Data collected from NASA’s Earth Surface Mineral Dust Source Investigation (EMIT) provides one of the first opportunities to use imaging spectroscopy for mapping geologic units from space. Based on the hypothesis that there exist predictable linkages between geochemical characteristics of the lithosphere and the spectra of the land surface, we used EMIT imagery coupled with the geologic map of the McDermitt Caldera, NV, to train partial least squares discriminant analysis models for differentiating geologic units. Initial validation results showed mixed but consistent outcomes, with some units predicting up to 90% accuracy. Consolidating geologic units based on their chemical compositions and textural similarities helped increase classification accuracy overall, but some units remained poorly predicted. For those units, we found remarkable improvement through the inclusion of vegetation spectra that was originally omitted. This highlights that vegetation is reflecting important features of the bedrock that aids in remotely mapping geologic units.