Use of Geochemistry Data Collected by the Mars Exploration Rover Spirit in Gusev Crater to Teach Geomorphic Zonation through Principal Components Analysis
Rodrigue, Christine M., Journal of Geoscience Education
This paper presents a laboratory exercise used to teach principal components analysis (PCA) as a means of surface zonation. The lab was built around abundance data for 16 oxides and elements collected by the Mars Exploration Rover Spirit in Gusev Crater between Sol 14 and Sol 470. Students used PCA to reduce 15 of these into 3 components, which, after quartimax rotation, very strikingly divided the surface traversed by Spirit's into three distinct zones. Students then used such concepts as the Bowen reaction series, typical minerals in Earth's basalts and andesitic arcs, the periodic table, and the Goldschmidt classification, together with Pancam images from Spirit and the Mars Orbiter Camera, to interpret the surfaces over which the rover moved. Students found this foray to Mars a challenging but enjoyable project, and it made PCA memorable to them long after the class had ended. Some variant on this lab could work for multivariate statistics courses in geology, geography, and environmental science, as well as advanced courses in the content of those disciplines, particularly those dealing with zonation. © 2011 National Association of Geoscience Teachers. [DOI: 10.5408/1.3604826]
This paper presents an exercise that uses Mars Exploration Rover geochemical data to teach principal components analysis (PCA) for geomorphic or geological zonation. The data come from the Spirit rover's Alpha Particle X-ray Spectrometer (APXS), which collected spectra from 93 rocks and soil samples (Geliert et al, 2006) during its travel over three distinctive zones on the floor of Gusev Crater. These zones consisted of a cratered basaltic plain, the West Spur of the Columbia Hills with bedded materials and evaporites, and the northwest side of Husband Hill where very diverse aqueous and acid-aqueous altered rocks and soils were found. PCA is a data reduction technique that has increasingly been used in the geosciences since the early 1960s, making its acquaintance of value in the education of geoscience majors. The APXS data can make the technique memorable to such majors as it produces a coherent zonation from 15 different oxides and elements.
A classic task in the geosciences is zonation of complex surface patterns into areal units and demarcating transition zones or boundaries between them, often along a transect in the field. So, for example, a soil catena can be zoned by changes in soil particle size, underlying bedrock and regolith, topographic relief, drainage, erosion and deposition processes, weathering, organic matter, and geochemistry (Milne 1935; Bushneil, 1942; Webster, 1973; Raynolds et al, 2006). Ground-penetrating radar can be used along a transect to infer subsurface stratigraphy for geological mapping (Baker and JoI, 2007). An environmental ecotone might be zoned by field sampling of soils and censusing of species presence and abundance along a transect. For example, a transect could be taken down a catena, across a wetland-upland interface, or through a seasonal surface water and groundwater boundary (Fortin et al, 2000).
Zonation can be vertical and temporal in geological usage, not just horizontal and spatial in mapping usage. So, for example, fossils, grain size, bulk density, and geochemistry can be used for temporal zonation and sequencing of stratigraphie units (e.g., Patterson et al, 2000; Brown and Pasternack, 2004; Peterson et al, 2008).
Zonation, then, is a common task in the field and laboratory activities of geoscientists. The process can seem superficially straightforward, but the zoning schemes that result can color analytic results. Complications include scale, edge effects, spatial autocorrelation, and aggregation effects. These distortions and biases are collectively called the Modifiable Areal Unit Problem or MAUP (Dark and Bram, 2007) or the analogous Modifiable Temporal Unit Problem (MTUP). The MTUP is less commonly discussed, largely in criminology contexts (e. …