Real-time knowledge of whale distributions would allow imminent harmful activities to be temporarily redirected, relocated, or postponed, helping to reduce threats to the endangered north atlantic…

This Cooperative Agreement award, funded by the National Oceanic and Atmospheric Administration (NOAA) under the Unallied Science Program (CFDA 11.472), is aimed at building a near real-time modeling system to predict the density of the endangered North Atlantic right whale (NARW) population along the U.S. Eastern Seaboard. The $1,381,943 award to Duke University will support the development and testing of this system, which will link visual and acoustic observations of right whales with environmental data to generate daily predictions of whale density. Initially, the system will serve as a research and development testbed, with input from managers and stakeholders, to ensure the reliability and utility of the automated predictions. Once the model predictions are deemed effective, they will be made publicly available on an operational basis. The system is designed for future evolution, allowing new candidate models to be developed and assessed. If suitable methods are developed, the system may also be expanded to issue short-term forecasts and be applied to other marine species.

Real-time knowledge of whale distributions would allow imminent harmful activities to be temporarily redirected, relocated, or postponed, helping to reduce threats to the endangered north atlantic right whale (NARW) population. In partnership with NOAA fisheries and the wider community of marine researchers and stakeholders, duke mgel is building a near real-time modeling system for the NARW. By linking the real-time flow of visual and acoustic observations of right whales to contemporaneous environmental covariates, the system will predict whale density across the eastern seaboard at a daily cadence. Initially, the system will serve as a research and development testbed for a collaborating team of developers to work out the statistical challenges and software workflow needed to obtain reliable automated predictions and accompanying skill assessments, with managers and stakeholders providing guidance and assessing the utility of the results. Once model predictions are judged to be sufficiently effective by a steering committee of researchers, managers, and stakeholders, daily predictions will be made operationally and released to the public. The system will be designed from the start for future evolution, allowing new candidate models to be developed and assessed against each other and the currently-operational model before eventually replacing IT. If data allow and suitable methods are developed, the system can be extended to issue short-term forecasts, and expanded to other species.