Benjamin Moyer

My primary research interest is to better characterize the subsurface of planetary bodies by leveraging modern computational resources in order to make inferences from a variety of geophysical datasets, particularly magnetic and gravitational anomaly data. Magnetic and gravity anomaly surveys are relatively inexpensive, provide wide coverage, and do not require physical coupling with the planetary body of interest. However, the constraints they provide result in a high degree of non-uniqueness, which has traditionally been removed by introducing assumptions about the interior. I choose to embrace this uncertainty within a trans-dimensional Bayesian framework. My efforts are focused on novel approaches to model proposal generation, greater incorporation of prior information, and improvements in computational efficiency through the aggressive exploitation of parallel algorithms, clever forward modeling computation, and GPU acceleration. These efforts will make the three-dimensional model domains tractable, enable the practical addition of other geophysical measurement types for simultaneous joint inversion, and facilitate the inversion of even larger datasets.

In addition to my primary research, I explore the use of machine learning to estimate the size of seismic events from a single station, using broadband three-component waveform data. This departs from traditional approaches that rely on network-scale observations and instead investigates what information can be extracted from minimal data. This work has relevance for national security and monitoring applications, where rapid characterization of both natural and human-generated seismic sources is important, particularly in data-limited environments.