Astron. Astrophys. 362, 595-598 (2000)

2. New photoelectric measurements and data analysis

The observations were made applying the 85 cm reflector installed at the Xinglong Station of Beijing Observatory, China, with a six-channel two-filter photometer which is a new version of the one once used in the STEPHI Network campaign (Michel et al. 1995). During the observations, a v filter with a center wavelength of 411 nm and bandwidth of 20 nm and a y filter with a center wavelength of 550 nm and bandwidth of 40 nm were used. The stars SAO 59589 and SAO 59576 were selected as comparisons. The observations in the v filter were of high quality, a good data set covering eight nights (about 70 hours) was obtained within the measuring error of about 6.0 mmag. For the y filter, the observation error is a little larger than that of the v filter. In some nights a drift was very obvious, which maybe caused by the multichannel photometer or the variable star. The measuring error for y is around 7.0 mmag; the data set also covers eight nights (about 70 hours). No evidence of variability of the comparison stars was found.

A multiple-frequency analysis of 59 Aurigae was performed with a package of computer programs employing single frequency (Fourier) and multiple-frequency least-squares techniques (program PERIOD, Breger 1990a) which utilize Fourier as well as multiple-least squares algorithms. The latter technique fits a number of simultaneous sinusoidal variations in the magnitude domain and does not rely on prewhitening. As the quality of the y band is not high enough, our period analysis is mainly based on the v filter data.

Fig. 1 shows the power spectra of the v data before and after subtraction of the best single to eight-frequency solution. At the top of the figure the spectral window based on the times of available observations is plotted. We note that alias patterns, including the 1 c/d daily aliases, are quite low in the spectral window. The existence of eight frequencies of pulsation (6.8283, 6.3179, 5.0074, 5.9535, 4.3285, 5.6854, 9.0206 and 8.6260 cycles per day) is suggested. This is also shown by their S/N ratios. The noise has been calculated as a function of frequency after subtraction of the best eight frequency solution. The eight detected frequencies have amplitude S/N ratio of 11.0, 9.1, 9.6, 9.9, 5.4, 5.3, 4.0 and 3.8, respectively. The first 6 frequencies are convincing and are accepted as real. The seventh and eighth frequencies lie in the region between being selected or abandoned. Considering that the eighth frequency is the best solution of observations of Gupta (1980), we chose to keep the last two frequencies in our frequency solution. After removal of the eight frequencies, the residuals are of the order of 7 mmag. More peaks, which are a little above the noise level, also can be picked up. It seems that other pulsation frequencies may also exist in 59 Aurigae.

Fig. 1. The power spectra of 59 Aurigae for the v band. The spectra are shown before and after applying multiple frequency solution.

As the y band data are not of high quality, it is difficult to obtain the same solution from the period analysis of y data alone. However, the eight-frequency solution fit the light curves very well (see Fig. 3). The fitted residuals are around 7.5 mag; this is very close to the observed uncertainty. Table 1 shows the eight-frequency solution to v and y bands data with their amplitude as well as significance of detection from v. The fits to the data are displayed in Fig. 2 and Fig. 3. It is necessary to point out that not all the amplitude differences between v and y are completely real, especially for and . We suggest that some system error caused by the binary system or the drift between different channels may be significant.

Fig. 2. The fit of the eight-frequency solution to v band data.

Fig. 3. The fit of the eight-frequence solution to the y band data.

Table 1. The frequency spectrum of 59 Aur.

In order to check the reliability of our eight-frequency solution, we fit our frequency solution to the data obtained by Gupta from 1975 to 1976. We list the frequency solution in Table 1. With the exception of the low-frequency term =4.3285 that is not clear and is omitted, other frequencies fit the original solution very well. The fitting residuals lay in the range of 7.1 mmag, which is better than that of the solution of Gupta (1980). As to why the frequency 4.3285 disappears, we think there are two possibilities. One possibility is that the value 4.3285 is not accurate and may be caused by the extinction or drift of photometer. The other possibility is that the value 4.3285 is correct, but the data set of Gupta is not large enough to cover this frequency. In all observations of Gupta (1980), the longest continuous observation is around five hours and among all observations, only two nights are consecutive. In this kind of data file, it is difficult to cover a frequency that is smaller than 4.5 c/d. We display the fits in Fig. 4. The result is much better than the frequency solution obtained by Gupta himself.

Fig. 4. The fit of the eight-frequence solution to the data obtained by Gupta in 1975-1976.

© European Southern Observatory (ESO) 2000

Online publication: October 24, 2000
helpdesk@link.springer.de