and the dispersed spectrum of that source, a different flat is needed for each location in the field. It might be possible to produce a sensor PRNU map for a slitless spectrograph but defects such as vignetting and dust are difficult to correct for because they depend on the geometry and where the vignetting occurs. For example vignetting before or after the dispersion element produces a different effect (Before dims the complete spectrum depending on the location of the zero order, after dims by different amounts along the spectrum depending on its location in the field. Also the position of the shadow from dust donuts on the sensor cover glass for example depends on the angle of illumination which depends on both the location of the spectrum and the distance along it. The solution for professionals appears to be to produce a "flat cube" ie effectively a flat for each location of the spectrum in the field. The potential magnitude of the problem for a given setup can be checked by taking spectra of the same star in different locations in the field and comparing them.
Cheers
Robin