湖冰物候是反映区域气候变化的直观指标。由于青藏高原湖冰物候的地面观测不足，遥感与模拟成为动态监测湖冰物候变化并揭示其变化机理的重要途径。本文以纳木错为例，通过不同遥感方法获取了纳木错2000年—2015年湖冰物候的动态变化。在此基础上，将遥感与物理基础清晰的湖泊过程模型相结合，重建了纳木错1963年—2018年的湖冰物候序列，并分析了近60 a来纳木错湖冰物候的变化规律。研究结果表明，气候变暖影响下，湖冰融化日期显著提前，纳木错湖泊的冰期以6.4 d/（10 a）的速率显著缩短。未来气温升高2 ℃的情景下，湖冰融化日期平均可提前12.4 d。

Lake ice phenology refers to the dates of lake freeze-up and break-up and period of ice cover; it is considered a valuable indicator of regional climate change. The shifts of lake ice phenology in association with a warming climate is widely interesting because it not only serves as evidence of the changes in climate but could show substantial impacts on regional hydrological processes and the aquatic ecosystem. Ground-based records of lake ice phenology over the Tibetan Plateau are limited because of the harsh geographical conditions and the high observation costs. Satellite-based observation and modeling are expected to be effective in investigating the long-term changes in lake ice phenology for regions with poor ground observations. We aim to reconstruct the lake ice phenology time series and to identify the long-term changes of lake ice phenology in responding to the climate of Nam Co Lake at the Tibetan Plateau and for the past 60 years based on a process-based model, where remotely sensed lake surface water temperature is used to calibrated the process-based model.The research framework includes retrieving lake surface water temperature and lake ice phenology information from remotely sensed data, calibrating the process-based model against the remotely sensed lake surface water temperature, determining lake ice phenology according to the simulated water temperature, validating the simulated lake ice phenology by comparing against that derived from the remotely sensed data, detecting the long-term trends in the reconstructed lake ice phenology, and modeling the response of lake ice phenology to changes in air temperature. Four different remotely sensed datasets and the corresponding approaches are used to retrieve lake ice phenology of the Nam Co for the period 2000—2015. The process-based model (LAKE 2.3) is a 1D lake surface energy balance model. It is used to reconstruct lake ice phenology of Nam Co for the period 1963 to 2018 and investigate the sensitivity of lake ice phenology to climate change. The Mann–Kendall nonparametric statistical test approach is used in detecting the trend of lake ice phenology.Lake ice phenology derived using different remotely sensed data and approaches with consistency in the trend but with considerable uncertainties due to the temporal and spatial resolution of the sensors. The reconstructed lake ice breaking-up date based on the model is more comparable to that remotely sensed data than the other lake ice phenology indicators. The reconstructed time series of lake ice phenology shows that, during the previous 57 years, the freezing-up date was significantly delayed whereas the breaking-up date was earlier, thereby resulting in a shortened ice cover duration. The ice cover duration is shortened at a rate of 6.4 days/10a during the period 1963 to 2018. Sensitivity analysis shows that the breaking-up date would be significantly earlier in a warm climate. Under the 2 °C warmer scenario, the breaking-up date would be 12.4 days earlier on the average, and the ice cover duration would be shortened by 19.7 days, on the average.This study combines the strengths of remote sensing and numerical modeling in forming a novel research framework to reconstruct lake ice phenology of regions with poor ground-observation, such as the Tibetan Plateau. The results show that the framework is reliable and valuable to explore the long-term changes in lake ice phenology and its response to climate change. However, uncertainties exist in the remotely sensed lake ice phenology and the numerical modeling, which needs to be improved and further validated where or when ground-based observations are available.