We have acquired hundreds of thousands of short duration (few seconds) unfiltered images of regions containing open clusters. We have been analyzing these images to test their suitability for photometric studies. We break the images from a given night into groups of several hundred. Within a single group we force the average star flux to remain constant from frame to frame. We then histogram the brightness of each star and apply a Gaussian fit to the result. We have found that on most nights 200 to 300 stars meet our arbitrary criterion of the standard deviation of the fit divided by the mean of the fit less than 6 percent. Of these stars we look for excess deviations of a certain size, either high or low.
One goal is to be able to detect objects in the outer solar system that occult a star for a fraction of our 3.5 second exposure. The result would be an apparent dimming of the background star. These objects might be part of the Kuiper Belt known to exist beyond the orbit of Neptune or they might lie beyond these previously detected objects. We are also searching for positive-going events that could arise from gravitational microlensing that occurs when a dense, massive object passes near the direct line between an observer and a distant star. We have found that many objects, such as a passing satellite, can mimic the signal we are searching for. (See below.) Our current efforts are directed at improving the calibration of our images, understanding spurious events, continuing data acquisition, simulating possible events and searching previously acquired data for signal.
Our ability to learn something by counting fluctuations is critically dependent upon acquiring the most data possible. We acquire 1 to 12 Gb of data each clear night, depending on where we are in a given data cycle and the time of year. When we are in a data cycle we are compelled to take data every clear night. Sometimes, however, you need to know when to give up, go home and get some sleep. This work is also tough on the equipment. Camera shutter failure is the most common form of mechanical trouble. The tell-tale sign of failure.
|This plot shows the histogram for a star, on July 16, 2002, with a Gaussian applied to it. There was a 7.4 σ low event in this data set. This was caused by a satellite passing near the star causing the sky background (and not the star itself) to increase in brightness.|
Not surprisingly, our data sets contain interesting information on populations of variable stars in open clusters. We have developed a novel technique to search for variable stars in unfiltered images. Read an abstract for a presentation we made at the 94th annual meeting of the AAVSO in Newton, MA, October 2005.
A video of the presentation can be found on the AAVSO website. Go visit the site and explore.
Our densely sampled data sets are particularly useful for detecting stellar variability on timescales of a few to several hours. Since we are our looking at fields containing open star clusters of all different ages we are are particularly interested in the development and evolution of short period eclipsing binary stars in open star clusters. These systems consist of two stars in orbit around each other, often with orbital semi-major axes such that the surfaces of the stars are in contact. It is not completely understood how such star systems form and evolve. We hope to help constrain models of evolution by surveying these systems both within and outside the clusters in our fields. Since the ages of open star clusters are fairly well known, we might be able to use our data to help determine the rate of appearance of these systems as a function of age. This information would be key in refining models of contact binary formation. Of course, nothing is ever as simple as it sounds. To do the work well we need a broad sample of cluster ages and we only add one or two new clusters per year. In addition we only find a few short-period eclipsing binaries in each field and those systems may or may not be part of the cluster. It is often not easy to tell. All we can do is keep plugging away. We also hope to detect any variations in period and amplitude of the systems we are studying. Below is a sample lightcurve (=brightness versus time graph) from one of the variable stars we have found.
At some point in their evolution some stars enter a phase of significant pulsation. One type of pulsation causes stars to to get brighter and dimmer with a period of scores to hundreds of days. We are also able to detect this type of variation quite efficiently. Below is a lightcurve of one such star we have been monitoring for a few years now.
The two images below show small region of our field (about two arcminutes on a side) containing one of our variable stars. The image on top was taken on April 6, 2009 and the image below was acquired on August 16, 2009. Clicking either image will play a movie (.wmv) of 16 nights of images of this field spanning all of our 2009 data.
It is important for us to know the colors of the stars we are studying. Since we are using unfiltered images the conversion from measured signal to star brightness is strongly color-dependent. Also, we expect variables stars of particular types to have specific colors. Thus, we take some images of our fields through standard BVRI filters. One way we use these data is in the construction of color-magnitude diagrams. These are graphs of brightness measured in the magnitude system versus color for all the stars in the field. The color is the brightness (again in the magnitude system) of a star in one filter minus the brightness in a different filter. A color-magnitude diagram for the brightest third of the stars from one of our fields is shown below. Stars we have measured no variability in are represented with black dots. Long period pulsating stars are represented by red squares and short period eclipsing binaries are represented by red diamonds. The main sequence for the cluster appears neatly as the nearly vertical band of stars to the left. The eclipsing binary stars fall near this band and the long period pulsating stars show up as very red, as expected.
As part of the ongoing project we return to the field containing the open cluster M23 each year. In 2010 we are acquiring our seventh season of data from this field. The above color-magnitude diagram is, in fact, for the M23 field. During our first three observing seasons we detected unambiguous variability in thirty stars in this field. We continue monitoring this field looking for changes in the periods and amplitudes of these variables as well as luminosity "drifts" in apparently normal stars. The announcement of the discovery of the M23-field variables can be found in an abstract for a presentation we made at the 210th meeting of the American Astronomical Society in Honolulu, HI, May 2007.
Read an Astronomy Magazine Web News Item describing this work.