A METHODOLOGY FOR EVALUATING REMOTELY-DERIVED DATA AND ITS APPLICATION TO LUNAR AND MARS ROVER ASTROGEOLOGY EXPLORATION

The analysis and interpretation of robotically derived data presents new and unique challenges not associated with more traditional methods of field geology. One key factor currently missing in planetary geology missions and rover test simulations is the ability to link higher-level interpretations and results directly to low-level observations and the specific data from which they are derived. This limits our ability to critically evaluate the quality of data being produced, the effectiveness of the instrumentation package being utilized, and the functionality of the scientific working team to generate the most effective and informative interpretation of the evidence possible. We propose that documenting the working methodology of geologists as they interact with remotely-derived data will allow us to reconstruct the link between data and conclusions. This will allow us to critically evaluate, within the context of the complex phenomena and asymmetry of causation inherent in geologic systems, individual lines of scientific reasoning that we will be able to trace from inception to fruition. Our methodology is to create a semiotic network, or explicit representation of the concepts and relationships, as used by a group of mission scientists when evaluating data, and more specifically when interpreting rock type from digital images and geochemical data. By understanding the cognitive and reasoning structures actually used in practice, rather than a rational reconstruction of what the scientists did (as is generally presented in final documentation), we will be able to critically evaluate the scientific productivity of future exploration missions and target efficient optimization of instrumentation.