Abstract:

NOAA’s Mussel Watch Program has trackedcoastal contamination since the 1980s, but its sediment chemistry, toxicity,benthic, and tissue records have been analyzed separately. This project testedwhether they can support one national framework. Benthic condition was scoredwith AMBI and M-AMBI and linked to watershed urbanization at 732 sites fromEPA’s 2015 National Coastal Condition Assessment and 33 stations from a 1991NOAA survey of the Hudson-Raritan Estuary, an urban system with legacySuperfund sediment. Random forest and Bayesian models showed that urbanizationpredicted benthic condition in the New York metropolitan area (ρ = 0.62) but not nationally (ρ = 0.05). A Delft3D model ofthe estuary reproduced observed tides (R = 0.97-0.98) and showed sea-level riseraising bed shear in the deep channels.

The results presented are from the NOAA EPP CSC NERTOgraduate internship project conducted under the mentorship of Dr. KimaniKimbrough and Dr. Erik Davenport, National Ocean Service (NOS), NationalCenters for Coastal Ocean Science (NCCOS), Mussel Watch Program. This NERTOexperience aligns with the NOAA Cooperative Science Center in CESSRST-II,supporting the Center’s goal of conducting NOAA mission-aligned collaborativeresearch to understand and predict changes in land and water, oceans andcoasts. This project, titled Using AI/ML and Bayesian Modeling Techniques toStudy Chemical Stressors and Their Impacts, addressed the research question:does urbanization predict benthic condition, and do sediment contaminants carrythat link? The work provides value to the scientific community and programstakeholders by providing a standardized national benthic metric and a modelfor testing whether Superfund cleanups hold under sea-level rise. Through thisinternship, the student deepened their understanding of NOAA mission areas andgained enhanced skills in machine learning, Bayesian statistics, geospatialanalysis, and hydrodynamic modeling.