GLOBAL MULTISPECTRAL MAPPING OF THE MOON BY CLEMENTINE 
A.S. McEwen and M.S. Robinson 
U.S. Geological Survey, 2255 N. Gemini Drive, Flagstaff, Arizona 86001, USA 
KEY WORDS: Remote Sensing, Geology, Multispctral, Image, CCD 
ABSTRACT: 
The Clementine spacecraft was built and operated by the Naval Research Laboratory, carried a suite of lightweight 
sensors designed by the Lawrence Livermore National Laboratory, and mapped the Moon from Feb. 19 to May 3, 
1994. Near global coverage of the lunar surface was acquired in 11 spectral bandpasses from 415 to 2780 nm and at 
resolutions of 80-250 m/pixel (UVVIS and NIR cameras) (Nozette et al., 1994). A thermal-infrared camera (LWIR) 
sampled ~20% of the surface at 8575 nm in thin strips of data that stretch from pole-to-pole. A high-resolution camera 
(HIRES) mapped the polar regions and acquired visible color in four wavelengths (415 - 750 nm) for areas of high 
scientific interest. Finally, a lidar altimeter (LIDAR) mapped the large-scale topography of the Moon up to latitudes of 
75 degrees N and S. Clementine was in a polar, elliptical orbit, ~400-450 km periselene altitude. Periselene latitude 
was -30 degrees for the first month of systematic mapping, then moved to +30 degrees. This strategy maximized the 
resolution uniformity for the 11 band global color dataset. NASA is supporting the archiving and analysis of the ~1.8 
million lunar images and ancillary datasets. The raw data are stored on CD-ROM (available through INTERNET 
access) that are being widely distributed to the science community and the general public. 
The first major step in the systematic The global base map is being constructed 
processing of the UVVIS and NIR global imaging with 750 nm filter images. After this mosaic is 
data is the production of an accurate base map, to completed the corresponding UVVIS (415, 900, 950, 
which all other products will be geometrically 1000 nm) and NIR (1100, 1250, 1500, 2000, 2600, 
registered. Current maps and control points of the 2780 nm) bands are to be coregistered. This 11 band 
Moon are not adequate. The previous RAND control image cube will provide the opportunity to map global 
network (Davies., et al., 1994) is accurate to 500 m in lunar mineralogy at a resolution of ~250 m/pixel. 
the regions of the Moon covered by the Apollo These wavelengths were specifically selected to 
missions (15% of the Moon's surface). This previous diagnose known spectral characteristics of returned 
control network is accurate to about 1-2 km for lunar samples. In addition to the multispectral 
regions covered only by telescopic, Galileo, and mapping, Clementine also acquired stereo image data 
Mariner 10 observations. However, most of the far for the lunar polar regions, the Orientale basin region, 
side is not included in the network, and the only other and other selected targets. These data can be 
positional dataset for these regions (Duxbury, ef al., processed to obtain digital terrain models (DTMs) that 
1994) contains errors as large as tens of kilometers. portray topography on a pixel-by-pixel basis (Oberst, 
Based on our current best measurements of the et al, 1995). Combined with global scale lidar 
spacecraft orbit and pointing, UVVIS geometric topography (Zuber, et al., 1994) these high resolution 
distortions, and time tags for each observation, we stereo models will allow for quantification of internal 
expect the SPICE data alone will provide positional and surficial lunar processes. 
accuracies better than 1 km over most of the Moon 
(McEwen, et al., 1995). Many scientific questions concerning the 
origin of the Moon, the nature of its crust, styles of 
Our goals are to provide better than 0.5 volcanic eruptions, surface weathering, and impact 
km/pixel absolute positional accuracy everywhere on history (to name just a few) remain in question. This 
the Moon except for gaps in nominal coverage that new dataset will greatly alter our understanding of the 
were filled by highly oblique data (< 1% of the Moon. Early science results include: 1) a global 
surface). The new global geodetic network is being model of crustal thicknesses (Zuber, ef al., 1994); 2) 
constructed from ~43,000 images from which 0.5 new information on the topography and structure of 
million match points have been selected. The average multiring impact basins (Zuber, et al., 1994, Spudis, et 
relative positional error, after match point comparison al., 1994); 3) evidence suggestive of water ice in large 
and camera updating, is about 80 m, less than 1 pixel. permanent shadows near the south pole (Nozette et al., 
The average absolute positional accuracy is estimated 1994); 4) global determination of crustal iron 
to be better than 250 m/pixel. This vastly improved abundance (Lucey et al., 1995); 5) reevaluation of the 
control network will facilitate future lunar Copernican impact crater population (Moore and 
exploration, as well as provide an invaluable base for McEwen, 1996); 6) an extension of known regions of 
geologic mapping. anorthositic crust (McEwen et al, 1994); 7) 
improvements in lunar stratigraphic relationships 
702 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996