Scientists are refining our understanding of the world based on denser data across rapidly increasing spatial and temporal scales. My research enables discovery by developing theoretical and computational methods for efficiently analyzing data at unprecedented scales, particularly continuously streaming data from many-sensor networks.
More broadly, my research spans data-intensive high performance computing (HPC), imaging science, inverse problems, signal processing, and data science for physical sciences. In recent years, I have focused on seismology with distributed acoustic sensing (DAS), a rapidly developing technology repurposing fiber optic cables as a means of data transmission and a dense meter-scale (now pushing even to centimeter-scale) array of sensors. I have applied this to monitor permafrost thaw, earthquake detection, imaging for earthquake hazard analysis, studying hydrology, and near-surface geohazards, all in urban areas or near infrastructure.
I am involved in data acquisition, both fit-for-purpose experiments and new piggy-back experiments using fiber in existing telecommunications. In this way, I investigate novel data that motivates algorithm development needs. In detecting and extracting weak signals (close to the noise level), and in searching for repeating patterns in data, we often need to cross-correlate data. This common signal processing technique can require a great deal of data movement per floating point operation, so I have focused recently on developing techniques in multiple dimensions to make correlation analysis more scalable to many sensors, long time series, and large images.