Testing and Data Reduction of the Chinese Small Telescope Array (CSTAR) for Dome A, Antarctica
ABSTRACT The Chinese Small Telescope ARray (hereinafter CSTAR) is the first Chinese astronomical instrument on the Antarctic ice cap. The low temperature and low pressure testing of the data acquisition system was carried out in a laboratory refrigerator and on the 4500m Pamirs high plateau, respectively. The results from the final four nights of test observations demonstrated that CSTAR was ready for operation at Dome A, Antarctica. In this paper we present a description of CSTAR and the performance derived from the test observations. Comment: Accepted Research in Astronomy and Astrophysics (RAA) 1 Latex file and 20 figures
arXiv:1001.4935v1 [astro-ph.IM] 27 Jan 2010
Research in Astron. Astrophys. 2009 Vol. 9 No. XX, 000–000
Testing and Data Reduction of the Chinese Small Telescope Array
(CSTAR ) for Dome A, Antarctica
Xu ZHOU1,4, Zhenyu WU1, Zhaoji JIANG1,4, Xiangqun CUI2,4, Longlong FENG3,4, Xuefei
Gong2,4, Jingyao HU1,4,Qisheng LI1, Genrong LIU2, Jun MA1, Jiali WANG1,4, Lifan
WANG3,4, Jianghua WU1, Lirong XIA2, Jun YAN1,4, Xiangyan YUAN2,4, Fengxiang
ZHAI2, Ru ZHANG2, Zhenxi ZHU3,4
1National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China;
2National Astronomical Observatories/Nanjing Institute of Astronomical Optics & Technology
3Purple Mountain Observatory
4Chinese Center for Antarctic Astronomy
Received [year] [month] [day]; accepted [year] [month] [day]
Abstract The Chinese Small Telescope ARray (hereinafter CSTAR) is the first Chinese
astronomical instrument on the Antarctic ice cap. The low temperature and low pressure
testing of the data acquisition system was carried out in a laboratory refrigerator and on
the 4500m Pamirs high plateau, respectively. The results from the final four nights of test
observations demonstrated that CSTAR was ready for operation at Dome A, Antarctica.
In this paper we present a description of CSTAR and the performance derived from the
Key words: instrumentation: detectors — techniques: photometric — stars: variables
Site testing at the South Pole (90◦south, 2835m elevation) and Dome C (123◦east, 75◦south, 3260m
elevation) over the past decade has shown that the Antarctic plateau offers outstanding sites for as-
tronomical observations. The extremely cold temperatures lead to very low infrared backgrounds and
atmospheric water vapor content. The very low wind speeds and stable middle and upper atmosphere
result in favorable seeing conditions for high-resolution imaging (Storey et al. 2007). The median free-
atmosphere seeing in Dome C is 0.27 arcsec, and it is below 0.15 arcsec for 25 per cent of the time
(Lawrence et al.2004). In addition, the long dark winter on the Antarctic plateau allows continuous ob-
servations of variable astronomical objects.
Dome A (77◦21′east, 80◦22′south, 4093m elevation), the highest point on the Antarctic plateau,
is widely predicted to be an even better astronomical site even than Dome C, based on the topographic
similarityandDomeA’s higheraltitude.InJanuary2005,via overlandtraverse,DomeA was first visited
by the Polar Research Institute of China (hereinafter PRIC). This provides astronomers with a good
opportunity to explore this special area for astronomy. PRIC plans to establish a permanently manned
station at Dome A within the next decade, with astronomy as one of the scientific goals of the station.
As part of this program, PRIC conducted a second expedition to Dome A, arriving via overlandtraverse
in January 2008. On this expedition, the first Chinese Antarctic astronomical instrument, CSTAR, was
deployedto DomeA.Beside thetask oftheastronomicalsite testing,themainscientific goalsofCSTAR
2X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.
include variable star light curves and statistics, supernovae studies, gamma-ray burst optical afterglow
detection and exoplanet detection.
CSTAR was designed and constructed during 2006 – 2007 at the National Astronomical
Observatories of China (NAOC) and the Nanjing Institute of Astronomical Optics Technology of China
(NIAOT). A series of tests were performedon CSTAR beforethe second expeditionto Dome A to enure
that it was ready for deployment. As a result of this careful preparation, CSTAR operated successfully
during 2008, as part of the Plateau Observatory (PLATO) at Dome A (Yang et al. 2009). In Section 2,
we describe the design and constructionof CSTAR . Test observationsat the Xinglong station of NAOC
and the data reduction are presented in Section 3. Finally, a summary is given in Section 4.
CSTAR is a small 2×2 Schmidt-Cassegrain telescope array. Each telescope of CSTAR has an entrance
pupil diameter of 145mm (effective aperture of 100mm) and a focal ratio of f/1.2, giving a field of
view of ∼ 4.5◦× 4.5◦. Fig.1 shows the optical design of the CSTAR telescope, which consists of a
catadioptricobjectivewith sphericalprimarymirror,deliveringlow chromaticaberration.Thefirst plano
lens serves both as a window and as a filter. In order to keep the focus unchanged through a ∼ 100◦C
temperature difference (from 20 to −80◦C), Zerodur and fused silica are used for the main optical
components and Invar 36 is used for the telescope tube. The tube is designed to be light weight, well
sealed, and easy to assemble. The inside of the telescope tube was filled by pure nitrogen to avoid ice
and frost formation on the internal optical surfaces. Each telescope tube is hermetically sealed and an
ITO (Indium-Tin-Oxide) film was coated onto the front window. An electric current is passed through
this film, providing ∼ 10 W of power to keep the surface of the window warmer than ambient. CSTAR
is specifically designed for Antarctic operation, having no moving parts at all—including the optics and
mechanical supporting system. The four telescopes are installed in a steel enclosure, see Fig.2, and are
pointed at the South Celestial Pole; ie., each telescope is inclined 9◦38′from the zenith. Details of the
CSTAR telescope structure are described in Yuan et al. (2008).
The three telescopes CSTAR #2, #4, and #1 have fixed filters: g, r, and i, the fourth telescope
CSTAR #3 is filter-less. The main parameters of the three filters are listed in Table 1 and the transmis-
sion curves of those filters are presented in Fig.3. The filters are designed to be similar to the corre-
sponding filters of the SDSS (Fukugita et al. 1996). Using these filters, CSTAR can obtain multicolor
photometric data for each object simultaneously.
An Andor DV435 1K×1K frame transfer CCD with a pixel size of 13 µm is used for the detector.
Frame transfer technology is ideal for fast imaging as it has the advantage of requiring no mechanical
shutter. Avoiding the need for moving parts is very desirable on the Antarctic plateau. The CCD was
enclosed in a control box, as shown in Fig.4. The cable at the back of the box connects to the PCI
controller card installed in the control computer. The typical readout noise of the CCD is ∼ 3e with
maximum of ∼ 12e, and the gain is set to 2.0 e per A/D. The peak quantum efficiency of the Andor
CCD at −90◦C is ∼ 95%. During the typical exposure time of 30s and under the typical ambient
temperatures of less than −50◦C on the Antarctic plateau, the dark current of the Andor CCD is only
0.5e. The dark current can thus be negligible under Antarctic conditions.
Each Andor CCD is controlled through the CCI-010 PCI controller card installed in an industrial
control computer for each telescope. The control computeris composed of a 1TX-i7415VLmain board,
Intel Centrino 1.6 GHz CPU, and 1 GB of memory. Two kinds of storage disks are used for the control
computer. One is a 4 GB CompactFlash (CF) disk which can operate at low temperatures (down to
nearly −45◦C); the other is a normal 750 GB IDE hard disk. Fig.5 shows the four control computers.
Eachcomputerweighs8.3kg.TheWindowsoperatingsystem is installedontotheCF diskbecauseofits
greater reliability under low temperature conditions, while the 750 GB hard disk is mainly used as data
storage. The CCD control and data collection software were developed based on the Andor-SDK-CCD
software development kit for the Windows-XP operation system. The time of the controler computer of
CSTAR #3 is synchronized by GPS and the other computer correct its clock by CSTAR #3.
CSTAR : testing and data reduction3
The real time data reduction process start automatically after the controller computer booting. The
image coorected by bias and flat-field frames, and the catalogues objects is produced. The brightest
3000 stars of the catalogue from 1/3 images is movedto a special directory for data transfere via iridium
3 TESTING AND DATA REDUCTION
In order to assure the performance of CSTAR under the extremely low temperature conditions of Dome
A, the CCD system and several different industrial control computers were tested. Finally, the whole
CSTAR system was tested at low temperature in the laboratory of NAOC. These tests indicated that
the four telescopes and the CCD can work at low temperatures down to nearly −80◦C, while the four
control computers can work down to −30◦C. In 2007 February 6 – 9, the CCD and control computers
were tested at Kalasu. Kalasu (see Fig4) is located in the Tajik Autonomous County of Taxkorgan,
on the Xinjiang Pamirs of China at an elevation of 4450 m. We chose Kalasu as the test site because
of its low temperature and low atmospheric pressure conditions similar to the Antarctic plateau. The
atmospheric pressure was ∼ 58.6 kPa and the temperatures ranged from −5◦C to −18◦C during the
testing process. Both the CCD and the control computers were shown to work normally during the two
days of testing, and there are 4 750GB normal hard disks were selected as data storage of CSTAR.
In 2007 September 3 – 7, test observations of CSTAR were performed at the Xinglong station
of NAOC. The four CCDs were cooled down to −40 – −50◦C by electronic cooling system of the
camera. The weather was goodin most of the time duringfour observationnights, and more than 20,000
images were obtained. The typical exposure time was 20s. Fig6 shows the ‘super’ bias images for each
telescope, which are the median of 100 bias frame images for each telescope. There is no obvious
variation and structure in the ‘super’ bias images. These ‘super’ bias images are unique bias frames to
be used for reduction of data both from observations at Xinglong and also from Dome A.
Variations of night-sky background are obvious even in the zenith direction. If one takes the time
during a photometric, moonless night to obtain a long series of sky-dominated images pointing directly
at the zenith,the effects of the nonuniformityof the night sky can be minimized.However,our telescope
observes the polar sky area at an airmass of 1.54 at Xinglong station. The median sky background can
only be used as an initial flat-field for image correction. Thus, we typically obtained ‘supersky’ flat-
fields by combining images of the sky (Zhou et al. 2004). During this combination, the bright stars in
the images were masked and rejected, and only the areas free from stars were used. By comparing
the images, the median level of each pixel could be selected to derive the final ‘supersky’ flat-field.
100 images of ‘supersky’ for each telescope of CSTAR were used to obtain the ‘supersky’ flat-field.
These flat-fields mostly reflect the small, pixel to pixel variations in the images. Fig.7 shows the final
‘supersky’ flat-field images for each telescope. Some obvious structures can still be seen.
3.2 Data reduction
First, for each filter a ‘super’ bias frame was subtracted from each image, then the ‘supersky’ flat-field
was divided by the bias-corrected images. The bias and flat-field corrected data of ∼ 20000 images
obtained by CSTAR during the four test-observation nights were processed with the automatic data
reduction software developed by Z. J. JIANG and X. ZHOU based on the DAOPHOT photometric
package(Stetson 1987),whichwasusedinthedatareductionofBATC(Fan et al. 1996;Wu et al. 2007).
Because CSTAR has a large field and is undersampled, obtaining an accurate point-spread function
(PSF) for the sources detected across the whole view of field is very difficult. The DAOFIND program
was used to find stars in each image and DAOPHOT was used to performsynthetic aperture photometry
on the objects detected by DAOFIND. All instrumental magnitudes of the four telescopes were then
normalized to the V band magnitudes of stars in the image 39530013.fit, which was observed by #3
telescope on 2007 September 5.
4X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.
3.3 Error analysis and correction
There are obvious systematic errors in the derived aperture-photometry magnitudes. The errors mainly
come from following sources:
1. The bias stability of each CCD
Due to the continuous observation during exposures and the frame transfer mode of the CCD, there
is no opportunity to obtain real-time bias frames. The bias frames obtained at one time must be
used for observations from another day at Dome A. Because of variations in the environmental
parameters, such as temperature and instrumental status, the bias of each CCD camera may change.
This variable bias will introduce linear errors in the observed magnitudes.
2. Non-uniformityof the ‘supersky’ flat-field.
The flat-field images were not obtained during ideal photometric nights, and not from the zenith
sky. A brightness gradient and asymmetry may exist in the flat-field frames. The variation in tem-
perature from −40 to −80◦C may also change the characteristics of the flat field. During the polar
observations by the fixed CSTAR telescopes, every star will trace out a circle on the CCD, and the
residual flat-field error will give a false variation in the observed magnitude of each star.
3. Variable PSF for stars in different positions in the images of CSTAR .
The telescopes of CSTAR have a large field of view. The optical design cannotkeep the PSF exactly
uniform over all parts of the image. When we use a fixed aperture to measure the magnitudes of the
stars,thePSF dependsonthelocationontheimageandthiswill causeavariationintheinstrumental
magnitudes of each star relate to the other stars.
Because we are observing a single area of the sky, and the sky’s image is rotating on the CCD, we
have the opportunity to correct the main residual system errors mentioned above. Using thousands of
stars with very different magnitudes, we can easily determine the variable component of the bias resid-
uals based on the different magnitudes of those stars in two different images. Using all of the circular
traces of the stars, the large-scale residual flat-field correction can be obtained. Using the instrumental
magnitudes from several different apertures for each star, the aperture photometry curve-of-growth can
be obtained in all parts of the image but is mainly correctedwith residual flat-field correction mentioned
above. The instrumental magnitude obtained from different aperture were calibrated to the standard
system. After all these corrections, the systematic errors in the derived photometric magnitudes can
be reduced to the level of 0.01 mag for the brightest stars in most of the images. Some sudden ab-
normal variations, where they exist, mostly come from the cirrus clouds in the sky. Fig.8 presents the
magnitude-corrected flat-field images for each telescope using thousands of stars., and shows the obvi-
ous circular structures that match the traces of stars on the CCD.
Two kinds of error estimates have been performed. One is theoretic statistical estimation based
on star’s magnitude and its sky background. The other is obtained by real repeated observation of all
the objects in the images. By comparing the errors resulting from different images of the same field
with the same filter, we find that the measurement errors are normally ±0.01 mag for bright stars. The
statistical errors can be regarded as the lower limits of the measurement errors. In the error estimates,
we ignore points with abnormally large deviations to calculate the root mean square (rms) errors. The
abnormal variation may come from the true star brightness variation, or defect of the image (cosmic
ray, satellites, bad pixels, etc). Fig. 9 shows the photometric errors for each telescope of CSTAR at
different magnitudes. Because the errors estimated by this method include real statistical errors and
residual system errors from the bias and flat-field correction, the errors shown in Fig.9 should be larger
than the actual observational errors. Fig.9 also shows that the efficiency of telescope #2 of CSTAR is
very low and that the limiting magnitude of this telescope is about 2 mag lower than that of the other
three telescopes. We knew that the CCD camera for CSTAR #2 was much noisier than the others, but
we were unable to change it in the time available.
As an example of the data obtained, the light curve of one of the bright stars from the four CSTAR
telescopes are shown in Fig.11. The main scientific objectives of CSTAR are to assess the site quality
of Dome A and to study the variable objects in the region of the South Pole. Fig.10 shows the image
CSTAR : testing and data reduction5
Table 1 Passband parameters of filters used by CSTAR .
CSTAR #2CSTAR #4CSTAR #1CSTAR #3
effective Wavelength (nm)
39530013.fit obtained by CSTAR during the test observations at the Xinglong station of NAOC. The
variable stars detected by CSTAR are labeled by green circles. The light curves of those variable stars
are presented in Fig.12.
CSTAR, China’s first Antarctic astronomicalinstrumentis described.CSTAR is composedof foursmall
Schmidt-Cassegrain telescopes. Each telescope has an effective aperture of 100 mm and a field of view
of ∼ 4.5◦× 4.5◦. Three of the four telescopes are equipped with g, r, i filters, the fourth one is filter-
less. A frame-transferAndorDV4351K×1K CCD is used as the detectoron each telescope. A specially
designed control computer for each telescope is used for data acquisition and data reduction.
Low-temperature laboratory testing demonstrates that the telescopes and the CCD can work under
extremely low temperature (down to nearly −80◦C), while the control computer can work at temper-
atures as low as −30◦C. Actual test observations at Kalasu in the Xinjiang Pamirs indicated that the
CCD and control computer can work at these low temperatures and under low atmospheric pressure
‘Super’ bias and ‘supersky’ flat-field images were obtained during the test observations at the
Xinglong station of NAOC. These test observations and the subsequent data reduction indicate that
CSTAR can work stably and obtain a large volume of scientific data. A special data reduction method
was used to reduce the observational errors for each of the objects detected by CSTAR . The data reduc-
tion process is done automatically in real time, and catalogue of brightest star from 1/3 of the images
obtained are prepared for further data transfer via iridium satellite communication. Finally, Eight vari-
able stars were detected by CSTAR during the test observations.
Acknowledgements This work was supported by the Chinese National Natural Science Foundation
grands No. 10873016, 10633020, 10603006, and 10803007, and by National Basic Research Program
of China (973 Program), No. 2007CB815403. We thank our colleagues at the University of New South
Wales, Australia, for assistance in editing this paper.
Lawrence, J. S. Ashley, M. C., Tokovinin, A., & Travouillon, T. 2004, Nature, 431, 278
Fan, X.H., et al. 1996, AJ, 112, 628
Fukugita, M., et al. 1996, AJ, 111, 1748
Stetson, P.B. 1987, PASP, 99, 191
Storey, J. W. V., Lawrence, J. S., & Ashley, C. B. 2007, RevMexAA, 31, 25
Wu, Z. Y., et al. 2007, AJ, 133, 2061
Zhou, X., et al. 2004, AJ, 127, 3642
Yang, H.G., et al. 2009, PASP, 121,174
Yuan, X.Y., et al. 2008, SPIE, 7012,152
6X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.
Fig.1 Optical design of CSTAR telescope.
CSTAR : testing and data reduction7
Fig.2 A picture of CSTAR enclosure was taken in XingLong station of NAOC.
8X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.
Fig.3 Transmission profiles of the 3 CSTAR filters. The filter codes (see Table1 are labeled
on each filter. Note that CSTAR #3 has no filter.
CSTAR : testing and data reduction9
Fig.4 The Andor CCD enclosed in its control box. This picture was taken at Kalasu in the
Tajik Autonomous County of Taxkorgan,Xinjiang Pamirs of China.
10X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.
Fig.5 Computer control equipment.
Fig.6 Bias frame images for each telescope.
CSTAR : testing and data reduction11
Fig.7 Flat-field images for each telescope.
Fig.8 Images of the corrected flat field for each telescope.
12X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.
Fig.9 Photometric errors for each telescope of CSTAR . The vertical scale is in magnitudes.
CSTAR : testing and data reduction13
Fig.10 The image 39530013.fitobtained by CSTAR . The variable stars detected by CSTAR
are labelled by green circles in the image.
Fig.11 The light curve of one of the bright sources from the CSTAR telescopes.
14X. ZHOU, Z. Y. Wu, Z. J. Jiang, et al.