|
|
|
|
|
|
Primary Quantities Measured with System Typical |
|
|
Section 1: Hardware
1.1) Why does the WSI need to use an imager?
1.2) How did you decide to use the 16-bit CCD, as opposed to using film, video camera, or intensified CCD?
1.3) What is the purpose of the occultor?
1.4) Why is the occultor so big?
1.5) Why do you need the occultor under moonlight?
Section 2: Observed Images
2.1) The image appears round. What is the user looking at?
2.2) Why do you use 16 bits, if you can only display 8 bits?
2.3) Can you see the whole sky?
2.4) What can the WSI see under starlight? Is it really visible and not IR?
2.5) How does the sensor adjust to the changing light conditions?
Section 3: Cloud Decision Images
3.1) How is cloud cover computed?
Section 1: Hardware
1.1) Why does the WSI need to use an imager?
Originally, for acquiring radiance distributions, we used scanning photometers. Whereas they provided excellent radiance distributions, the more demanding task of high quality cloud detection requires more simultaneity in the data acquisition. The imager acquires measurements in all directions at once, so cloud motion presents little problem. By using a 512 by 512 imager, we essentially have nearly 250,000 calibrated radiometers staring at the sky simultaneously.
1.2) How did you decide to use the 16-bit CCD, as opposed to using film, a vidicon, video camera, or intensified CCD?
We used film fisheye systems in the 50's, 60's and 70's, however they cannot be processed quickly, and the radiometric content of the data is severely degraded or lost. For measuring daylight skies only, we moved to a CID-based video camera in the 80's. This had the advantage over vidicons that calibration and geometric positioning accuracy are maintained. (This is not true of all video cameras however; it is very important that they have fixed gain, and electronic characteristics such that they can be well calibrated.) In moving to the Day/Night application, we determined that the video cameras we evaluated did not have the sensitivity we required. Likewise, intensified CID systems did not have sufficient stability and noise control for our application. We have been very happy with the performance of the slow-scan low noise 16-bit CCDs we are using.
1.3) What is the purpose of the occultor?
The occultor is used to shade the lens and dome, so we don't get significant stray light in the image. Most photographers have at one time taken pictures looking toward the sun and seen the resulting artifacts in the image; the occultor prevents these artifacts by shading the lens.
1.4) Why is the occultor so big?
It is not sufficient to cover the Ì solar disk. It is necessary to shade the full physical extent of the lens and the dome. Since the solar rays are parallel, the shade must be at least as big as the dome, which is 7". In order to avoid obscuring too large a solid angle, we place it 24" from the dome. The actual solid angle obscured by the shade is less than 1% of the sky dome. At the remote sites, we will be using a larger arc shade instead of the troley. This shade is fixed in the N-S direction and moves only in the E-W direction; it will obscure 6% of the sky for sites near the equator, and 14% of the sky for sites near the poles, but should be more reliable for rigorous environments or sites without technical personnel on-site.
1.5) Why do you need the occultor under moonlight?
The flux control is set to bring the radiance of the moonlit sky and clouds on scale. Under moonlight, the moon is as bright relative to the sky (in the absence of urban lights) as the sun is relative to the daylight sky. Thus it is equally important to shade the lens under these conditions.
Section 2: Observed Images
2.1) The image appears round. What is the user looking at?
The center of the image is the zenith (overhead), and the edge of the circular image is the horizon. The view cannot be the same as looking down at a map, because the WSI is looking up. If you imagine lying on the ground looking up, with your toes to the north, you can imagine the view of the WSI: E is right, W is left, S is top, and N is bottom.
2.2) Why do you use 16-bits, if you can only display 8-bits?
An 8-bit image has only 256 grey levels, which means it has a limited radiometric resolution and/or range. By saving and using the original 16-bit data, data over a larger range and with a finer resolution are available to the numerical processing.
2.3) Can you see the whole sky?
Yes, because we underfill (rather than overfill) the chip so that the full optical image falls on the chip.
2.4) What can the WSI see under starlight? Is it really visible and not IR?
We do use visible light at night, although currently we open up the passband to roughly 400 - 900 nm. We can see the milky way, stars such as the stars in Orion's belt, clouds that are not visible to the human at night, and the light of distant cities. Often when the night visually appears clear, there are cirrus layers easily detectable by the sensor.
2.5) How does the sensor adjust to the changing light conditions?
The flux control algorithm embedded in the control software computes the solar zenith angle and the lunar zenith angle, phase angle, earth-to-moon distance, and resulting relative brightness. Using these values, the desired combination of filter and exposure selection is chosen. The imager has sufficient dynamic range within each image, that the system need not correct for variations in lighting conditions due to cloud cover in order to acquire onscale data.
Section 3: Cloud Decision Images
3.1) How is cloud cover computed?
Please see cloud cover determination in the Theory of Operations section. In general, an opaque cloud is defined as a cloud of a given whiteness (or measured red/blue ratio). A thin cloud is defined as a region which is whiter than the clear sky would be. In this process, an adjustment is made so that haze is normally not identified as thin cloud. The algorithm does a very nice job of detecting clouds whether they are large or small (the resolution is about 17 m for a cloud at 3000m, and 6 m for a cloud at 1000m). It is not fooled by dark clouds, and it has no problem seeing high clouds. Sometimes thin clouds near the sun will be incorrectly labeled opaque clouds, and the choice of when to define an increasingly thick haze as a thin cloud is something that we hope to better define in the future.
Produced
by the Marine Physical Laboratory, SIO.
Send questions, comments and suggestions about the Atmospheric Optics
Group website to:
webmaster@mpl.ucsd.edu
Copyright © 2002.
Official web page of the University of California, San Diego |