地球信息科学学报 ›› 2019, Vol. 21 ›› Issue (11): 1790-1801.doi: 10.12082/dqxxkx.2019.190204

• 遥感科学与应用技术 • 上一篇    下一篇

基于SAR干涉数据的东帕米尔高原冰川变化

郭唯娜, 柯长青*(), 范宇宾   

  1. 南京大学地理与海洋科学学院,南京 210023
  • 收稿日期:2019-04-30 修回日期:2019-09-18 出版日期:2019-10-25 发布日期:2019-12-11
  • 通讯作者: 柯长青 E-mail:kecq@nju.edu.cn
  • 作者简介:郭唯娜(1997-),女,陕西渭南人,硕士生,主要从事冰川变化研究。 E-mail: mg1827056@smail.nju.edu.cn
  • 基金资助:
    国家自然科学基金项目(No.41830105)

Glacier Changes on the East Pamirs Plateau: Based on SAR Interference Data

GUO Weina, KE Changqing*(), FAN Yubin   

  1. School of Geography and Ocean Science, Nanjing University, Nanjing 210023, China
  • Received:2019-04-30 Revised:2019-09-18 Online:2019-10-25 Published:2019-12-11
  • Contact: KE Changqing E-mail:kecq@nju.edu.cn
  • Supported by:
    National Natural Science Foundation of China(No.41830105)

摘要:

冰川变化监测对生态灾害预防、区域水资源调控、气候变化研究等意义重大。利用冰川在雷达干涉影像上表现出失相干这一特性,选用1998年ERS 1/2与2018年 Sentinel-1A重轨单视复数SAR数据,通过相干系数取阈值的方法获取东帕米尔高原两个时期的冰川边界,以Landsat TM/OLI影像和全球陆地冰川空间监测计划发布的数据验证本文冰川边界提取的精度,从而分析冰川变化。结果表明:① 拟合研究区相干系数图上相干系数γ与对应像元个数的曲线关系,冰川区像元个数会在低相干区域积累形成一个小的波峰。曲线一阶导数变缓的点(冰川区向非冰川区过渡的转折点)即为所选阈值点,利用SAR相干系数取阈值法提取的冰川边界与光学遥感影像结合RGI6.0数据提取的验证冰川边界具有较好的一致性,SAR干涉相干系数提取冰川边界的方法是可行而有效的,ERS 1/2与Sentinel-1A提取的冰川总面积精度均在90%以上,而且SAR数据能够有效提取光学遥感影像难以识别的冰川表碛覆盖;② 1998年和2018年东帕米尔高原冰川总面积减少了318.59 km2,年平均变化速率为-15.93 km2/a,冰川退缩面积占冰川总面积的23%;③ 对大、中型规模冰川来说,表碛覆盖型冰川退缩较其他冰川明显;从坡向上来看,20年各个坡向冰川均有所退缩,其中东南坡冰川退缩最多,西坡冰川退缩最少;从海拔上来看,1998年冰川集中分布在4519~5421 m海拔区间内,2018年集中分布在4682~5320 m海拔区间内;在3325~5710 m海拔区间内冰川退缩明显,4915 m海拔附近达到退缩极大值。

关键词: ERS1/2, Sentinel-1A, 相干系数, 表碛覆盖型冰川, 冰川边界, 东帕米尔高原, 坡向, 海拔

Abstract:

Glacier change monitoring is of great significance for ecological disaster prevention, regional water resource regulation, and climate change research. Based on glaciers' showing incoherence on radar interference imagery, 1998 ERS 1/2 and 2018 Sentinel-1A repeat orbit single-view complex SAR data were chosen to obtain the glacial boundary of the East Pamirs Plateau at the two periods, by taking the threshold of the coherence coefficient. The accuracy of the glacier boundary extraction was verified by Landsat TM/OLI and data published by GLMS, for further analysis of the glacier changes. Results show: (1) While fitting the curve relationship between the coherence coefficient γ and the number of corresponding pixels in the coherence coefficient diagram of the study area, the number of pixels in the glacial area accumulated in the low-value region of γ to form a small peak. The point where the first derivative of the curve becomes slower (the turning point of the transition from the glacier to the non-glacial region) was identified as the threshold point. The glacial boundary extracted by the SAR coherence coefficient threshold method was consistent with the result of the optical remote sensing imagery combined with the RGI6.0 data, suggesting that the proposed method based on SAR interference coherence coefficient for extracting the glacier boundary is feasible and effective. The glacier total area extraction accuracy of ERS 1/2 and Sentinel-1A was over 90%; moreover, SAR data can effectively extract glacier debris difficult to identify by optical remote sensing imagery. (2) The total area of glaciers on the East Pamirs Plateau decreased by 318.59 km 2 from 1998 to 2018, the annual average rate of change was -15.93 km 2/a, and the area of glacial retreat accounted for 23% of the total area of glaciers. (3) For large and medium-sized glaciers, the debris-covered glaciers retracted more than other glaciers. During the past 20 years, glaciers of all slopes have retreated. Glaciers in the southeast slopes retreated the most, and glaciers in the west slopes retreated the least. In 1998, glaciers concentrated in the 4519-5421 m altitude range; while in 2018, they concentrated in the 4682-5320 m altitude range. In the altitude range of 3325-5710 m, glaciers retreated obviously, and the maximum retraction value was at around 4915 m.

Key words: ERS1/2, Sentinel-1A, coherence coefficient, debris, glacier boundary, the East Pamirs, aspect, altitude