I started rambling about this at the SIG mtg on Nov13.
Is anyone calibrating to absolute flux and deriving magnitudes? ; submitting mags to AID?
I was looking for a project for a SA-100. You could use a number of digital filters on a single image, transformation shouldn’t be needed and it validates the SED.
Just starting and looking at errors and such.
Sounds complicated and I am clueless.
Is there a way to calibrate an SA-100 image to absolute flux?
I don’t even know an easy way to calibrate pixels to absolute flux. Are the units Watts per pixel squared?
By digital filters, do you mean software applied to the image matrix? Are you just adding up the ADUs in pixels that correspond to spectrum pixels that occupy the nominal filter bands with band-edge attenuations typical of those filters?
If so, it seems that transformations would still be needed.
ray
Hello Ray,
This question came to me while I was taking the Best Practices course this past summer and the discussion and exercises were about transformations. I understand the need to correct between slight differences in filters, but it seemed to me that this wouldn’t be necessary using a “calibrated” spectrum and a standard filter transmission table.
There is a way to do the calibration. I included a link to a good website that discusses the methods. There are also a set of digital filter tables, e.g., Bessel V, there. There are a couple of other papers if interested. It’s like zero-order photometry in a way; you’re comparing a known star and your target.
I use the units I typically for energy as ergs/cm2/sec/A …
To illustrate what I have tried so far, using RSpec, get an SED from a source like CALSPEC which is in calibrated ergs as a profile. Also input a V filter (above) as your reference. Cut the data below 4700A and above 7300A. Multiply the two and move to profile. Enclose with meas lines and check “Area under curve”. Beware this number might need to be corrected for the wavelength delta of your curve, e.g., if 3A/pix then multiply by 3 to correct. In this case, the area is in ergs, so divide by 2600 (7300-4700) to get the average flux per A. Using the ST mag system, MagV = -2.5log10(avg flux ergs) - 21.1…This also works in reverse, in a sense. If my SED is in relative flux I can still do the filter multiply and get an area for V in total relative flux. If I know the star’s magnitude I can convert to ergs = 10^((-.4*MagV)-8.449) = 10^((MagV-21.1)/(-2.5))…
This becomes a multiplier to the relative flux curve.
All of this said, you use your measured ADU per A. Correct the ADU to ADU/sec. Also, divide ADU/sec by number of pixels in the value, i.e., the aperture (y column). I think RSpec adds them but not sure and not sure what too many pixels would cause. Divide this by the A/pixel you measured. You now have ADU/pix/sec/A. Divide this by the stars catalog SED in ergs/cm2/sec/A and you will wind up with basically an instrument response curve in ADU/ergs. Divide your target by this to get your target in absolute flux.
Note: I had some issues with RSpec; the area problem I mentioned and problems with number scales. Ergs is a very small number here; ADU/ergs is a very large number. RSpec had problems doing some of the calcs. I just rescaled the numbers, did the calcs then unscaled the results (if I don’t forget…).
That was way too much. I’m sure I’ve made it more confusing and, I’m open to being corrected or shown a better way. I just thought, why buy a bunch of filters when you only need one and it does double duty (photo and spectro)…
Good luck, Paul
http://www.astrosurf.com/buil/calibration2/absolute_calibration.htm