**DPS 2001 meeting, November 2001**

*Session 19. Mars Atmosphere Posters*

Displayed, 9:00am Tuesday - 3:00pm Saturday, Highlighted, Wednesday, November 28, 2001, 10:30am-12:30pm, French Market Exhibit Hall
[Previous] |
[Session 19] |
[Next]

## [19.10] Principal Components Analysis Studies of Martian Clouds

*D. R. Klassen (Rowan University), J. F. Bell, III (Cornell University)*

We present the principal components analysis (PCA) of
absolutely calibrated multi-spectral images of Mars as a
function of Martian season. The PCA technique is a
mathematical rotation and translation of the data from a
brightness/wavelength space to a vector space of principal
``traits'' that lie along the directions of maximal
variance. The first of these traits, accounting for over
90% of the data variance, is overall brightness and
represented by an average Mars spectrum. Interpretation of
the remaining traits, which account for the remaining
~10% of the variance, is not always the same and
depends upon what other components are in the scene and
thus, varies with Martian season. For example, during
seasons with large amounts of water ice in the scene, the
second trait correlates with the ice and anti-corrlates with
temperature. We will investigate the interpretation of the
second, and successive important PCA traits.

Although these PCA traits are orthogonal in their own vector
space, it is unlikely that any one trait represents a
singular, mineralogic, spectral end-member. It is more
likely that there are many spectral endmembers that vary
identically to within the noise level, that the PCA
technique will not be able to distinguish them. Another
possibility is that similar absorption features among
spectral endmembers may be tied to one PCA trait, for
example ''amount of 2 \micron\ absorption''. We thus attempt
to extract spectral endmembers by matching linear
combinations of the PCA traits to USGS, JHU, and JPL
spectral libraries as aquired through the JPL Aster project.
The recovered spectral endmembers are then linearly combined
to model the multi-spectral image set. We present here the
spectral abundance maps of the water ice/frost endmember
which allow us to track Martian clouds and ground frosts.

This work supported in part through NASA Planetary Astronomy
Grant NAG5-6776. All data gathered at the NASA Infrared
Telescope Facility in collaboration with the telescope
operators and with thanks to the support staff and day crew.

The author(s) of this abstract have provided an email address
for comments about the abstract:
klassen@rowan.edu

[Previous] |
[Session 19] |
[Next]