Probabilistic Geobiological Classification Using Elemental Abundance Distributions and Lossless Image Compression in Fossils, Meteorites, and Microorganisms

摘要

Last year at this symposium we introduced a strategy for the automated detection of fossils during robotic missions to Mars using both structural and chemical signatures. The strategy employs a measure derived from information theory, lossless compression of photographic images, to estimate the relative complexity of a putative fossil compared to the rock matrix. Following target selection unsupervised multifactor cluster analysis of elemental abundance distributions provides an initial classification of the data. This autonomous classification is then confirmed using a non-linear stochastic neural network to produce a Bayesian estimate of classification accuracy. We have now employed this strategy to explore extant and fossil cyanobacteria from a variety of extreme terrestrial environments and microfossils and abiotic microstructures found in-situ in freshly fractured internal surfaces of carbonaceous meteorite. Elemental abundances (C, N, O, Na, Mg, Al, Si, P, S, Cl, K, Ca, Fe) obtained for both extant and fossil cyanobacteria produce signatures distinguishing them from meteorite targets and from one another. Fossil cyanobacteria exhibit significant loss of C, N, O, P, and Ca and increases in Al, Si, S, and Fe relative to extant organisms. Orgueil structures exhibit decreased abundances for C, N, Na, P, Cl, K, and Ca; and increases in Mg, S, and Fe relative to extant cyanobacteria. Fossil cyanobacteria are distinguished from Orgueil samples by relative increases in Al, Si, and Fe; and by diminished O and Mg. Compression indices verify that variations in random and redundant textural patterns between perceived forms and the background matrix contribute significantly to morphological visual identification. The results provide a quantitative probabilistic methodology for discriminating putatitive fossils from the surrounding rock matrix and from extant organisms using both structural and chemical information. The techniques described appear applicable to the geobiological analysis of meteoritic samples or in situ exploration of the Mars regolith.