The SkyProbe Camera
SkyProbe consists of a short focal-length lens (aoubt 50 mm) and a
small SBIG CCD. It images a 5x7 degree patch of the sky roughly
centered on the CFHT field of view. The goal of this camera is to
measure the transparency of the sky and to provide in real time a
quantitative indication of the current attenuation by clouds.
In the current setup, a 35 second exposure is taken approximately
every minute. These images are automatically analysed with tools from
Elixir: dark and flat correction with Flips, followed by stellar
photometry with Sextractor (E. Bertin). The stellar measurements are
calibrated astrometrically and photometrically by reference to the
Tycho Catalog. The resulting calibrated measurements are incorporated
in an Elixir photometry database.
In a typical image, there will be approximately one to two hundred
stars matched with the Tycho Catalogue. The mean of the magnitude
differences is computed and this information provides a measure of the
sky transparency. Each data point in the magnitude attenuation plots is the
mean of the magnitude differences of a single image.
Currently, the error in the magnitude attenuation is at ~2% level, and
is succeptible to systematic errors caused by the sample of observed
stars. As a result, it is sometimes hard to distinguish between
actual attenuation due to clouds and other effects. For this reason,
we provide here some references on how to interpret the data.
Interpreting The Plots
The following picture shows an example of a photometric night as seen
with SkyProbe. Ideally, on a photometric night, the magnitude
attenuation will center around zero with an error that is
characteristic to our system. At this moment, the error in our
measurement is at the 2% level, i.e., deviation from the baseline will
in general be within 0.02 magnitudes.
Unfortunately, our measurement of the magnitude attenuation is
sensitive to bias, as discussed above. This leads to distinct
segments seen in the plot. In general, abrupt changes like this are
most probably artifical. Within each segment, the deviation from the
mean is quite small. This lends further support to the shift being
caused by biased measurement rather than real attenuation. Notice the
slight swelling around t = 7. At this moment, the analysis procedure
is not accurate enough to tell us conclusively whether it is due to
clouds drifting over or some other artificial effects in the analysis.
The bottom panel in the plot is the standard deviation of the mag
attenuation in an 1 hr window.
Below we show an example of a non-photometric night. See the archive for more example nights.