基于2009-2014年Landsat和PALSAR数据的藏东南冰川编目

A glacier inventory for the southeastern Qinghai-Tibet Plateau based on Landsat and PALSAR data(2009–2014)

本数据集是基于2011~2014年多时相Landsat光学影像、2009~2010年的L波段PALSAR雷达影像和改正的SRTM数字高程模型(DEM)得到的最新藏东南冰川目录。大致空间范围在28°N^31°N、93°E^97°E内,包括念青唐古拉山中部和东部,以及横断山西部,覆盖面积达11.5万平方公里。数据集内包括三个文件:1)定义研究区范围的矢量文件;2)冰川目录矢量文件;3)统计每条冰川特征的文档,参数包括GLIMS编号、冰川面积、最大和最小高程、平均高程、平均坡度、平均朝向、有无冰碛覆盖,以及冰碛覆盖面积等。为克服藏东南地区多云雨对光学影像的影响,对无冰碛覆盖冰川的提取采用了一种基于自动识别云和冰雪覆盖的方法,实现多景影像信息半自动融合;并将从光学影像中提取的地面信息和PALSAR雷达影像得到的相干图以及坡度结合起来,实现了冰碛冰川的半自动化单独提取。在后处理阶段,采用人工编辑提高数据精度:比如控制无冰碛覆盖冰川与其冰碛覆盖部分的连接,调整单条冰川界限,以及改正部分阴影、水体的影响。与人工数字化提取的冰川边界相比较,本数据集的冰川面积精度总体在3%以内。此编目为目前最新的冰川编目,有效地减小了云、季节性冰雪和冰碛覆盖对冰川范围识别的影响,可作为基础数据集,应用到各种冰川研究中。

This dataset provides the latest glacier inventory for the southeastern Qinghai-Tibet Plateau based on multi-temporal Landsat images acquired during 2011–2014,L-band PALSAR data obtained during 2009–2010,and the modified SRTM(Shuttle Radar Topography Mission)Digital Elevation Model(DEM).Within the spatial extent of 28°N–31°N and 92°E–97°E,the region covers the central and eastern Nyainqêntanglha and western Hengduan mountains,with a total area of around 115,000 km2.Three sub-datasets are provided:①extent of the study region in shapefile format,②glacier outlines in shapefile format,and③parameter statistics for each individual glacier.The parameters for each glacier include the glacier area,maximum and minimum elevation,mean elevation,mean slope,mean aspect,flag indicating presence or absence of debris cover,and area of debris-covered glaciers.To mitigate influences of cloud cover on optical images,we adopted semi-automated identification of cloud cover and clean snow/ice to combine observations from multi-temporal Landsat images.Debris-covered glaciers were separately determined with a semi-automated method that integrates information derived from Landsat images,coherence information generated from PALSAR(Phased Array type L-band Synthetic Aperture Radar)image pairs,and slope information.In the post-processing stage,we performed manual editing of the glacier outlines to improve the data quality,for example,to ensure seamless shared boundaries between clean-ice glaciers and their debris-covered parts,to adjust the individual ice divides,and to correct the effect of shadow and water bodies.Compared with manually digitized glacier outlines,this dataset has an overall error within 3%.The quality of the latest glacier inventory was improved;it has less uncertainty associated with cloud and seasonal snow cover in glacier maps and is a fundamental dataset for future glacier-related studies.