Sloan Digital Sky Survey Telescope Technical Note 19950517-01
Astronomers have traditionally monitored cloud cover visually. However, this technique is difficult under dark conditions, insensitive to thin cirrus, and impossible for remote telescope use. A sensitive camera operating in the thermal infrared escapes these problems. Unfortunately, commercially available cameras have smaller than desired fields of view, are not intended for continuous operation, and are expensive. Consequently, a single channel all-sky scanner has been constructed to continuously monitor observatory cloud cover and motions ("Sloan Digital Sky Survey cloud scanner", C. L. Hull, S. Limmongkol and W. A. Siegmund, Proc. of S.P.I.E., 2199, 1994). Operating at a wavelength of 10 µm, it produces a 135°x135° image with a resolution of 0.9°. These data are suitable for returning to remote observers and for archiving with astronomical data. The detector is a single channel HgCdTe photoconductive liquid-nitrogen cooled detector. It is scanned by two flat mirrors on an altitude-altitude mount driven by stepping motors. An off-axis parabolic mirror images the sky onto the detector. The beam is mechanically chopped at 400 Hz and the signal is synchronously sampled by an analogue to digital converter.
On the afternoon of May 14, 1995, these four images were obtained between 1532 and 1544 Mountain Daylight Time (MDT). They were taken 4 minutes apart (in order from left to right and top to bottom). The infrared (IR) images were stretched to emphasize cloud details. The warmest features are bright yellow and the coldest are dark brown. Each image is 135°x135° centered on the zenith with north at the top. The images extend to within about 20° of the horizon and include utility power poles and wires, trees and the roof of a nearby building. The clouds move from the west-southwest. The sun is the brightest feature just above the center on the right. Cloud and apparent solar diurnal motions are seen easily in the movie. The clouds scatter IR radiation from the sun enlarging the solar image. This effect is particularly apparent in the second image. At the same time as the IR images, visible light images were taken with an Apple Computer QuickTake electronic camera. The IR images were stretched to emphasize cloud details, then cropped and scaled to approximately match the visible images. The visible images were not taken at the same instant as the IR image or from exactly the same location. This and the difference in the physics of the radiative and scattering properties of the clouds in the two wavelength regions account for differences in the visible-light/IR pairs. The cropped IR images took nearly one minute to scan. This finite duration could cause shearing of the image of a moving cloud although this is not apparent in these images. Line scanning was left-right for even lines and right-left for odd lines. Frame scanning was bottom to top. A color image in visible light (left) from 1532 MDT is compared with an infrared image taken about the same time (right). The utility power pole provides a familiar reference object. At 1536 MDT, a clear patch at top left is evident in both images. At 1540 MDT, the large central cloud is visible in both images. At 1544 MDT, a distinctive cloud is apparent near the center in the visible image and right of center in the IR.
The performance of the SDSS cloud scanner is excellent. Thin cirrus clouds are detected readily. The low intensity of solar radiation in the thermal infrared allows day and night operation of the scanner. Visible and IR images can be compared easily and common features identified. This suggests that IR images obtained on dark nights (when it is not possible to obtain visible images) will be straightforward to interpret. The registration of fixed objects on successive images is excellent (see movie). It should be possible to accurately determine the location of cloud features. This will be important in linking a specific cloud feature to extinction measured in a particular direction with a nearby telescope.
