Caig: predicting the distribution of rare-earth element-rich sediment using artificial intelligence-based geospatial mapping and fish tooth fossils

The National Science Foundation Office of Integrative Activities awarded Woods Hole Oceanographic Institution $1,356,482 on September 1, 2026, under the Geosciences program (CFDA 47.050) to predict the distribution of rare-earth element-rich sediment in the Pacific Ocean using artificial intelligence-based geospatial mapping and fish tooth fossil analysis.

The project applies machine learning to sediment metal and fish fossil datasets to identify processes that result in high concentrations of rare-earth elements and improve predictions of where rare-earth element-rich deposits occur. The researchers will interrogate variables affecting rare-earth element concentrations, including scavenging from seawater by iron and manganese oxides, incorporation into calcium phosphate minerals such as fossil fish teeth, and the duration sediment layers are exposed to ocean bottom water. The project will employ core-top artificial intelligence-enhanced elemental mapping to determine modern drivers of rare-earth elements in surficial marine sediments and improve fish tooth biostratigraphic dating of red clays through artificial intelligence methods. Complementary statistical approaches will explore relationships between rare-earth element patterns and sediment property datasets to evaluate the null hypothesis that rare-earth element concentrations are randomly distributed on the seafloor through space and time. The project will develop classroom activities addressing deep-sea sediment, marine fossils, and how artificial intelligence supports ocean exploration.

Performance occurs at Woods Hole, Massachusetts, with a period of performance from September 1, 2026, through August 31, 2029.

Abstract

Rare earth elements (rees) are essential to modern technology, U.S. Economic competitiveness, and national security. Technological sector growth has motivated interest in mining REE-rich deep-sea sediment. Some of the highest known REE concentrations in sediment occur in the deep pacific ocean. However, their distribution is not well known. Traditional sampling methods are inefficient for mapping their spatial variability. This project will identify the processes that result in high concentrations of rees in sediment and improve predictions of where REE-rich deposits will occur. The researchers will apply machine learning to sediment metal and fish fossil data. This combination will reveal when, where, and how rees become enriched in sediment. The project will develop classroom activities on topics like deep-sea sediment, marine fossils, and how artificial intelligence supports ocean exploration. There is a fundamental GAP in understanding basin-wide controls on REE enrichment in marine sediment, with the current best-guess models assuming that REE concentrations are randomly distributed on the seafloor through space and time. This project evaluates that null hypothesis by interrogating variables that may impact REE concentrations, including scavenging from seawater by iron and manganese oxides, incorporation into calcium phosphate minerals such as fossil fish teeth, and the length of time a sediment layer is exposed to ocean bottom water. A complementary suite of artificial intelligence (AI) tools and other statistical approaches will explore relationships between REE patterns and sediment property datasets. The researchers will use core-top AI-enhanced elemental mapping to determine the modern drivers of rees in surficial marine sediments. Fish tooth biostratigraphic dating of red clays will be improved through AI-driven image classification. Finally, big-data exploration of the relationship between rees and scientific ocean drilling core physical properties will expand the predictive power of these relationships to the full pacific ocean basin. The resulting models will test whether REE enrichment follows predictable spatial and stratigraphic patterns, increasing knowledge of REE abundances in regions where no direct REE measurements have been made. The outcome of the project will advance our understanding of marine metal cycling and demonstrate how data-driven approaches can transform the interpretation of seafloor geochemistry. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are planned for this award.