I have been working on this project for several weeks, carefully optimizing the photometric aperture and sky subtraction method for each T dwarf, using the RMS scatter of photometry of field stars of similar brightness. Optimal apertures tend to be smaller for fainter objects and for nights of better seeing, while brighter objects or nights of bad seeing require larger photometric apertures. The optimal type of sky subtraction for the images as a whole can vary from no sky subtraction, to sky subtraction based on a single median-stack of normalized images from the entire night, to sky-subtraction based on multiple sky frames such as I described in my previous post. It has been a surprise to me that the multi-frame sky subtraction did not always produce the best photometry. There may be a number of reasons for this. The shot noise is higher in sky frames created from an average of fewer images, and this puts the multi-frame approach at a disadvantage. The masking of stars that I employ in this approach can (if the data are sparse or affected by clouds) introduce image artifacts that affect the photometry. However I must say that although I think I understand the big picture of why the optimal aperture changes from night to night and from object to object, I do not feel that I have a similar understanding of why the optimal type of image-based sky subtraction varies.
Whatever type of image-based sky subtraction I have used, I always additionally subtract the mean in an annulus around each object for which I obtain photometry (as is, of course, standard practice for aperture photometry in astronomy).
One of the last problems I had to deal with was the satellite streak in this image:
it wiped out two of the 54 stars I was using as photometric references on this frame. My relative photometry algorithm is designed to deal with outliers, but this streak is so intense that it not only made the two stars that are behind it positive outliers -- it made all the other stars in the image negative outliers, since the outlier identification uses the ratio of each star to the summed fluxes of all the others. So the satellite really wreaked havoc on this frame even though only two stars were directly affected. I could have thrown out the frame, but that was a more serious loss of data than rejecting the two stars. However, since I use a consistent set of reference stars across all nights, the rejection of two stars meant the re-doing of photometry on other nights where the same object was observed. This is all done and finalized now, however.
What I am planning to do next is obtain absolute photometry for each object, just to have a reference-point for comparing fluxes. I am also planning to stack photometric monitoring data from different T dwarfs on different nights as a function of airmass, to look for a systematic dependence of the relative brightness on airmass. There should be a weak dependence due to the increased effect of telluric water vapor absorption on T dwarf photometry relative to stars (because the T dwarf flux is heavily weighted toward the red edge of the f814w filter I used for the observations, and that includes the telluric water feature near 950 nm). The airmass effect is not visible in the photometry of individual T dwarfs, but if I can measure it in the aggregate over all the nights, it will give me a measure of the mean precipitable water vapor in the atmosphere during these observations.
No comments:
Post a Comment