基于SAR数据的山地冰川表面运动速度分析

Analyzing Mountain Glacier Surface Velocities Using SAR Data

合成孔径雷达(SAR)因其可全天时、全天候工作且不受云、雨的影响而成为遥感应用的前沿领域。SAR干涉测量(Inteferometry)利用SAR数据的相位信息可获得大地表面厘米级的形变而成为冰川表面流速监测广泛使用的手段;SAR图像相关方法(SRFT)能克服干涉测量方法因失相干严重而难以产生清晰的干涉条纹以及可见光图像质量由于云遮、雪盖限制的不足而成为目前山地冰川表面流速遥感监测的首选方式。为深入探讨SAR图像相关方法的适用性,以天山科契卡尔巴西冰川为研究区域,分析使用不同时间基线的ALOS PALSAR数据与ENVISat ASAR C-band的图像相关方法估计冰川的表面流速,并使用实地测杆的DGPS(Differential GPS)测量流速进行对比验证,发现在冰川表碛覆盖区域使用图像相关方法测量的值与实测值有很好的一致性,而在裸露冰区域或坡度较大区域,误差比较大。比较长时间基线的SAR数据对特征识别的结果发现:时间基线为1a的冬季获取的数据对估计值与实测值在表碛覆盖区域比较一致,这可能是由于前后两次获取图像时天气或地面状况比较接近。比较ALOS PALSAR数据与ENVISat ASAR数据发现:波长较长的L-band(23.5cm)比C-band(5.7cm)SAR数据更适合山地冰川的表面流速估计;另外在运用SAR数据特征匹配方法时也可能是极化方式的差异使得ALOS PALSAR(HH极化)数据比ENVISat ASAR(VV极化)数据更适合冰川研究。

Synthetic Aperture Radar leads on the Remote sensing technology due to its excellent advantage such as independence of sunlight can work in all weather conditions and all day and all night,not easily affected by cloud and rain.Synthetic Aperture Radar data Feature-Tracking（SRFT） procedures can overcome the shortcomings of Interferometric Synthetic Aperture Radar（InSAR） due to decorrelation to obtain glacier surface velocities robustly.In order to discuss the suitability of the Feature-tracking procedures,To find the suitability of different SAR data sets to derive glacier surface velocity,SRFT of ALOS PALSAR Data and ENVISat ASAR data were employed and compared to obtain the Koxkar Baxi Glacier surface velocities.Data with different time intervals（temporal baseline） were used to calculate the glacier surface velocity.Which were validated by the in situ measurements.SRFT is a robust technology to derive mountain glaciers surface velocity while SAR interferometry may fail.The uncertainties were analyzed in the same time.We find that the velocities obtained in debris-covered area were well agreement with the in situ measurements.There exist much more uncertainties in the exposed ice regions where there are no apparent characteristics observed.And for long baseline datasets the weather conditions when SAR data acquired in the glacier region should be considered or lead to the occurrence of more errors.It＇s clear that L-band ALOS PALSAR data are more suitable for Feature-Tracking than ENVISat ASAR data in estimating mountain glaciers surface velocities due to the wavelength or polarization difference.