next up previous contents index
Next: The HST-GHRS Spectrum Up: StarsStellar Populations, Previous: HST Studies of

Science with the Hubble Space Telescope -- II
Book Editors: P. Benvenuti, F. D. Macchetto, and E. J. Schreier
Electronic Editor: H. Payne

Eclipse Decomposition of Time-Resolved UBVRI and HST Spectra of the Dwarf Nova IP Peg

S. Wolf
Universitätssternwarte München, Scheinerstr. 1, D-81679 München, Germany

Keith Horne
University of St. Andrews, School of Physics and Astronomy, North Haugh, Fife KY16 9SS, Scotland, UK

T.R. Marsh
University of Southampton, Department of Physics, Highfield, Southampton S017 1BJ, England, UK



A light curve decomposition tool ( LFIT ) developed by Keith Horne and Tom Marsh and its application on HST-data of the eclipsing dwarf nova IP Peg will be presented. This paper is based on observations collected at Wendelstein Observatory (Germany) and HST-data obtained from 1992 to 1994.

Keywords: binaries, cataclysmic variables


The eclipsing dwarf nova IP Peg has been monitored since 1990 using the high-speed multichannel multicolor photometer (MCCP) of the Universitaets-Sternwarte Muenchen (USM). Several UBVRI eclipse runs in 1992 and 1994 are simultaneous with HST observations, affording a very wide 1200--8000Å wavelength range to study the radiation characteristics of different components ranging from the cool secondary star to the hottest accretion regions in the inner disk and the bright spot where the gas stream crashes into the disk rim.

Our LFIT code decomposes time-resolved spectral data into N components with parameterized light curve shapes, solving for the spectra of each component. We used this successfully on HST spectra covering an OY Car eclipse to isolate separate spectra of the white dwarf, bright spot, and disk components (Horne et al. 1994). To apply this algorithm to the HST and UBVRI eclipses of IP Peg, we add an ellipsoidal light curve for the companion star, and a more detailed description of the bright spot. A program demonstration and first results of component light curves and spectra are presented.

How does LFIT work?

LFIT is a program which decomposes eclipse light curves by fitting a N-component light curve model to the observational data. Each component is described by parameters. The whole set of parameters ( , number of parameters) defines the shape of the model light curve which are fitted to the observed data at each phase and wavelength . The scaled sum of all N light curves builds the total model light curve. The wavelength depending scaling factors derived by a linear least square fit to the data finally represent the mean spectrum of each component:

Thus this algorithm allows a simultaneous decomposition of light curves and spectra.

In this case the model light curve is divided up into 4 components: the primary (white dwarf), the secondary, the accretion disk, and the bright spot.

Figure 1 (top left) shows the assumed fundamental CV-Model (bright spot profile and orbital plane), the -evolution of the fit, the final parameter values and at the bottom the fit to the HST-data light curve of IP Peg on the 4th of November, 1992.

The HST data

The following part will be confined exemplarily to the third run observed on the 4th of November, 1992. For details on the data processing and the observational equipment see Baptista et al. (1994).

The bright spot is the most prominent feature seen in the light curves of IP Peg, manifested by the hump around phase -0.2, before going into a remarkably deep eclipse due to the occulting secondary star. At the end of the eclipse, the white dwarf is first to come into the line of sight followed by the egress of the bright spot---both phenomena represented by small steps in the light curve at phase 0.04 and 0.08, respectively.


Figure 1 (bottom left) shows the best light curve fit to the HST data of run 3: The fit itself, the component light curves (dashed and dotted lines), the observation and the residuals of the fit at the bottom of the plot. The latter may be interpreted as flaring or flickering, especially during phases before the eclipse. The fit reproduces the shape of the eclipse in almost every detail---the egress of the white dwarf and the bright spot are well copied. Only during the hump phase () the fit deviates largely, as none of the components describes such a high frequency feature. The corresponding parameter space which leads to this best fit is displayed on the top panel of figure 1.
The parameters are basically not in contradiction to values derived by an UBVRI fit. Fixing the mass ratio and the white dwarf's width of eclipse while running the fit to the UBVRI data, the rest of the parameters did not extremely deviate from those of the HST-fit. The disk radius increased by 20%. That is not unusual, as the disk radius may change between successive outbursts. The direction of maximal bright spot flux emission changed to a smaller angle, so that the emission is not as perpendicular to the gas stream's trajectory as in the case of the HST observing run.

As already seen in the component light curves, the primary is the faintest object of IP Peg in the UV and so is the spectrum, there is almost no signal (figure 1, right). The brightest feature is the hump, thus the spectrum of the bright spot is the strongest. The most striking feature in the disk spectrum is the CIV line, which is not present in other components---in the bright spot spectrum it seems rather to be in absorption. The HeII line which is usually seen during outburst, disappeared in these quiescence data.

In future projects we will try to improve the quality of fit. The fit seems to prefer larger values at the short side of Ly. Looking at the light curve's behavior approaching these wavelengths, you will find a change of the eclipse shape, which may be the cause of the tendency in the fit (figure 1, right). Moreover, our biggest efforts will be, to build a model of the most prominent UV feature, the bright spot, by fitting a series of bright spot profiles to the region, where the gas stream crashes into the disk. Together with the long timebased UBVRI-data and the HST-UV-data we should be able to determine temperature distributions and fluctuations as well as mass transfer rates which are closely connected to the crucial parameter, the viscosity of the disk.


Baptista, R. et al. The HST Observations of IP Pegasi: A First Look at the Data 1994, ASP Conf. Series, 56, 259.

Horne, T. et al. 1994, ApJ, 426, 294.

next up previous contents index
Next: The HST-GHRS Spectrum Up: StarsStellar Populations, Previous: HST Studies of