The seismic noise recorded by the InSight seismometer (SEIS) has revealed numerous transient micro-events pointing to an active Martian environment, including wind bursts, pressure drops and thermally induced cracks. Identifying these micro-events is crucial to ensuring that none are wrongly classified as a “Marsquake.”

GMIG has successfully carried out such identifications with an unsupervised deep-learning approach built on a deep scattering network and GMMs. The clustering and detection efficiency for pressure drops and glitches are superior to current manual detection techniques. Because interest in space travel is consistently on the rise, as are questions about the presence and extent of biological activity outside Earth, the applications of GMIG’s research to space exploration have led to exciting opportunities for collaboration with corporate partners and academic institutions.

GMIG’s has produced a deep catalog for the different types of Marsquakes by adapting and modifying its polyphonic recurrent scattering neural network for seismo-volcanic monitoring. The goal of this neural network is to better understand Mars’ stratified interior, applying deep-learning approaches to inverse problems and inference while remaining on the leading edge of the study of Mars.

Further reading: Anatomy of continuous Mars SEIS and pressure data from unsupervised learning