Short Duration Exposures of Rapidly Variable White Dwarf Stars via the Whole Earth Telescope Ian Radtke (Minnesota State University, Mankato) Dr. Steve Kawaler (Dept. of Physics and Astronomy, Iowa State University) Dr. Reed Riddle (Dept. of Physics and Astronomy, Iowa State University) Abstract We describe a new CCD-based time-series photometer currently under development for use with the Whole Earth Telescope and compare this emerging system to the current three-channel PMT photometer. of I. Introduction Texas Astronomy Department In astronomy, as in all walks of science, established a world-wide network of improvements in technology are coupled cooperating astronomical observatories with improvements in theory. As time- to series photometry has matured over the measurements of variable stars. last two decades, advances in the “Whole Earth Telescope”, hereafter theories stellar referred to as WET, currently consists of pulsations and stellar seismology have 23 observatories in 16 countries around quickly the world. of stellar interiors, outpaced instrumentation. advances in obtain uninterrupted time-series This This cooperative optical This discrepancy is observing network is distributed in primarily due to the lack of hardware longitude around the Earth and is capable of coordinated from a single site so that it “uncluttered” resolution that is fine acts as one instrument. Nightly data sets enough to provide direct observation of from each telescope are sent to the WET current theoretical predictions (Nather et headquarters via email allowing for real- al, 1990). time data reduction, combination and of providing a level analysis, maximizing the effective use of In 1986, to fill this gap in hardware the entire distributed instrument capability, scientists from the University (http://wet.iitap.iastate.edu). By making proper use of the WET, gaps in time- brightness of white dwarfs can be most series be readily thought of as “starquakes”. minimized or completely eliminated, Temperature fluctuations within the star allowing for greatly improved sensitivity result in the vertical movement of dense and resolution. material above its equilibrium point. photometric data can This high-density material is relatively II. Science hotter, therefore brighter, than the Photometry is, literally, the measurement surrounding low-density material which of light. Viewed from an astronomical causes variations in the amount of light perspective, photometry generally refers emitted across the surface of the star. to stellar The small vertical motions of surface Time-series material also generate larger, horizontal photometry extends these measurements motions (e.g. waves across the surface of to observe time dependent variations, the star) (Kawaler & Dahlstrom, 2000). such as the multi-periodic oscillations To fully understand such rapid and observed in rapidly variable white dwarf complex stars. Simply put, the goal of time-series observational coverage is essential. the measurements magnitudes and colors. of oscillations, extended photometry is to extract, from the light curve of a variable star, information of The collection and study astrophysical interest about its internal information, composition and behavior. asteroseismology, has been the driving now of known this as force behind the WET since its inception. Pulsations of variable white dwarf stars With a distributed instrument such as reveal information about their internal WET, it is possible to minimize or structure and composition, age, rate of remove the gaps in the time-series rotation, magnetic field, mass and photometric data acquired. distance. These stars often pulsate in a WET run will produce roughly ten times hundred or more modes simultaneously the amount of data that a series of and have periods on the order of two to observations at a single site. twenty phenomenal minutes (Kleinman, 1996). These periodic changes in the relative gain in A typical The capability, compared with that of a single site, resulting in such a high data density has sky brightness, it became possible to virtually limitless possibilities. (Nather observe et al, 2000). telescopes than were originally thought dimmer stars on smaller useful. Such a run is controlled by a laptop PC which records the data and III. Equipment astronomy plots a real-time light curve of the target observation is always the equipment star, comparison star and sky. (Nather et used. al, 1990). The backbone of any Improvements in equipment almost always lead to improvements in the quality of data collected which, in The next evolutionary step in WET turn, leads to improvements in analysis hardware and understanding as well as more comparatively bulky PMT photometer to accurate theories. a much svelter CCD-based time-series is to shift from the photometer. A CCD photometer offers Although the distributed nature of the the ability to track multiple companion WET allows for continuous coverage of stars and has a higher sensitivity than its a target object, it also adds the difficulty PMT predecessor. Again, a laptop PC of multiple, often overlapping data sets. controls the instruments, records the data, To time-series and plots near real-time light curves of observations from multiple telescopes the target star, up to four companion into a single, comprehensive light curve, stars, and sky brightness. best combine the uniformity of hardware, software and observing procedure is a must. When As this system is aimed at taking short the WET was created, a two-channel exposures (on the order of 10-15 seconds) photomultiplier tube (PMT) photometer of rapidly variable stars, keeping the was adopted as the hardware standard. camera exposed to the target is of utmost As the WET has matured, the original importance. To do this, the CCD must two-channel PMT photometers were read out its information as quickly as retrofitted with a third channel allowing possible in order to minimize exposure for a continuous measurement of the sky down-time. Unfortunately, this read out brightness. By keeping better track of latency is the bottleneck of the current hardware. The CCD system currently IV. Software under development runs on the Linux As the hardware of an observing system operating system with an Apogee AP7 is as the backbone of the body, the camera. While the AP7 is an excellent software driving such a system is its camera, it is severely limited by its ISA nervous system. The software must be interface which throttles read out speed. agile enough to juggle the tasks of data While a faster PCI-based interface is acquisition available, Apogee has yet to write simultaneously. Thanks to the excellent microcode to drive an AP7 at these work of Dave Mills from The Random higher speeds. As it now stands, reading Factory (http://www.randomfactory.com) out an unbinned 512x512 pixel image, a software development kit (SDK) for roughly 1 MB of data, takes about ten Apogee CCD cameras under Linux is seconds. available in the public domain. For single image exposures, and rough analysis the ten second delay is a small hindrance but for time-series photometry it is a The Apogee SDK is coded in a mixture serious problem. Thankfully, Apogee is of TCL/TK for the user interface and currently developing a new line of CCD basic cameras that incorporate an RJ-45 SAOImage DS9, also written in TCL/TK; 10BaseT Ethernet connection. http://hea-www.harvard.edu/RD/ds9) With image manipulation (via such a lofty read out speed, taking full, while the “heavy lifting” of camera I/O unbinned images will become a trivial in done in C. task. If the image size remains the same, interface (i.e. the light curve and reading out 1 MB of data on the new comparison windows) are written in camera will take just under one second. TCL/TK while the image correction, Such a dramatic improvement will centroiding, and count extraction are eliminate written in C. virtually all gaps in All additions to the observation. Technical specifications on the Apogee AP7 CCD camera can be Using the routines provided in the SDK, found direct interaction with the AP7 is easily http://www.ccd.com/apseries.html. at accomplished. In addition to low-level I/O, the SDK provided a framework for specific requirements of our time-series Preliminary Image is Taken photometry, a variety of routines were written to extend the capabilities of the base package. While the generic control window was little modified (as seen in Fig. 2), entirely new structures for light Subregions are Selected curve plotting and manipulation have been coded. As seen in Figure 1, the process of image acquisition and analysis can be explained through a Image Correction simple flowchart. As with any observing run, a series of bias frames, dark frames and flat field frames must be taken to allow proper Centroiding on all Stars image correction. Through a simple pull-down menu, these three libraries are very simple to create. A bias library is created simply by inputting the number Count Extraction of frames desired. The dark library is one step more complex, requiring not only a number of frames but also an exposure time no less than the exposure time to be used in observing, optimally Next Image is Taken identical. Flat fielding is equally complex, requiring both an exposure time and the number of frames desired. Once these libraries are created, a Figure 1: Basic process of image acquisition and rough analysis. preliminary image is taken and saved as a standard FITS file. From this image, a rudimentary graphical user interface to ease operation of the system. Due to the the target star, up to four companion stars and a sky region are prompted for. robust telescopes. From these corrected The software allows for either a Box subregions, photon counts are extracted subregion or an Annulus subregion to be and plotted. The next exposure is then selected. Using either subregion type, taken and the above process repeats. the aperture is adjusted on the displayed image and sized by the observer to best As the light curve is produced, direct fit the target star. Any companion stars manipulations of the resulting plot are that the observer elects to use are forced available to the observer. (Figure 3) to use an identical aperture size to Both the X- and Y-axes can be moved to provide for direct comparison between view a particular region of the data and plotted light curves. If the Box both are independently scalable to allow subregion has been chosen, a Sky for a maximum of functionality. Other subregion is prompted for. The Annulus options include viewing the entire data subregion will not prompt for a discrete set simultaneously, a “live” view that Sky subregion as it is already accounted will adjust the plot as new data points for through the outer annulus of each are created and an option to display the chosen star. After subregions have been last corrected image with associated selected, the frame is corrected by the subregions in DS9. bias, dark and flat libraries which will minimize any intrinsic flaws in the CCD hardware. While not necessary after only one exposure has been taken, each chosen star undergoes a centroiding process which adjusts apertures to correct for unavoidable drift between the camera and the sky. Regardless of the quality of stellar tracking inherent to the telescope used for observation, the centroiding process is an easy step to take which greatly improves the observing capabilities of small, less Although the software supports up to five simultaneous stars, it is often beneficial to directly compare only two of the light curves. The “Compare” button from the main light curve window prompts the observer to select which two targets are to be viewed and new, larger plot is created. Both light curves are combined in the new window with a similar set of control and zooming options are available. In this “Light Curve Comparison” window, alterations to the X-axis will affect both plots while instruments, the increased portability is the Y-axes are both independently still a benefit. movable and scalable. Again the “All” and “Live” buttons are available. (Figure 4) As the summer winds to a close, the last two and a half months of development have shown TCL/TK to be a less than V. Summary and Future Plans optimal development language for our While the CCD system is still under purposes. development, a few distinct advantages development is to rewrite the entire code over the PMT photometer are apparent. base in a more robust, lower level Firstly, up to four companion stars can language such as C, C++ or Java. This be selected where the PMT can only will allow for much faster image accept one companion star in addition to processing simply due to improved a discrete sky brightness reading. These memory additional companion stars allow for handling solutions that can be found more precise tracking of the target star, outside of TCL/TK. More functionality especially when mounted on an Alt- can be gained by fully multithreading the Azimuth telescope which will cause the code to allow image acquisition to image to rotate as the observing run continue progresses. Secondly, not only is a light encountered in analysis (e.g. a star curve produced but all of the CCD moving outside its aperture or off the images are saved allowing for a much field entirely). Once the error has been more thorough post-processing. This is corrected, the images taken since the an advantage over the PMT system error occurred can be processed and where only the light curve is retained. added Lastly, the difference in size and weight Multithreading will also decrease overall between the PMT system and CCD latency by allowing the acquisition system is significant. While not an issue thread to run unhindered by any analysis. of major importance, as each telescope All processing will occur independently houses the majority of their own of the image acquisition process. Our first priority in further management regardless to the and of light memory any errors curve(s). While much has yet to be done to finalize the CCD time-series photometer as the new WET standard, large strides were taken in the past ten weeks, development continues and, with some perseverance, the new hardware standard will be available to the WET for its next observing run. References: Kawaler S.D, Dahlstrom M., 2000, American Scientist, 88 Kleinman S.J., Nather R.E., Phillips T., 1996, PASP, 108, 356 Nather R.E. et al, 1990, ApJ, 361, 309