Modeled River Ice Thickness and Phenology, Alaska and Yukon, Canada, 1990-2021

Climate change thinning river ice and shortening the duration it is present. With limited in-situ observations in Alaska, modeled river ice data fills a critical gap in our understanding of these processes across temporal and geographic scales. We constructed river ice models using statistical, machine learning, and remote sensing techniques to assess river ice presence and thickness across Alaska and the Yukon River basin. This dataset was created using these river ice phenology and thickness models. The models were run with inputs from a regional climate model and Sentinel-1 backscatter. A machine learning river ice phenology model produced break-up and freeze-up dates for 34,208 segments in Alaska and Yukon, Canada from 1990 to 2021. Combining statistical, machine learning, and remote sensing methods yielded ice thickness data for 8,792 river segments from 2016 to 2021.