Progressive Cloudcover Penetration by Multiple Satellites or Multiple Satellite Passes

Bill Stoney, Alan Goldberg, Stacy Bunin, & Rob Groff
Mitretek Systems, Inc.

The material presented here is preliminary, and has been placed on the WWW for internal review. It is not to be duplicated without the permission of the authors

Set your browser to view the full length of the line below


The animated GIFs linked below show the progressive increase which can be expected by multiple attempts to image the earth from satellites such as Landsat.

Proper viewing requires Netscape 2 or other browser or helper application which supports dynamic GIFs. The full animation rate typically requires a 50MHz or faster computer.

Each animation begins with the distribution of recorded cloudcover in one full set of land scenes taken over one complete 16-day cycle at the beginning of the indicated season. "Seasons" were defined for computational convenience, and do not correspond to astronomical or meteorological norms, but their start dates are separated by 91 days.

The first animation shows the improvement obtained by continuing observations over 2 to 6 cycles, covering the entire season.

The other animations show the effects of multiple satellites in each season. Subsequent overlays show the improvement which is obtained by using more satellites simultaneously with different ground starting positions. Over the range of one to four satellites, their is no significant difference between using multiple satellites or multiple cycles to improve the likelihood of achieving adequately low cloud obscuration. However, in most applications, timeliness of cloudfree collection may be the driving requirement, leading to the need for multiple satellites as the only satisfactory solution.

Base satellite viewing geometry is that of the current Landsat series, i.e., sun-synchronous circular orbit at 705 km altitude and with 10:30 AM descending nodal crossing time. Inclination is 98.2 deg. The EOS AM series satellites will operate in the same orbit, with a later nodal crossing time.

Map projection is based on the Landsat 4 Worldwide Reference System (WRS), which identifies each possible Landsat scene by a path (corresponding to one of 233 unique daylight subsatellite tracks) and a row (corresponding to one of 121 standard points along a path, separated by nominally 24 s through an orbit). Path numbering increases toward the right, row number toward the bottom. At the equator, paths are separated by slightly less than a full Landsat scene's 185 km width, and rows by slightly less than a scene's 170km height. This map projection can be considered a conceptual extension of Snyder's Satellite Oblique Mercator (SOM) projection. It suffers from the typical mercator distortion toward the poles, and has an S-shaped distortion due to the orbit not being inclined exactly 90 deg.

Base cloud data is derived from a standard USAF 1977 synoptic database. It has been resampled to the local satellite viewing time over each WRS scene. Average cloudcover in the original database has been resampled to the WRS and averaged over the Landsat scene dimensions.

Color coding on the maps ranges from nearly or totally obscured (black), to nearly or totally clear (red). In particular ...

Additional assumptions include the following:


Spring season, one satellite in orbit for one to six 16-day cycles


Spring season, one to four satellites in orbit for one 16-day cycle


Summer season, one to four satellites in orbit for one 16-day cycle


Fall season, one to four satellites in orbit for one 16-day cycle


Winter season, one to four satellites in orbit for one 16-day cycle


Alan Goldberg, Mitretek Systems, Inc., McLean VA 22102.

Return to "Land Sensing Satellites in the Year 2000"
Return to recent papers subpage.

Return to Landsat homepage