A Search for Novae in the Andromeda Galaxy (This research report has been
published in the RBSE Journal By the 1998-99 Astronomy classes of Grosse Pointe North High School. Jeff G., Student P.I. Ardis Maciolek, Teacher. The research team discovered 21 possible novae in the Andromeda Galaxy (M31). Sixteen data fields taken at Kitt Peak by the 0.9m telescope from 1995-98 were inspected for stars that showed sudden brightening in the hydrogen alpha portion of their spectrum. The magnitude and location of all candidates were determined, and light curves were developed for novae that appeared on consecutive fields. Of the 9 light curves analyzed, five appeared to be type NA "fast" novae, and two each of type NB and NC. Novae appeared to be distributed close to the galactic core, probably due to a density effect- more stars, more novae. "Fast" type novae were closer to the center and the slowest type farther out. The purpose of this research project was to search for novae in the Andromeda Galaxy, and once identified and located, to observe their changes in magnitude as a function of time. The National Optical Astronomy Observatories supplied digital CCD images on CD-ROM. The CD’s contained images of M31 taken on 16 nights, from September 1995 to November 1998. All were 10-minute exposures through a hydrogen alpha filter at 656.3 nm. The images were analyzed using a combination of two different software packages and computers: Scion Image Beta-3 on PC, and NIH Image 1.62 on Macintosh. Special macros written for both of these programs allowed for importing the FITS images, and determining magnitude and location by celestial coordinates. In the search for nova each student received one of 16 subrasters. Each subraster was part of the original image of M31. In each subraster folder there was 16 epochs. Each epoch was the time that the pictures were taken. Approximately 40 students participated in searching for nova candidates by using software to stack the images and to "blink" them. Novae would appear in some images but not all. If seen in several images, their brightness would change from one image to the next. A hydrogen alpha filter removes all but one specific wavelength of light for observing. Novae emit most of their light in this wavelength so the use of this filter makes it easier to detect them. Once a possible nova was identified, its location was recorded using both the "x, y" pixel coordinates and a special macro that located the star by right ascension and declination. The magnitudes of the novae were determined by using a special photometric macro routine and a set of "standard stars" for each subraster, supplied by the NOAO. The known magnitudes of these standard stars were entered, and the accuracy of the photometry was enhanced by using rescaling and aperture "sizing" routines. This proved to be the most complicated part of the project and most of this work was done by five people. Rescaling the pixel range was a key consideration both in searching for novae and in getting the magnitude macro to perform properly. Images near the galaxy core, such as subrasters 6,7,10 and 11 were rescaled from 0-3000. Images at the edges of the galaxy were rescaled as small as 0-500. A rescale value of 0-2400 seemed to be a good place to begin. For very faint novae, altering the star’s effective aperture to "1-2-4" in the magnitude routines allowed the macro to be used to determine stars as faint as 17.6. Taking into consideration certain factors eliminated false nova candidates. One of the factors was weather. A bad weather field made it hard to tell if it was a nova or just a star that sometimes wasn’t imaged. The edges of the pictures were also a factor when eliminating nova candidates. In some pictures there may have been jagged edges or missing data. One more thing that can also eliminate nova candidates is an artifact. When magnified, square or rectangular shapes indicate bad pixels in a CCD camera. In scientific research it is important that one’s data can be reproduced, proving that one’s work was done with a high regard to precision. With this concern the Nova Search Team had to take some extensive measures to back up its findings. At least two students were assigned to each subraster with the duty of searching for novae. This led to the next step of comparing the results of students in the same subraster. This double check of data could result in a conflict, where a third person was then needed to recheck each student’s claims. The rechecking of results was possible by the location of the nova’s right ascension and declination. Some additional debate concerned novae appearing in one picture, but gone in the next. The presence of a reoccurring nova takes a decade or longer, making it impossible to occur during this research project due to the time lapse of only a few years. This information allowed elimination of some of the suspected novae that repeatedly appeared and disappeared. More than half of the nova candidates were seen only in one image. Due to the sporadic dates of the images, it is possible that the ones reported are novae, even though there is not enough data to develop a light curve. The error for the novae magnitudes was determined by using the magnitude macro to calculate the magnitude of standard stars. At least two and usually three or four standard stars per image were checked. The largest deviation from known magnitude on each image was recorded as the "Image Error". Then the largest reported error range (by NOAO) of the standard stars from each image was referenced. This was recorded as the "SS (Standard Star) Error". These two errors were summed quadratically to create the "Combined Error" for each magnitude. The following chart shows all available data on each of the novae. Column 1 is the team designation for the nova discovered. The X and Y coordinate values reported are the numbers listed nearest the R.A. and Declination values in the dialog box- this clarification is necessary because the dialog box simultaneously shows two different sets of X,Y values.
Twenty-one nova candidates were located and verified on a total of forty-eight images. Nine of the 21 were identified on multiple images. Magnitudes were determined for each nova image, and light curves were calculated for the nine repeated novae, using Cricket Graph. These were then analyzed by extrapolating the light curves to three magnitudes and/or 100 days past maximum light. The light curves were tested and fit in the following way: Three different fits were tested: a first order polynomial (line), a second and a third order polynomial. The R^2 value for each calculated best fit equation was examined and the fit that gave the best correlation and a logical plot was chosen. A "logical plot" would approximate what would be typical of a nova- either a steady decline and/or rise or a fairly constant magnitude. Curves were rejected if, for example, they showed the nova brightening, then dimming, and then brightening again, even if that correlation was higher than other curve fits. For all graphs, magnitudes were entered as negative values, so that brighter magnitudes appear as higher points. All magnitude values are actually positive. Nova Light
Curves - Type NB/NC The location and characteristics of each nova was noted, in order to develop a model for the typical nova observed. By definition, novae that fall by >3.0 magnitudes in less than 100 days are type NA or "fast" novae (Sterken and Jaschek, Light Curves of Variable Stars, 1996.) Type NB or NC novae decline more slowly. The decline of NC class nova is very slow, and the star’s light may be near maximum magnitude for years. The following chart summarizes the findings of the team based on the above definitions. The Curve Fit shows P1 for a linear best fit and P2 for a second-order polynomial fit. The Correlation column shows the value for R^2. Nova Analysis Chart
The Nova Search research team discovered a total of 21 candidate novae, most in the high-density areas of the spiral arms of the Andromeda Galaxy. Of these, five were determined to be NA novae, two of type NB and two of type NC. The average magnitude of the novae was 16.46, and they ranged from 17.66 to 15.61. By examining the locations of the novae, there is no obvious reason why they are distributed this way, other than to conclude that where there are more stars, there are likely to be more novae. However, all the fast (NA) novae tended to be located nearer the center, and the slowest (NC) ones were distributed farther away from the core. The Nova Search team would like to thank the following people for their help with software or research issues: Joe A., Jamie Elsila, Tom Gehringer, Larry Kendall, Travis Rector In addition to our P.I., recognition for special dedication to this research project goes to the following students: Michelle S. and Brent H., George K. Chaisson, Eric and Steve McMillan. Astronomy: A Beginner's Guide to the Universe, 2nd edition. Prentice-Hall, 1998. Scovil, Charles E., editor. The AAVSO Variable Star Atlas, 2nd edition. 1990. Sterken, C. and Jaschek, C. Light Curves of Variable Stars. Cambridge University Press, 1996. |