Author(s): Karl-Erich Lindenschmidt; Haishen Lu; Mingwen Liu; Yu Lin

Linked Author(s): Karl-Erich Lindenschmidt

Keywords: No keywords

Abstract: Ice-jam flooding (IJF) is a major hazard in cold-region rivers, influenced by climate change and reservoir regulation. This study presents an integrated probabilistic modeling framework to quantify IJF risk and evaluate the effects of regulated and naturalized flows under current and future conditions. The approach combines flow simulations from the RIVICE river ice dynamics model embedded in a Monte Carlo Analysis framework, with machine-learning algorithms, such as Long Short-Term Memory (LSTM) and interpretable stacking ensembles, to capture uncertainty in ice-jam formation and backwater staging. Hydrodynamic simulations with climate projections were incorporated to assess changes in ice processes and ice-jam flood severity due to natural/regulated flow and status quo/climate change scenarios. Results show that cascade reservoir regulation generally increases IJF backwater levels and inundation hazard for moderate events, while offering some mitigation for extreme floods. Frequency analysis indicates that regulated flows during breakup periods amplify discharge and water

DOI:

Year: 2026