Modeling Martian Thermal Inertia in a Distributed Memory High Performance Computing Environment

Modeling martian surface properties fusing high resolution, spatially enabled, remotely sensed data and derived thermophysical modeling is an essential tool for surface property characterization studies. In this work, we describe the development of a thermal inertia modeling tool that integrates the KRC thermal model and a nine-dimensional parameter interpolation with inputs draw from remotely sensed data. Our model is classifiable as operating in both the Big Data and Big Process domains. We utilize the KRC thermal model to generate a dense lookup table. We show that the overall size of the lookup table necessary to derive thermal inertia can be reduced, through sampling, by approximately 82% while maintaining a high level of accuracy at those regions of the parameter space where thermal inertia is most frequently derived. This level of data reduction supports the distributed, in-memory application of our model and we illustrate the computational performance through a classic scaling experiment. This work extends previous modeling efforts by allowing for pixel scale thermal inertia modeling at the highest spatial scales allowed, and enabling surface properties investigations at spatial scales relevant to addressing high-priority science and engineering questions.