Image reconstruction of remotely-sensed planetary data

Abstract

Planetary data measured from orbiting spacecraft are defined on a sphere. Therefore, the geometry of the sphere must be taken into account in their analysis. In this thesis, I develop a Bayesian image reconstruction technique, based on pixon image reconstruction, but allowing the solution to tailor its fit to structure in the image. Unlike older implemen- tations this new method is capable of reconstructing data defined across the entire sphere without the need to partition the surface into approximately flat regions.

The thorium line data from the Lunar Prospectors Gamma Ray Spectrometer are enhanced using pixon reconstruction. The enhancement shows that at the Compoton- Belkovich Volcanic Complex, thorium is distributed over a larger (40 km × 75 km) area than the (25 km × 35 km) high albedo region normally taken to define the feature. The thorium must have been deposited during the formation of the volcanic complex, because subsequent lateral transport mechanisms, such as small impacts, are shown to be unable to move sufficient material. The morphology of the feature is consistent with pyroclastic dispersal and we conclude that the present distribution of thorium was likely created by the explosive eruption of silicic magma.

Application to Mars Odyssey Neutron Spectrometer epithermal data yields an improvement in resolution from 520 km to 290 km. This new high-resolution data is used to infer the global distribution of Water Equivalent Hydrogen (WEH) and reveals WEH-rich features near the equator at the Medusae Fossae Formation and the Tharsis Montes, which may indicate the presence of water ice. The data are also used to infer the polar abundance of CO2 throughout the Martian year, which is found to exhibit behaviour not predicted by the Martian general circulation models.