Hi folks,
Here is the problem.
As I understand matters, I should set the inner annulus (circle) to capture the light from the target star (Iris provides a "growth curve" tool to assist this), have a buffer to avoid close stars and set an outer annulus far enough away that it is not affected by light from the target or nearby stars. Then (ideally) I should use the same aperture settings for ALL measurements.
BD Pavonis presents me with a bit of dilemma. Referring to chart 12409GS, the problem is achieving the ideal aperture settings for the stars in the chart given as 12.6, 13.8 and 14.3 in particular.
The issue is finding a suitable common outer annulus.
Do you simply fit the outer annulus as best as the star field allows?
Or do you keep the inner measuring annulus the same and fit the outer annulus for each star idividually?
Thanks in advance.
- Carl.
Hi Carl,
The usual guideline is to keep the measurement aperture fixed within one image - use the same one for the target and the comparison stars. Stellar profiles do not have a specific outer edge, so any measurement aperture is sampling only part of the profile. By using the same aperture on all stars, you are measuring the same fraction of the profile, and so the relative brightnesses of the objects are correct.
The sky annulus is considerably more flexible, as its role is to find the sky brightness per pixel, and that value is then subtracted from each of the measurement aperture pixels. As long as the sky brightness is properly measured, it does not matter whether the sky annulus contains 100 or 1000 pixels.
There is a caveat, however. The typical sky annulus is centered on the star being measured. Since the star profile has no edge, there is actually star contribution within the sky annulus. If the inner radius is pretty far out from the measurement aperture, that contribution is small. On the other hand, if the outer radius is large, then your sky annulus might also contain nearby stars, which, while mostly removed during the sky calculation, will leave some residual flux. You usually set the sky annulus so that it is reasonably distant from the star in question, has many more pixels than the measurement aperture, but also avoids bright stars that might fall within the annulus. If you can do that for the stars you want to measure by changing the width and location of the sky annulus, that is perfectly ok.
Arne
Hi,
I find setting aperture annulus especially difficult for super novae, like the one that can currently be observed well in the northern hemisphere, SN 2013dy.
Bright SN that can be observed by amateurs are usually located in a host galaxy that should be clearly visible in the light images. To do accurate photometry, you need to subtract the background from the SN's pixels ADUs, but in this case the host galaxy of the SN is contributing to the background. If the host galaxy is seen more or less edge on and/or is not covering a big number of pixels, the usual circular sky annulus will always capture some pixels from the host galaxy at some distance form the SN, and some pixels from the sky without host galaxy content, which must hurt accuracy. I wonder if there are software solutions where you can restrict the annulus to a sector perhaps?? Or should one measure the background at the location of the SN more accurately after the SN has faded away and correct the initial photometry??
Cheers
HBE
You raise good questions about the photometry of objects near galaxies or other sources of diffuse background.
Yes, the best solution is to subtract an image of the galaxy without the supernova, if one happens to have it; that requires either some luck in acquiring images before the supernova occurred, or waiting until the supernova fades completely.
As a second-best solution, one might do an approximate subtraction in the following simple way: take an image of the galaxy which includes the supernova. Rotate this image by 180 degrees around the center of the galaxy. Subtract this rotated version from the original image. Since some galaxies are at least approximately axisymmetric, this might do a decent job of removing the diffuse background.
Hi Heinz-Bernd, Carl,
One solution is the software rejects the pixels of faint stars or other flare from a large background area. This is what makes my own software. It's also available in IRIS, just check the "median" button of the photometry process.
I use a significantly more sophiticated technique, a first median (like IRIS) determines a first background level, the faint stars being mostly eliminated. Then the pixels above that level plus 2 or 3 noise sigma are indexed and rejected from the final background calculation. A margin of one pixel can be set. But ok, not a simple subject ! There are complex cases that are difficult to process.
A lot can be done for the background but if a faint star or a flare fell into the foreground area the only solution is the subtraction process like described by HB. I did test something like that for our common paper last year, where we had number of blending cases. It used the catalog mag of the faint star and the PSF, the results were well improved but not perfect.
Cheers,
Roger (PROC)
I'd like to also point out that thanks to the way magnitudes work that unless the brightness of the background galaxy is comparable to the brightness of the object you are trying to measure, its contribution is bassically equal to the normal background.
To put that in terms of arithmatic the linear fluxes add so if you have two sources convoluted together the total flux will be: Ft = F1 + F2
But remember to compute the magnitude you take the log of the flux so that the combined magnitude (mt) ignoring the constants is roughly:
mt = Log ( F1 + F2)
So if F2 << F1 then mt ~ Log(F1) = m1
In the case of a bright supernova embedded in a galactic disk you don't need to worry about the disk contribution until you can't really pick out the supernova from the disk visuallyl anyway.
So yes, the absolute correct thing to do is subrtract out the flux from the disk and perform the aperture photometry, but under most circumstances the errors you will introduce by that operation exceed the increase in precision of your flux measurement converted into magnitudes. (Not to mention your the cost to your own patience and sanity in trying to get the subtraction just right.)
Thanks for your replies!
Roger, I'll have a look at IRIS again.
uis01, I don't think I agree with "you don't need to worry about the disk contribution until you can't really pick out the supernova from the disk visually".
It all depends on the level of accuracy you are aiming at. If you are doing CCD/DSLR photometry of an object of around 13 mag, I think it';s not overly ambitious that you would like to keep the error below (say) +/- 0.05mag (if you browse recent observations of SN2013dy in the AAVSO database, you can find observations with this accuracy). An error of 0.05 mag equals 1- 2.51^0.05 ~ 5 % in terms of flux.
Let's say the flux (summed over the pixels of the SN) from the SN itself is 5 times higher than the (unknown) flux from the galaxy background. If my math is correct, then to get an overall error within 5% for the flux, the background flux estimate must be correct within a factor of 5 - (5-1)/1.05 ~ 1.19, so within 19% .
So in this example, if the flux from the galaxy is just ~ 20% higher than the background outside the galaxy, and the SN flux is at most 5 times higher than the flux from the disk, the overall error will exceed 0.05 mag by neglecting the disk contribution.
CS
HB
EDIT: upps, my initial math was wrong, I corrected it now.
Right. But remember that you are talking about total integrated surface brightness in your apperture. M74 has an average surface brightness of about 15.5/arcsec^2. The nucleus is much brighter than the disk. Let's be generous though and say that the disk has a surface brightness of 17/arcsec^2 (probably an over-estimate). Assuming your apperture is 5 arcsec^2 (typical seeing) in area the supernova will have to drop in brightness below 15.5 before the size of the disk contribution is enough to cause a systematic error of 0.05 mag.
Generally lacking a pre-supernova disk image, I'll measure the disk contribution with the same apperture and annulus for all my images. Then once the supernova has faded to nothing, I'll subtract the result for "no supernova" from my earlier measures. Again though, this has little impact on the measurements around the peak and only matters well into the decline for all supernova by virtue of their prodigious relative luminosity at peak.
Hi,
I just rechecked the risk of galaxy contribution into the foreground pixels. I used a Seligman site image (cseligman.com) to estimate its extent into my foreground pixels (I do not use circles but various shapes). At present (12.8 V) it's no issue against a 0.05 accuracy, the contribution shoud be below 0.01. There are two or three faint stars that could also contribute at a very low level. As John said it would be more a problem when the SN is back to 15.
The background calculation process is not affected by the surrounding stars and the galaxy, that well below 0.001 . I also checked a simple median like used in IRIS, the difference against my own process was just zero ADU.
Clear Skies !
Roger (PROC)
Supernova photometry is complex for many reasons. When they are bright and isolated, any method works. However, this is often not the case.
When the SN is close to the nucleus (which happens far too often!), then the nucleus does impact the photometry, even when the SN is relatively bright. As it fades, that nuclear contribution only gets worse.
Often the SN occurs in a star forming region of the galaxy, such as a spiral arm. There are usually HII regions nearby, or even luminous blue stars, that get in your aperture and again contribute at relatively bright SN magnitudes.
As the SN fades, any nearby contribution becomes important, and determining proper sky when the sky annulus does not contain sky, just a lumpy galaxy, gets tough.
The four phases of SN photometry are: early rise, near peak, down 3 mags, late time. There is very little data for the early rise and late time segments of the light curve; the more SN we detect early, the better. Even at the 3-mag point, which is critical for estimating the absolute brightness of the SN, the time it reached this point can be systematically wrong if your photometric measurements include 0.1-0.2mag contribution from effects of the galaxy.
These are some of the reasons why we don't "alert" on every SN. We only pick the bright ones (so you can monitor them with less galactic influence), and usually try to avoid the ones close to the nucleus or other contaminating background. You CAN go after other SN, and get reasonable results, especially if you only monitor around peak brightness, but you always have to watch. As others have said, it depends on how accurate your results need to be. Michael Richmond has a couple of papers on how to perform SN photometry even in poor locations by, for example, using a template prior image for subtraction that I'll look up.
Arne
Look at Richmond, et al.
1995AJ....109.2121R
for some ideas. I've followed several SNe a year or more after outburst, so it can be done. For near-peak photometry, as John mentions, the galactic contribution is usually small, so your normal aperture photometry will work.
Arne
Thanks for the reference, very interesting read!
In general I might experience greater problems with background contributions than others see simply because my backyard equipment isn't the best for these types of targets, e.g. I get pixels that are itself ca 2.5 arc sec wide (!) so for a photometry aperture that is (say) a few pixels in diameter, even for targets with low surface brightness I do get quite some flux per pixel. I think I will execute the "template" method from the paper that Arne cited in a few weeks and then see how that compares to simple aperture photometry.
CS
HB