Project Grant 2616414

The National Science Foundation Office of Integrative Activities awarded the University of Miami $277,083 on August 15, 2026, under the Geosciences program (CFDA 47.050) to develop artificial intelligence tools for analyzing dense seismic array data from temporary field deployments in Oklahoma, Texas, and Hawai'i.

The project will create a geometry-aware self-supervised learning framework that processes paired time-domain and frequency-domain seismic recordings while accounting for variations in station layout, noise conditions, and deployment geometry. The learned representations will connect to source properties through downstream tasks including source-physics mapping and rare or under-recognized event discovery. The work aims to help scientists distinguish natural from human-related seismic activity, recognize emerging event types, and extend the utility of temporary dense-array observations by linking short-term field data to long-term monitoring records. The project will also produce reusable workflows and curated training resources for the research community.

Performance extends through July 31, 2029, with work conducted in South Miami, Florida. The award supports training of students and early-career researchers at the intersection of geoscience and machine learning, addressing the challenge that dense seismic arrays generate data volumes too large and complex for manual analysis despite their value for monitoring fluid movement through the subsurface in geothermal fields, wastewater-disposal regions, and active volcanoes.