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

A Search for Variable Stars based on the HST-Fine Guidance Sensor Photometry

R. Kuschnig, W.W. Weiss and R. Bahr
Inst. f. Astronomy Univ. Vienna, A-1180 Vienna, Austria



A time series analysis is applied to the photometric data obtained by the HST Fine Guidance Sensors (FGS), in order to discover new variable stars and to derive information about microvariability of a large sample of objects. We briefly describe the photometric properties of the FGSs and present a typical result.

Keywords: stars, variable stars


The HST is equipped with three Fine Guidance Sensors (FGS) which serve two principal functions. First, they provide a pointing reference to the telescope Pointing Control System (PCS). Second, they can be used for astrometry. FGS3 is presently allocated to this task. The technical concept and layout of the FGSs is described in detail by Bradley et al. (1991). The light of a Guide Star (GS) is split into four channels. The intensities are measured by photomultipliers (PMTs) with an integration time of 25 msec. Usually, two of three FGS instruments are used simultaneously to guide the HST.

Our project is based on a time series analysis of all data sets recorded by the FGSs in order to detect variability among the used Guide Stars. Since the start of the HST operation in November 1990, more than 6000 individual stars selected from the Guide Star Catalog (GSC) have been observed.

The brightness of Guide Stars ranges from 8.5 mag to 15 mag in Johnson-V. No further information (spectral type, temperature, luminosity, etc.) is listed in this catalog. Therefore, a cross-check with astronomical archives, based on the GSC coordinates, is necessary to find an astrophysical description of the object, which is successful only in about 5% of all cases. The Guide Star distribution across the sky is obviously determined by the scientific programs of the HST, but on a large scale it is homogeneous.

In our program, we check a huge sample of field stars for variability with high accuracy. As we will show, the HST-FGS photometry is only limited by photon noise, but regular data gaps due to the earth limb occultation of the telescope deteriorates the window function. It is planned to perform a consistent analysis of all available FGS data during the lifetime of the HST.

FGS Data Format

The FGS data are presently extracted from the Astrometry and Engineering Data Processing System (AEDP) by the so called Observatory Monitoring System (OMS). This software package produces the two groups of files which are essential for our analysis. The FGS-SAVE files consist of the four PMT count rates, including the start time of each integration. The OMS-SUMMARY files contain in chronological sequence the FGS operation data, the Guide Star numbers and some important spacecraft parameters. Combining the contemporaneous information of both files defines which FGS unit has acquired which Guide Star during a given observing session, as well as the orientation of the HST relative to earth limb, moon and sun. Due to the limited data transfer rate from the spacecraft, either one or 20 of the 40 actual measurements per second are stored and written to the FGS-SAVE files. One datapoint per second is transmitted in the telemetric ``A format'', and 20 datapoints in the ``F format''. Up to fall 1994 the A format was predominantly used, but later the F format was chosen for most of the guiding sessions.

Photometric Properties


: Histogram of mean amplitudes derived from 1000 synthetic white noise data sets with identical parameters, but different random seed.

In order to analyze the photometric performance of the FGSs, we took a sample of 1100 stars which showed no indication for variability in the frequency range from 0 to 600 c/d. This group of suspected `constant' stars was observed with all three FGS's, but the FGS2 instrument was used by a factor of two less frequently than the other two units. With no disturbing turbulent atmosphere, the photometric noise should follow a Poisson distribution. If true, a direct relation between the standard deviation of all (ndat) measurements and the square root of the mean intensity (mint) should exist for constant stars. And, indeed, for faint Guide Stars this relation is fulfilled. For high intensities, a deviation is caused by dead-time effects. Furthermore, we found that FGS1 is more sensitive by a factor of 1.2, and FGS3 even by a factor of 1.5, relative to FGS2. Dead time constants and the sensitivity of all three FGSs which were determined by us are in good agreement with STScI internal documents, and will be monitored over the life time of the HST.


: The light curve of GS 0825203215 obtained with FGS2

Independently, we checked the amplitude spectra of our sample of `constant' Guide Stars in order to confirm the white noise characteristic predicted by photon statistics. More than 150 cases with comparable frequency resolution were used for this study.

From these investigations we derived a `noise model' for the FGS photometers, which is defined as the relation of the mean amplitude (mamp) over the full frequency range with the total number of registered photons. To study the influence of parameters, like duty cycle and time base, on the noise level in the amplitude spectrum, we computed synthetic white noise data. Plotting the mean amplitudes obtained for 1000 individual synthetic data sets (with all parameters kept constant, except the seed-value for the random noise generator) results in distributions of which one is presented in Fig.1 with a typical set of parameters (ndat, mint, time base, duty cycle). Based on such diagrams, it is possible to derive a confidence level for variability of actual HST Guide Star observations.


Presently we have a list of about 250 suspected variable Guide Stars, which is about 15% of the targets analyzed up to now. The verification for at least few typical cases by ground-based observations is important for an independent confirmation of the variability found by the FGS photometry. Based on ESO 1.5m Danish DFOSC photometry, we have verified variability for three Guide Stars.

Fig.2 shows a light curve of the Guide Star 0825203215 (mv=9.28, F8, HD116979). From this figure, upper panel, it can be seen that the guiding session covered four orbits of the HST in April 1995, and again in May 1995 (lower panel). A well defined light curve is visible for most of the orbits except for the 3 orbits on April 4th, where for telemetry the A-format was used.

Our short term goal is to publish the results as a catalog of variable Guide Stars and which will be periodically supplemented. Furthermore, an observing program will be initiated to determine the spectral type of all interesting objects.

Based on a sufficiently large sample of Guide Stars, we will be able to investigate the pulsational stability of stars which are distributed over a wide parameter range in the HR-diagram. This study should contribute interesting information to the theory of stellar stability.


The project is supported by the Digital Equipment Corp. (European External Research Program, project STARPULS) with computer hardware, and funded by the Austrian Academy of Science (project FGS-Präzisionsphotometrie). P. Bely and H. Jenkner from the STScI provided access to the FGS archive and contributed with many valuable comments and suggestions.


Bradley, A., Abramowicz-Reed, L., Story, D., Benedict, G., & Jefferys, W. 1991, PASP, 103, 317

Eaton, D.J., Whittlesey, A.R., Allen, B.W., Stoll, R., Abramowicz-Reed, L., & Margulies, M. 1993, Applied Optics, 32, 10