SAS, a global leader in data and AI, is partnering with Fathom Science Inc. a tech spin-off from North Carolina State University. Together they will help protect the critically endangered North Atlantic right whale. Fathom Science is developing digital twins of the ocean to deliver precise, real-time insights that support sustainable marine operations and conservation. One of their key innovations is WhaleCast, a predictive model that estimates whale activity along the U.S. East Coast, enabling ships to proactively reduce speed and avoid whale strikes.
Taylor Shropshire, Head of Marine Resiliency at Fathom Science, created WhaleCast. Taylor combined historical whale sighting data with Fathom’s proprietary ocean models. The result is a dynamic heatmap projection that can be integrated into vessels’ existing navigation screens. This allows mariners to anticipate right whale presence and adjust course or speed to minimize harm.
To enhance WhaleCast’s accuracy and reliability, Shropshire collaborated with SAS through its Data for Good program. A volunteer data scientist from SAS used advanced machine learning tools available in SAS® Viya® to validate and improve the model. To strengthen the statistical foundation.
SAS® Data Maker generated synthetic data mirroring real-world conditions, expanding the dataset to nearly 500,000 data points. This allowed for more rigorous model training, validation, and testing across seven machine learning approaches.
Beyond initial validation, the development team used SAS Viya Workbench, a standalone programming environmentto develop an additional model. This model calculates the probability of whale proximity to shore and further refines WhaleCast predictive capabilities.
“SAS’s speed and diversity in modeling are impressive,” Shropshire said. “They moved from simple models to complex neural networks, clearly demonstrating the strengths and trade-offs of each.”
Fathom Science’s work with SAS showcases how AI and analytics can power marine conservation and safeguard vulnerable species.
