Astron. Astrophys. 342, 745-755 (1999)

2. Observations

The multisite campaign on HS 2324 was performed during 2 weeks in August-September 1997, centered on new Moon. The journal of observations in Table 1 gives information on the observatories involved, telescopes and instruments used, and duration of the single runs. Most observations were obtained using two or three channel photometers with bialkali photomultipliers (EMI9784QB for Loiano, Hamamatsu R647 for Beijing and McDonald), no filters, and an integration time of 10 s, which was subsequently merged to 90 s. The leak of sensitivity to periods shorter than 180 s did not give us any trouble because no signals were detected in that range from a preliminary analysis of the photometer data alone. Only the Calar Alto data were collected using a SITe#1d CCD, B filter, and an exposure time of about 20 s (first two nights) or 40 s (following three nights) for each datum; the times between successive data points vary between about 70 and 100 s. The error introduced by the different effective wavelengths of each detector has been evaluated to be not more than 5 in amplitude.

Table 1. Journal of the observations

Despite the small number of participants in the campaign, only four, we obtained a good coverage, comparable with other multisite campaigns, thanks to the good weather conditions in most nights at the different sites. The complete (and combined) light curve is shown in Fig. 1; it has a total duration of 134.9 hours, with an overall duty cycle of 43. In the central part of the run (40.8 hours), the duty cycle is 98.

Fig. 1. The complete light curve of HS 2324; each panel represents 24 hours.

2.1. Details on data acquisition and reduction

We followed basically the same data reduction procedure as described in Handler et al. (1997). Here we summarize this procedure and give some detail on a few differences.

For the Beijing and McDonald photoelectric data, we chose the same comparison star already used by Handler et al. (1997), which was also one of the comparison stars used in the Calar Alto CCD measurements. For the Loiano photoelectric data this was not possible because the box of channel 2 is more distant from channel 1: therefore we used the same comparison star already used by Silvotti (1996). Both comparison stars were tested again for photometric constancy and found not to be variable. We then turned to sky subtraction. For the Beijing measurements, where a third channel was available, the sky background could be monitored simultaneously. In this case we subtracted the sky counts on a point by point basis. To reduce the scatter of the background measurements, some smoothing was applied whenever possible. At McDonald and Loiano only two channels were available. In this case sky was measured using channel 1 and 2 for about 1 min at irregular intervals of typically 20-90 min, depending on sky stability and presence of the Moon. The sky counts were then interpolated linearly and subtracted. In a few cases we used a cubic spline for the sky interpolation, when it was clear that this procedure was giving better results than the linear fit. All the PMT data were then corrected for extinction. Afterwards, they were used to examine possible transparency variations. In a few cases of high sky instability, the count ratio between channel 1 and channel 2 was used instead of channel 1 counts only. Some smoothing of the channel 2 data was applied when possible. Systematic long time scale trends ( 2 hours), probably due to tube drifts and/or to residual extinction, were finally compensated by means of linear or cubic spline interpolation.

For the Calar Alto CCD data, 10 comparison stars were selected, after having been tested for photometric constancy. Their average magnitude was subtracted from the HS 2324 measurements on a point by point basis. Differential extinction was corrected by means of a cubic spline.

Finally all the single data sets (PMT + CCD) were set to a mean value of zero. The times of all data were then converted to Barycentric Julian Date using the algorithm of Stumpff (1980). The accuracy of the original times was of the order of 0.3 s (Beijing), 0.01 s (Loiano), 0.2 s (McDonald) and 1 s (Calar Alto). For Calar Alto 1 s is also the time accuracy of each measurement. Moreover, to have a more homogeneous data set, we have binned the PMT data to an effective integration time of 90 s. The value of 90 s has been chosen because it corresponds to the mean distance between consecutive CCD observations. When more than one site was active at the same time, in the overlap regions, we applied a weighted average of the data obtained at the different sites. In this way even lower quality data can be used to improve the S/N ratio (see Moskalik 1993). In its final form, the data set is constituted by the time of each integration, the fractional departure of the count rate from the mean (modulation intensity), and the error.

© European Southern Observatory (ESO) 1999

Online publication: February 23, 1999
helpdesk.link@springer.de