Climate change is causing rapid warming and altered precipitation patterns in mountain watersheds, both of which influence the timing of ice breakup in mountain lakes. To enable predictions of ice breakup in the future, we analyzed a dataset of mountain lake ice breakup dates derived from remote sensing and historical downscaled climate data. We evaluated drivers of ice breakup, constructed a predictive statistical model, and developed projections of mountain lake ice breakup date with global climate models. Using Random Forest analysis, we determined that winter and spring cumulative snow fraction (portion of precipitation falling as snow) and air temperature are the strongest predictors of ice breakup on mountain lakes. Interactions between precipitation, cumulative winter air temperature and lake surface area indicate that shifts in air temperature and precipitation affect smaller lakes (< 2 km²) more than larger lakes (> 2–10 km²). A linear mixed effects model (RMSE of 18 d), applied with an ensemble of 15 global climate models, projected that end‐of‐century ice breakup in mountain lakes will be earlier by 25 ± 4 and 61 ± 5 (mean ± SE) days for representative concentration pathways 4.5 and 8.5, respectively.

中文翻译：

---

美国西部山区湖泊中冰物候的驱动因素和预测

气候变化导致山区流域迅速变暖并改变了降水方式，这两者都会影响山区湖泊结冰的时间。为了能够对未来的冰破裂进行预测，我们分析了从遥感和历史降尺度的气候数据中得出的高山湖泊冰破裂日期的数据集。我们评估了冰破裂的驱动因素，构建了预测性统计模型，并利用全球气候模型开发了高山湖泊冰破裂日期的预测。使用随机森林分析，我们确定了冬季和春季的积雪分数（降水量随雪而下降的比例）和气温是高山湖泊冰破裂的最强预测因子。降水之间的相互作用²）大于较大的湖泊（> 2-10 km ²）。线性混合效应模型（RMSE为18 d）与15个全球气候模型结合使用，预计到世纪末，高山湖泊的冰破裂将提前25±4和61±5（平均±SE）代表性浓度途径4.5和8.5天。