AEG-DMV May 2023 Meeting with Potomac Geophysical Society (PGS)

 Abstract:

The costly power and data volume limitations of delivering seismic data back to Earth from planetary missions requires the development of algorithms for lander-side signal analysis for telemetry prioritization. This is difficult to explicitly program, especially if no prior seismic data are available from the planetary body. We demonstrate that Convolutional Neural Networks can be used to accurately catalog planetary seismicity without local training data by building binary noise/signal classifiers from a single Earth seismic station and applying the models to moonquakes from the Apollo Passive Seismic Experiment (PSE) and the Lunar Seismic Profiling Experiment (LSPE). Instead of assessing the seismic signals by their time series, we convert the data to spectrogram images, which are easier to generalize with only a fraction of data. Simple two- to five-layer convolution models are built using solely 580 seismic events and tested against a subset of 200 Grade-A events from the PSE…