作者：Li, L., Long, D., Wang, Y. & Woolway, R.I.

Lakes are crucial for ecosystems1, greenhouse gas emissions2 and water resources3, yet their surface-extent dynamics, particularly seasonality, remain poorly understood at continental to global scales owing to limitations in satellite observations4,5. Although previous studies have focused on long-term changes6, 7-8, comprehensive assessments of seasonality have been constrained by trade-offs between spatial resolution and temporal resolution in single-source satellite data. Here we show that seasonality is the dominant driver of lake-surface-extent variations globally. By leveraging a deep-learning-based spatiotemporal fusion of MODIS and Landsat-based datasets, combined with high-performance computing, we achieved monthly mapping of 1.4 million lakes (2001-2023). Our approach yielded basin-level median user's and producer's accuracies of 93% and 96%, respectively, when validated against the Global Surface Water dataset7. Seasonality-dominated lakes constitute 66% of the global lake area and approximately 60% of total lake counts, with over 90% of the world's population residing in regions where such lakes prevail. During seasonality-induced extreme events, the impacts can exceed the combined magnitude of 23-year long-term changes and regular seasonal variations, doubling the contraction of 42% of shrinking lakes and fully offsetting the expansion of 45% of growing lakes. These results uncover previously hidden seasonal dynamics that are crucial for understanding hydrospheric responses to environmental changes9, protecting lacustrine systems10, 11-12 and improving global climate models13,14. Our findings underscore the importance of incorporating seasonality into future research and suggest that advancements in the fusion of multisource remote-sensing data offer a promising path forward.

湖泊对生态系统、温室气体排放和水资源至关重要，然而受限于卫星观测手段，在大陆至全球尺度上，其水域面积动态变化（尤其是季节性变化）仍缺乏深入认知。尽管已有研究聚焦于长期变化，但单一卫星数据源在空间分辨率与时间分辨率之间的固有矛盾，限制了对湖泊季节性变化的全面评估。本研究表明，季节性是全球湖泊水域面积变化的主导驱动因素。通过基于深度学习的时空融合方法，整合中分辨率成像光谱仪（MODIS）与陆地卫星（Landsat）数据集，并结合高性能计算技术，我们实现了 2001—2023 年 140 万个湖泊的逐月制图。以全球地表水数据集为验证数据，本方法在流域尺度上的用户精度与生产者精度中位数分别达到 93% 和 96%。季节性主导型湖泊占全球湖泊总面积的 66%，约占湖泊总数的 60%，全球超过 90% 的人口居住在这类湖泊广泛分布的区域。在季节性引发的极端事件中，其影响幅度可超过 23 年长期变化与常规季节变化的总和：使 42% 处于萎缩状态的湖泊收缩面积翻倍，完全抵消 45% 处于扩张状态湖泊的增长趋势。这些发现揭示了此前未被发现的湖泊季节性动态，这对于理解水圈对环境变化的响应、保护湖泊生态系统以及优化全球气候模型具有关键意义。研究结果强调将季节性特征纳入未来相关研究的重要性，并表明多源遥感数据融合技术的发展是极具前景的研究方向。

（来源：Nature 2025 Issue 8067 DOI: 10.1038/s41586-025-09046-3）