Spectral Parameter Mapping
Clementine calibrated nine band UVVIS/NIR spectra for a lunar feature of interest may be downloaded in 16 bit tif format from the USGS map-a-planet moon website at: http://www.mapaplanet.org/explorer/moon.html . The UVVIS spectra are downloaded from the "Natural" color Clementine UVVIS multispectral mosaic (415, 750, 900, 950, and 1000 nm). The remaining four bands are NIR spectra and are downloaded from the Clementine NIR multispectral mosaic (1100, 1250, 1500, and 2000 nm). Although two higher wavelengths are available, their calibration is not sufficiently perfected for inclusion and spectral maps here are based on 415 nm to 2000 nm Clementine images. Images will be perfectly co-registered as long as the menu is switched from the UVVIS multispectral to the NIR multispectral mosaic while the feature image remains on the screen. I am careful to only download a single band at a time. Once downloaded the images are named according to their wavelength and are imported into the free program ImageJ (see http://rsb.info.nih.gov/ij/ ) as an image sequence. The sequence is cropped as needed and saved as a txt image sequence.
I wrote an Excel 2007 VBA program that automatically generates a series of 8 spectral maps from the nine band txt image series. Each txt image is stored in a separate worksheet and the resulting spectral maps are formed in eight other worksheets within the Excel workbook. The spectral maps produced are as follows: orthopyroxene absorption band center wavelength minima (890 to 945 nm in 5 nm increments); percent orthopyroxene absorption band depth (890 to 945 nm in 5 nm increments); clinopyroxene absorption band center wavelength minima (950 to 1000 nm in 5 nm increments); percent clinopyroxene absorption band depth (950 to 1000 nm in 5 nm increments); olivine absorption band center wavelength minima (1005 to 1095 nm in 5 nm increments); percent olivine absorption band depth (1005 to 1095 nm in 5 nm increments); the FWHM (full width at half maximum) wavelength in nanometers; and the OMAT (optical maturity map) which indicates the relative "freshness" of a lunar feature and also provides some indication of its age. If desired, slope maps can be readily generated as well.
The series of maps is generated simultaneously from continuum divided spectra between 750 nm and 1500 nm for each map pixel. Although the continuum is adjustable, the default continuum line runs through the 750 nm and 1500 nm data points. New data points are interpolated at 5 nm increments using a cubic spline fit to the Clementine data between 750 and 1500 nm. Absorption trough minima between 890 nm and 945 nm; 950 nm and 1000 nm; and 1005 nm and 1095 nm are identified and olivine "shoulder" inflections are also detected. When a trough minima (or significant olivine shoulder inflection) is detected for each of these wavelength segments, the wavelength of the minimum is mapped and the percent absorption depth is mapped separately. Maps are generated in txt format and are converted to 32 bit tiff format following export to ImageJ. The maps are horizontally reversed in ImageJ since they are generated in column reverse order. Running the cursor over map pixels in ImageJ causes their spectral parameter values to be automatically displayed.
If the presence of olivine is suspected from the appearance of finished olivine band center and band depth maps, then the possibility of impact melt (which can have an absorption band center higher than 1000 nm) might be excluded by examining a 2000nm/1500nm ratio map of the lunar feature (olivine rich features relatively uncontaminated by clinopyroxenes will have a bright appearance). The FWHM map is generated from the spectral curve. The OMAT map is produced from its defining formula established by Lucey et al. (see Lucey PG et al. (1995) Science, 268, 1150-1153 and Lucey PG et al. (1998) JGR, 103, 3679-3699).
This automated mapping protocol in Excel 2007 is useful because the spectral parameters are viewed in the context of the spectral results for all map pixels for the feature of interest. If, on the other hand, the spectra of a group of pixels is analyzed purely manually then there is no spectral context of adjacent pixels unless these are also analyzed manually. So, it is obvious that automated spectral mapping saves a great deal of time and effort over manual analysis of a selected group of pixels. Manual analysis is saved for checking specific areas of interest demonstrated in the automated 8 spectral parameter maps described above.
In addition to the maps discussed above, it is also possible to generate maps of slope normalized to 750 nm, mafic absorption band depth, titanium content, and iron content according to the method discussed by LeMouelic (2000) (see: JGR, 105, 9445-9455) and Lucey (1998) (see: JGR, 103, 3679-3699). For the most part, this is easily and quickly done by applying equations for various parameters that are listed in these papers to Clementine 415, 750 , 950, and/or 1500 nm images. The necessary image arithmetic functions are available in the free ImageJ program described above under the Process menu by using either the Math or Image Calculator options as needed. The titanium equation requires that an arc tangent be applied and this can be done easily using Excel.
If all goes well, it is anticipated that an in depth discussion of the mapping program and numerous examples of generated spectral parameter maps for a number of different lunar features will appear in the free online journal Selenology Today (No. 14) which should be released some time in the spring or early summer of 2009. Meanwhile a general discussion is found in abstract #1093 (1093.pdf) of the 40th Lunar and Planetary Science Conference (R. Evans, C Wohler, R Lena. Analysis of absorption trough features using Clementine UVVIS+NIR imagery. LPSC 2009 abstract #1093). See: http://www.lpi. usra.edu/ meetings/ lpsc2009/ pdf/1093. pdf
In the interim, as time allows, 8 bit jpg map images for a few selected lunar features of interest should eventually be viewable here, although this format does not conserve the spectral parameter values directly as is the case for 32 bit tiff images. Black pixels indicate a negative result for all maps. A study of the crater Archimedes is below. Spectral mapping is a new technique for me and interpretations of the maps are likely to change and evolve as I have the opportunity to study more and more lunar features over time. Please note however, that when you see (for example) green in the false color maps and it is indicated as "olivine" that you MUST refer to the band depth percent to understand whether this is a rich olivine deposit (like in the Copernicus central peaks) or merely a small or trace amount of olivine present in a more complex rock/lava mixutre. This mapping method is extremely sensitive, and typically shows the presence of extremely small concentrations of pyroxines and olivine. There is the additional problem that although the green pixels are labelled as "olivine" they could alternatively represent impact melt or pyroclastic material (i.e. they show absorption between 1005 nm and 1095 nm and I have added the identifier "olivine" as an empirical identifier only) and for correct interpretation the geologic context etc. must be considered as well as ratio image information. I certainly am not trying to say here from this data that the moon is full of concentrated olivine 
Example: Spectral Maps for Hansteen Crater and the Mons Hansteen lunar "red spot" feature:
Note that BCW stands for absorption band center wavelength. Brighter pixels indicate a higher band center wavelength or band depth in the greyscale images above. This is also true within a particular color in the false color images.
Other Examples: Archimedes
Clementine Natural Color Image of Archimedes Crater
In the maps above, bcw refers to the band center wavelength of the absorption trough. Band Depth refers to the percent depth of the band center of the absorption wavelength. Op is orthopyroxene, Ol is olivine, and Cp is clinopyroxene. Archimedes is a lava flooded crater and this seems to correspond to the presence of the minor olivine component (green) seen within the crater floor. The clinopyroxene content of the wall varies from what appears to be an anorthosite with a minimal gabbroic component (darker blue) to areas (brightest blue) that are more gabbroic. Reddish areas correspond to anorthosite with a minor orthopyroxene component (darker red) to areas slightly more enriched in the orthopyroxene component (brighter red). Another example of spectral mapping, Hadley Rille, is given below.
The maps of the Hadley Rille below show essentially no orthopyroxene within the rille, which contains both clinopyroxene and olivine. This would be suggestive of an olivine basalt.
OPTICAL MATURITY MAP (OMAT):
Grey => White : 0.188 => 0.370
Another Example of Spectral Mapping: Manilius Crater
Referring to the 32 bit tiff images, the Clinopyroxene content of Manilius has an absorption trough centered at 995 to 1000 nm with an average depth of 8.1 % in the central peaks and 5.5 to 7.6 % in the crater rim. This likely represents a primarily anorthositic composition with a minor gabbroic component. Weak olivine content is present in basalts just outside the crater, especially to the southwest and has a band center between 1025 and 1095 nm with an average depth of 1.6 to 3.7%. Orthopyroxene content is present only focally in the eastern and southeastern portions of the crater rim with a band center of about 925 to 945 nm and an average depth of about 3.0 to 3.4 % indicating a probable anorthositic composition with a very minor noritic component.
Portion of Fracastorius Crater:
Fracastorius is a complex area and the interpretation of the spectral maps is difficult. For now, my initial impressions are given below. Fracastorius is mainly anorthositic but there are variations in the minor mafic component that is present that can be detected with the spectral mapping technique as shown in the graphics below. The crater rim has a very minor gabbroic component (shown in blue) and part of the floor has a very minor noritic components (shown in red). This may well represent a thin layer of ejecta present on the floor. This component has a band center between 890 and 930 nm. There are also some green areas most likely corresponding to a very minor olivine component. There is a large dome with a mixed composition most likely consisting of a weak olivine component with some ejecta contamination containing anorthosite with very minor orthopyroxene content. The dome area is circled in a broken yellow line.
