A regime shift in great lakes nearshore total freeboard dynamics driven by synchronized climate patterns

• First 75-year Great Lakes nearshore total freeboard reconstruction (1950–2025) using GNSS-IR and Random Forest deep learning. • In-situ data reveal winter water levels and lake ice decouple seasonally but exhibit lagged interannual correlation. • A regime shift around 2011–2012 quadrupled ∼4-year freeboard oscillations, strengthening synchronization with the EP-NP, NAO, and PDO. The nearshore total freeboard (NTF) of the Great Lakes serves as a sentinel of regional climate change, yet their role in capturing short-term climate signals has been largely overlooked. Here, we employ a Random Forest-based model to reconstruct NTF heights from 1950 to 2025, leveraging observations derived from ground-based GNSS Interferometric Reflectometry (GNSS-IR) signal of opportunity technique. Our reconstruction reveals a significant long-term thinning trend of -2.5 mm/decade since 1978. More importantly, we identify a dominant 4-year periodic oscillation in NTF heights, accounting for over 60% of the data variance. Strikingly, modulated by the synergistic enhancement of the East Pacific-North Pacific, North Atlantic Oscillation, and Pacific Decadal Oscillation patterns, the amplitude of this oscillation increased by 400% in 2011–2024 (12.6 ± 1.7 mm) relative to that in 1950–1980 (2.5 ± 0.7 mm). Our findings provide robust evidence that the winter lake ice-snow system is becoming increasingly vulnerable to climate change.