Data-driven Characterization of Crater Morphology During Plume-surface Interaction

The impingement of high-speed rocket plumes on regolith surfaces during propulsive landings on the surface of the Moon or planetary bodies erodes a crater and lifts ejecta that can compromise the lander and nearby infrastructure. There is a lack of engineering models that can aid risk prediction, stemming from an incomplete understanding of plume-surface interactions (PSI) and the diverse erosion mechanisms dominating in different regimes. In this work, we use data from recent sub-scale PSI experiments performed at NASA to advance the current understanding of erosion phenomenology and cratering predictions. Experiments involved a supersonic jet impinging on a granular bed, with a splitter plate aligned with the jet axis allowing visual access to the crater cross-section. This work uses the high-speed imagery of the crater as it evolves under the action of the jet. Custom image-analysis methodologies are used to extract the profile of the crater as the granular surface is eroded. For some of the test cases, the crater shape can be approximated using a family of standard Gaussian curves, which enables crater profiles to be defined with minimal parameters. Using the symbolic regression framework PySINDy, a dynamical system is identified that reduces the crater evolution to a set of ordinary differential equations. The dynamical system generated and presented in this work showed a good agreement with the experimental data.