青藏高原湖泊分布广泛， 且多为季节性冻结湖泊。在全球变暖的背景下， 青藏高原湖冰厚度及其物候发生着显著变化， 深刻影响着区域气候演变。然而目前对于高寒区湖冰厚度及其物候在气候学尺度上的演变特征认识还不甚清楚。因此， 本文利用青海湖下社水文站野外湖冰观测数据、 MODIS湖冰覆盖率数据集、 刚察气象站观测数据与CMFD数据， 结合湖冰准稳态分析模型， 研究了近40年（1979 - 2017年）青海湖湖冰厚度及其物候的演变特征。结果表明： 模拟的多年冰厚平均值为0.31 m， 与下社水文站实测值接近； 模型对湖冰融化结束时间的刻画较准确， 误差仅为0.07天， 开始结冰时间和结冰期长度的误差分别为5.60天和5.67天。1979 -2017年模拟的最大冰厚减小趋势与观测结果较为一致， 即每年冰厚减少0.003 m。模拟的1979 -2017年青海湖开始结冰时间延后（0.23 d·a^-1）， 融化结束时间提前（0.32 d·a^-1）， 结冰期长度缩短（1.02 d·a^-1）， 其中20世纪80年代结冰期缩短尤为明显（2.2 d·a^-1）。1979 -2017年青海湖结冰期内（12月至次年4月）向下长波辐射和气温（二者均呈现上升趋势）与平均冰厚以及最大冰厚之间存在显著的负相关关系， 向下短波辐射（呈现下降趋势）与最大冰厚以及平均冰厚之间存在显著的正相关关系。去趋势敏感性试验表明： 向下长波辐射、 气温、 向下短波辐射、 比湿是青海湖1979-2017年内平均冰厚和最大冰厚变率的主要驱动因子， 对平均冰厚变率分别贡献了42.08%， 40.93%， -36.99%和17.45%， 对最大冰厚变率分别贡献了44.48%， 44.68%， -34.77%和19.92%。所有气象驱动因子对二者贡献了83.40%和87.01%， 可以看出青海湖最大冰厚变率相较平均冰厚变率更容易受到气象条件的影响。本文研究结果对冰冻圈湖冰的长期演变趋势提供了认识， 为其他青藏高原湖泊冷季研究提供了参考依据。

Lakes widely distribute in the Qinghai-Xizang Plateau， and most of them are seasonally frozen lakes.Under the background of global warming， lake ice thickness and phenology are changing significantly， which has a profound impact on regional climate evolution.However， the evolution characteristics of ice thickness and phenology on the climatological scale are not well understood at present.Therefore， in this paper， the lake ice thickness and phenological evolution characteristics of Qinghai Lake during 1979 -2017 were studied by using the field lake ice observation data from Qinghai Lake Xiashe Hydrology Station， MODIS Lake ice coverage dataset， meteorological observation data from Gangcha Meteorological Station and CMFD， combined with a quasi-steady state model of lake ice.The results show that the simulated average ice thickness is 0.31 m， which is close to the measured value of Xiashe Hydrology Station.The error in modelling breaking-up end is only 0.07 days， and the errors of the freezing-up start and the ice duration are 5.60 days and 5.67 days， respectively.The simulated maximum ice thickness decreases from 1979 to 2017 is in good agreement with the observed trend， that is， the ice thickness decreases by 0.003 m per year.In the freezing periods from 1979 to 2017， the freezing-up start of Qinghai Lake is delayed （0.23 d·a^-1）， the breaking-up end is advanced （0.32 d·a^-1）， and the length of the ice duration is shortened （1.02 d·a^-1）， especially in the 1980s （2.2 d·a^-1）.During the freezing periods of Qinghai Lake from 1979 to 2017 （from December to April of the following year）， the downward longwave radiation and air temperature （both of which showed an increasing trend） have a significant negative correlation with the average ice thickness and the maximum ice thickness， while the downward shortwave radiation （which showed a decreasing trend） has a significant positive correlation with the maximum ice thickness and the average ice thickness.The detrending sensitivity test shows that： downward longwave radiation， air temperature， downward shortwave radiation and specific humidity are the main driving factors of mean ice thickness and maximum ice thickness variability in Qinghai Lake from 1979 to 2017， contributing 42.08%， 40.93%， -36.99% and 17.45% to mean ice thickness variability， and 44.48%， 44.68%， -34.77% and 19.92% to maximum ice thickness variability， respectively.All the meteorological driving factors contribute 83.40% and 87.01% to the two factors.It can be seen that the maximum ice thickness variability of Qinghai Lake is more susceptible to the influence of meteorological conditions than the average ice thickness variability.The results of this study provide an understanding of the long-term evolution trend of lake ice in the cryosphere， and provide a reference for the study of other lakes in the Tibetan Plateau in the cold season.