CAREER: GLOW: Sequencing rivers with machine learning and bioinformatics

Rivers are the arteries of landscapes and are deeply embedded in economies and ecosystems worldwide. They are reliable sources of fresh water and energy; vital habitats for fish and other wildlife; conduits for transportation; and hotspots for hazards from pollution and flooding. A fundamental aspect of rivers is that they rarely flow in straight lines, and instead meander through river bends of different shapes. For example, some bends are symmetric like bell curves, while other bends skew to one side like bananas. These shapes are widely used in both theory and practice but are based on subjective interpretation, which complicates efforts to predict how rivers will respond to a changing environment. This work will adapt computing methods, like those used in speech recognition and in DNA sequencing, to test whether river shapes are divisible into a geometric “alphabet,” and if so, how these shapes are ordered in sequences from upstream to downstream. Through partnerships with experts in music composition and science education at the University of Virginia, the project will also translate the geometry of rivers to music to illuminate the role of creativity in research. This fusion of art and science will be shared with the local community in Charlottesville, VA, and nationally through teaching materials for geoscience educators. Dozens of theoretical models describe the shapes of rivers, yet major knowledge gaps persist in comparing complex river shapes between models and observations. The research will address this need by blending approaches from geoscience, machine learning, and bioinformatics to derive a new language of rivers. The work will analyze over 100,000 meander bends using unsupervised machine learning; apply methods inspired by DNA sequencing to river channels; and test whether bend geometries are sensitive to bank materials such as sediment, bedrock and vegetation. This Division of Earth Sciences – Geoscience Lessons for and from Other Worlds (GLOW) award will also examine roughly 1,000 channel bends on the Moon, Venus and Mars in comparison to the overall meander bend classification. The educational activities will create a new entry point to science through music. The work will also support the PI’s deepened participation in organizations that are at the forefront of building the future for the field. This project is co-funded by the Directorate for Geosciences to support AI/ML advancement in the geosciences. 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.