[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
quick comment on photometry from DS23
I'm continuing to analyze properties of the "collected.big"
file Tom Droege created in DS23. A short summary of the photometric
results so far:
- both V-band and I-band measurements show much smaller scatter
within a single night than including many nights.
- the I-band measurements show strong residuals as a function
of position in the frame.
I tried concentrating on a small area on the sky, just 0.4x0.4 degrees.
Could I use ensemble photometry within this tiny area to reduce the
scatter of measurements taken on many nights? Yes! I explain this
as follows: measurements of a given star on different nights very likely
fall on very different parts of the chip, whereas measurements on
a single night (in "follow mode", where the camera follows a piece
of sky for two hours) all fall on the same area of the chip. If there
is a simple shift in response from one area on the chip to another
(due to imperfect flatfields, for example, or vignetting), then allowing
each frame to shift up or down a bit permits one to match responses
from night to night. The ensemble method of solution does just that.
With an ensemble solution, the V-band measurements come close
to the scatter predicted from the statistics of counts in each image.
There's still a small extra noise, which I suspect may be due mostly
to noise in the flatfield and dark frames. I'm pretty happy with the
V-band data -- the scatter in an ensemble solution is less than 0.01
mag up to V=10, and increases to 0.02 mag at V=11 and 0.10 mag at V=12.8.
However, the I-band data is much worse. Even with an ensemble
solution, the scatter is still about twice that predicted from
the statistics of counts in each image. The scatter reaches 0.01 mag
somewhere around I=9, up to 0.02 mag around I=10.5, and 0.10 mag
around I=13. I suspect that the I-band flatfielding could be
improved; perhaps vignetting by tree limbs or a roof might be
responsible for some of this, too. I haven't looked at the change
in PSF across the chip, so I don't know how important that might be.
If one is most interested in variability, then one should send
all the data through an ensemble solution. However, I don't see
immediately how to place the zeropoint of the ensemble solution
back onto the photometric system properly: each ensemble will likely
include stars from such a small region on the sky (less than one degree
on a side) that there will be no good standards in the area.
Sigh. Must think on it.
That's all for now. Tech Note will eventually appear, full of