As climate change is impacting water resources and aquatic ecosystems, there is a great need for natural resource managers to assess adaptation measures in a holistic manner. This can be done by integrating model predictions of climate, hydrology, and ecosystems with observational data to better refine estimates of conditions on-the-ground; however, it can be challenging to combine these different data types due to differences in the scale or accuracy of each model. Scientific machine learning a type of artificial intelligence that continues to learn as more data is available, could provide a novel and flexible way of combining observations and models. This project aims to develop a data-driven modeling framework that can help [...]
Summary
As climate change is impacting water resources and aquatic ecosystems, there is a great need for natural resource managers to assess adaptation measures in a holistic manner. This can be done by integrating model predictions of climate, hydrology, and ecosystems with observational data to better refine estimates of conditions on-the-ground; however, it can be challenging to combine these different data types due to differences in the scale or accuracy of each model. Scientific machine learning a type of artificial intelligence that continues to learn as more data is available, could provide a novel and flexible way of combining observations and models.
This project aims to develop a data-driven modeling framework that can help stakeholders incorporate the impacts of climate change into future predictions of aquatic flows. The proposed team will convene a working group composed of local, state, and federal resource managers who are responsible for mitigating the impacts of climate change on fish and wildlife. Informed by this working group, researchers will develop the data-driven framework by adopting modern computational methods (including scientific machine learning) that facilitate communication and ensure that the users of this framework are involved with its development from the ground up.
The focus area of the proposed research will be the Northeast U.S., although the modeling tool is robust and can be applied to any region of the United States. Detailed user documentation and a simple interface will allow resource managers to incorporate the modeling tool into their existing workflow, making its adoption more likely. The overarching goal of this research is to bring the very promising area of scientific machine learning into the hands of interested users and resource managers, ensuring fair access to such tools and demonstrating a different paradigm for climate change science.