Remote Sensing Monitoring Of Lake Ice Phenology In Qinghai-Tibet Plateau

Lake ice is sensitive to climate change,and its growth and melting processes,as well as its thickness,are affected by global warming.However,lake ice is still one of the relatively poorly studied components of the cryosphere.High-altitude lakes are less affected by human activities and can accurately record the changes of climate in the region where the lake is located and even globally.The large number and wide distribution of lakes on the Qinghai-Tibet Plateau make it an ideal area to study the response of lakes to climate change.Remote sensing technology has achieved the transformation of spatial scale of phenological observations from"point"to"area",making it the most suitable tool for monitoring lake ice changes.Conducting research on lake ice phenology in the Qinghai-Tibet Plateau using remote sensing technology can help to understand the spatial and temporal differentiation patterns of lake ice phenology in the region,and provide a cognitive basis for exploring the response mechanisms of lake ice phenology to climate change in the Qinghai-Tibet Plateau.In this study,we extracted long time series lake ice phenology dates of 40 lakes with areas about 200 km~2and above on the Qinghai-Tibet Plateau from 2000 to 2022,based on MODIS MOD09GA data and passive microwave remote sensing images including AMSR-E and AMSR2.And then,by analyzing the correlation and differences among results obtained from different data sources,a higher-precision dataset of lake ice phenology on the Qinghai-Tibet Plateau was constructed by integrating the advantages of multiple remote sensing images.Based on this dataset,we explored the spatial and temporal variation patterns of lake ice phenology in recent years.The results showed that:(1)When extracting the lake ice phenology from MODIS MOD09GA remote sensing images in the Qinghai-Tibet Plateau,the freeze-up start,freeze-up end,break-up start and break-up end dates of 40 lakes are both affected by cloud cover and other image quality issues,resulting in errors.Over the past 22 years,the average number of errors for these lakes has been 6 times,with an average duration of 2 days,2 days,3 days,and2 days for errors during the four phenological periods.There is no significant correlation between the number and duration of errors and the lake area or geographic location.It is essential to eliminate the influence of cloud cover and other factors when extracting lake ice phenology parameters based on optical remote sensing images to improve the accuracy of the phenology results.(2)The extracted result of phenological information for 30 lakes from 2002 to 2022,based on remote sensing images from AMSR-E and AMSR2,shows a strong correlation with results obtained from optical remote sensing data.Notably,the correlation between the extract parameters of different data sources is higher during the melting process than during the freezing process.The freeze-up start and complete freezing dates extracted based on passive microwave remote sensing are usually earlier than those extracted from optical remote sensing data,while the start of ablation and complete ablation dates extracted from the former are usually later than those extracted from the latter,with the mean difference of-5,-6,4 and 1 days,respectively.(3)Based on the high correlation between the lake ice phenology dataset v1.0extracted from optical remote sensing images and the phenology dataset v2.0 extracted from passive microwave data,the data were classified into two categories(correct and to be revised)by the error statistics results.The correlation and difference values were then calculated between the error-free dates selected in in the dataset v1.0 and the corresponding dates in dataset v2.0.For the to-be-revised dates,corrections were made based on the mean difference and the results of dataset v2.0 after determining the correlation.During the 22-years,the ice phenology dates of 19,16,15,and 19 lakes were revised for the four phenological periods,with 34,21,27,and 32 phenology dates revised,respectively.(4)On the Tibetan Plateau,lake freeze-up start and complete freezing events occur most frequently in November and December.Among them,the starting freezing date of lakes shows a more pronounced variation with latitude,with earlier freezing observed in the northern part of the plateau.And the onset of freeze-up also tends to be consistent with the distribution of lake altitude.The lakes on the Qinghai-Tibet Plateau mainly start to ablate in April and May of the following year,and the complete ablation dates are mainly concentrated in April to June.The spatial distribution characteristics of the two are similar,but are generally opposite to the spatial distribution characteristics of the beginning of freezing.That is,starting from lakes with a later onset of melting such as Kekexili and the West Kunlun Mountains,the time of lake ablation gradually becomes earlier towards lower latitudes.(5)From 2000 to 2022,the four phenological dates of 40 large lakes on the Qinghai-Tibet Plateau,including the onset and completion of freezing and melting,generally showed a trend of delay.The direction and degree of the long-term changes in lake ice phenology vary in different regions.Among them,the number of lakes with delayed and advanced freezing in the Kekexili region was approximately equal,while the two parameters of the melting process generally showed a trend of delay,with a faster rate of delay.(6)Apart from Zonag Lake,which experienced abnormal changes in its area due to a breach,most lakes on the Tibetan Plateau are expanding,with six lakes having an area increase of over 200 km~2 and four lakes having a relative change rate of over 100%compared to their area in 2000.The expansion of lakes may have an impact on lake ice phenology events such as delayed freezing,but the correlation is not significant.