基于Sentinel-1时序数据的山地冰川表面消融变化研究--以老虎沟12号冰川为例

Surface ablation changes of mountain glaciers based on Sentinel⁃1 time series data:Taking Laohugou Glacier No.12 as an example

冰川是影响气候变化的主要因素,合成孔径雷达(SAR)不仅具有较高空间分辨率和多极化等特点,而且对冰雪融化引起的介电常数变化较为敏感,已成为山地冰川表面消融变化监测的重要数据源。基于Sentinel-1时序数据,提出了一种综合应用多时相多极化SAR变化检测算法和Sigmoid函数的冰川表面消融监测方法。以祁连山老虎沟12号冰川为研究区,利用“中心线-圆”方法获得的冰川东、西支64个样区,通过各点2019-2020年124个时相的SAR后向散射系数变化特征确定了7个典型样区及2019和2020年冰川消融期;基于同轨道参考影像与多时相融雪影像对比,分别获得变化监测影像VV和VH极化后向散射系数,对其进行权重组合后提取湿雪像元;通过Sigmoid函数并结合湿雪平均海拔高度提取干雪,从而得到2年冰川消融期内16个时相的冰川表面干湿雪分布。利用2019年09月01日无人机数字正射影像(DOM)以及与Sentinel-1邻近日期5景Sentinel-2影像对提取结果进行精度验证,并结合冰川上气象站点气温和降水数据对误差结果进行深入分析。实验结果表明,该方法能够有效提取冰川表面干、湿雪分布,总体分类精度OA高达96%,Kappa系数高达0.84。老虎沟12号冰川消融期及覆盖各类型消融变化规律存在年际变化:2019年冰川的消融期为5月初至9月中下旬,2020年的消融期为6月中下旬至9月初;2019年从消融季开始,冰川表面干雪面积迅速减小,而湿雪与冰川冰面积在整个消融期均较大,但随着强降雪出现,干雪面积骤然上升,2020年冰川冰和湿雪面积变化明显,冰川冰面积持续增加,湿雪面积持续减小,而干雪面积在高海拔区趋于稳定。

Glaciers are main factors affecting climate change.SAR(synthetic aperture radar)not only has the characteristics of high spatial resolution and multi-polarization,but also is sensitive to changes in dielectric constant caused by ice and snow melting.It has become an important data source for monitoring the ablation changes of mountain glacier surfaces.Based on the Sentinel-1 time series data,this paper proposes a glacier surface ablation monitoring method based on multi-temporal and multipolar SAR change detection algorithm and Sigmoid function.Taking Laohugou No.12 glacier in Qilian Mountain as the research area,64 transects in the east and west branches of the glacier obtained by the“centerline-circle”method were firstly used,and 7 typical transects and the glacial retreat periods in 2019 and 2020 were determined by the SAR backscatter coefficient change characteristics of 124 time phases from 2019 to 2020.Then,based on the comparison of the reference image of the same track with the multi-temporal snowmelt image,the VV(vertical vertical)and VH(vertical horizontal)polarization backscattering coefficients of the change monitoring images were obtained,respectively,and the wet snow pixels were extracted after the weight combination was combined.Finally,the Sigmoid function was used to extract dry snow combined with the average altitude of wet snow,so as to obtain the distribution of dry and wet snow on the glacier surface in 16 time phases during the two-year glacial melting period.The accuracy of the extraction results was verified by using the UAV(Unmanned Aerial Vehicle)DOM(digital orthophoto map)on September 1,2019 and the five Sentinel-2 images of the adjacent date of Sentinel-1,and the error results were analyzed in depth based on the temperature and precipitation data of the meteorological site on the glacier.Experimental results show that this method can effectively extract the distribution of dry and wet snow on the glacier surface,and the overall classification accuracy OA(overall accuracy)is as high as 96%,and the kappa coefficient is as high as 0.84.There are interannual variations in the melting period and melting characteristics in Laohugou No.12 glacier.The melting period of glaciers in 2019 was from early May to mid-to-late September,and the melting period in 2020 was from mid-to-late June to early September.Starting from the melting season in 2019,the dry snow area on the glacier surface decreased rapidly,while the wet snow and glacial ice area were large during the entire melting period,but with the emergence of heavy snowfall,the dry snow area increased sharply.In 2020,the glacial ice and wet snow area changed significantly,the glacier ice area continued to increase,the wet snow area continued to decrease,and the dry snow area tended to be stable at high altitude.