Initial Results from a GIS-based Unsupervised Classification Study of the Martian Surface

Maps of thermal inertia-albedo units and thermal inertia-elevation units on Mars’ surface have been generated by choosing thresholds that fit the strongest peaks in the histograms of these datasets. The units thus defined were then interpreted as distinct mixtures of materials on the surface, such as: bright fines, rock + bedrock and ice. We have conducted an initial classification of Thermal Emission Spectrometer (TES) night-time thermal inertia and TES albedo using a hard classifier. The methods used here are largely unsupervised and differ from those of previous studies. The aim of our study is to investigate what information can be obtained by utilising unsupervised classification algorithms to investigate the distribution of thermal materials on the surface of Mars. We find that unsupervised classification reveals additional structure in the clustering and spatial distribution of surface materials with moderate-low albedo and moderate-high thermal inertia. We highlight a number of regions such as Acidalia and Valles Marineris for future detailed studies of this type.