Reading Snow with Thermometers: A New Way to Track Winter Water

What this episode covers

Snow is more than winter scenery. It is a slow-release water reservoir, a blanket over soils and tree roots, and a clue to flood risk when warm spells or rain-on-snow events arrive. This episode follows researchers in New York’s Adirondack Mountains who asked a practical question: can simple temperature sensors, paired with machine learning, tell us how deep the snow is when no one is there to measure it?We unpack how snow acts like insulation, why temperature changes inside a snowpack can reveal its depth, and how field scientists used iButton sensors, PVC pipes, snow stakes, trail cameras, and random forest models to estimate snow depth across a forested watershed. The result: errors as low as about 1.8 to 6.5 centimeters at trained sites, with larger uncertainty when applying the model to a new site. We also talk about bear-damaged cameras, melting midwinter snow, forest canopy effects, and why better snow monitoring matters for streams, forests, water supply, and climate adaptation.Citation: Gunn, Madison, James S. Mills, Michael Mahoney, Colin Beier, Tao Wen, and Samuel E. Tuttle. 2025. “A Machine Learning Approach for Snow Depth Estimation From Temperature Sensors.” Hydrological Processes 39: e70273. https://doi.org/10.1002/hyp.70273Disclosure: This Waterlines episode uses AI-generated voices for the hosts.

PodParley-generated summary based on available episode metadata and transcript content.