Probabilistic assessment of surface composition by remote imaging spectroscopy(NPO-51367-1)

The algorithm is being used primarily as a research tool. We ran experiments demonstrating this strategy with end-to-end accounting and propagation of uncertainty across calibration, reflectance, and mineral detection with closure of error budgets and residuals consistent with uncertainty predictions. To our knowledge, this is the first time statistical validation of second-order uncertainty predictions has been demonstrated in the do- main of remote mineralogy mapping. The approach applies to surface composition measurement in other planetary and Earth science domains.