Sloan Digital Sky Survey Telescope Technical Note 19910801
It has long been apparent that an all-sky picture of clouds over an observatory site and available in near real-time would be a great benefit to personnel using a telescope. The acquisition of such a device was rated very high by the Apache Point Observatory Science Advisory Committee a couple years ago, indeed higher than some astronomical instruments. For remote observing, such a cloud imager is almost essential. Our experience is, on nights with partial cloud cover, that most of the requests of the night assistant by the remote observer are for information on cloud distribution. Such interaction is a poor substitute for a cloud picture sent to the remote observer and is a poor use of the night assistant. Finally, protection of the telescope from precipitation can be better assured with continuously updated cloud images. This is important for both on-site and remote observers since a rain cloud can develop quickly in a direction away from the direction the telescope is pointed.
We have every reason to believe that clouds can be detected very simply and easily in the 10-12 micron wavelength window. The same factors which make low water vapor of crucial importance for IR astronomy, make this band useful for cloud imaging. We have anecdotal evidence from J.T. Williams of MMTO and David Westfall, formerly of CFHT and now of New Mexico Institute of Mining, that video-rate 10 micron cameras are very effective at imaging clouds. Fred Forbes and his colleagues of NOAO report that a water vapor monitor using an uncooled pyroelectric detector is very sensitive to clouds.
The video-rate IR cameras used by Williams and Westfall, are cryogenically cooled and not designed for continuous operation. Instead, we consider a single channel uncooled pyroelectric detector scanned using a mirror on an gimbal driven by stepper motors. This scanning scheme will generate scan lines much like lines of constant longitude on a globe. The equator, in this pattern, will be the great circle running east to west across the sky and through the zenith. We suggest this pattern since it can be displayed directly without remapping or written to a FITS format file, albeit with a little distortion. The scanner covers the sky down to two air masses except for a small portion in the direction of the motor driving the outer gimbal. The off-axis parabolic mirror, that images the sky on the detector, is mounted 15° above the gimbal axis. This enhances sky coverage in the direction of the motor driving the outer gimbal, but will cause further distortion of the image if the scanning of successive lines is not modified slightly.
The optical components of the system are listed below. An off-axis gold coated parabolic mirror images the sky onto the detector. A 10 micron filter above the detector determines the bandpass of the system (the detector sensitivity is not wavelength dependent). The focal length of the mirror is 25 mm. The detector is 0.5 mm in diameter. Thus, the field of view is 0.020 radians. To critically sample, we should sample every 0.010 radians. This gives about 200 samples across the 2 radians we wish to scan. Sampling at 200 Hz, a scan line will take 1 second and a frame will take 200 seconds or 3.3 minutes. Rather than scan in a TV raster pattern, we will scan successive lines in opposite directions to eliminate the deadtime during retrace.
For the detector, the noise equivalent power, i.e., the power incident on the detector needed to provide a signal to noise ratio of one, is
where A is the detector area and D* is the sensitivity parameter specified by device manufacturers.
We plan to sample at 200 Hz, so to avoid aliasing we need a low-pass filter with a cut-off at 100 Hz. Over this bandwidth, , the noise power is
Our goal (suggested by Fred Gillett) is to detect 1% of an IR opaque cloud at 0°C. The flux from a black body is
Evaluating this expression for 0°C and 10 microns gives 6.38E-12 W/m^2 Hz and 1% of this is 6.38E-14 W/m^2 Hz.
The power incident on the detector from such a cloud which fills the field of view is
where T is the throughput of the optical system, d is the diameter of the beam, is the solid angle viewed by the detector, and is the bandpass of the IR filter. The solid angle viewed by the detector is
where f is the focal ratio of the parabolic mirror.
Combining the above expressions we find that the signal to noise ratio is
Evaluating this expression for 3.8E12 Hz passband of the IR filter, the 0.77 throughput of the optical system (neglecting the detector window), and evaluating the rest of the parameters from the values given below, we find a SNR of 4.1.
Component Manufacturer Comments Detector EDO Corp. 300-2 Ø0.5 mm, D*=4E8 cm-Hz^1/2/W Filter OCLI W10672-9 Ø25 mm x 1 mm, 82% transmission, 10-11.4 µm Mirror Janos A8037-116 F.L. = 25 mm, Ø25 mm, 97% reflectance Mirror Edmund R32,088 38x54x9.5 mm elliptical, 97% reflectance
The 10 micron window has significant opacity, about 14% with 1 cm precipitable water. Since we plan to scan to about 2 air masses, the edge of the picture will be about twice as bright as the center. We may wish to subtract a function which compensates for this effect to make it easier to see clouds.
We do not intend to chop the beam against a reference. Instead, we will be looking for features in the cloud cover. This camera will not be sensitive to uniform cloud cover. However, this is a cloud pattern that the proposed Digital Sky Survey monitor telescope with its low spatial coverage, is sensitive to, and thus compliments the capability of the IR scanner.
The brightest astrophysical sources (outside the solar system) are about 1E21 W/m^2 Hz. This is far below the sensitivity of this system. This is one great advantage of imaging clouds at this wavelength, i.e., the interpretation of cloud images is not confused by background sources.
1. Forbes, F., Morse, D.A., Poczulp, G.A., Site Survey Instrumentation for the National New Technology Telescope (NNTT), Optical Engineering, 27, no. 10, p. 845.
