ABSTRACT
Mountain shadows in optical satellite images complicate the mapping of glacial lakes. Due to the rugged topography in periglacial alpine regions, many glacial lakes, especially smaller lakes, are partially shaded by mountain shadows in remotely sensed images. Shadows not only reduce the accuracy of lake mapping but also make changes in lake area hard to detect. In this paper, the characteristics of mountain shadows in remotely sensed imagery are explored, and their spatial relationships with regards to glacial lakes are modelled. Building on the previously developed Glacial Lakes Iterative Local Mapping (GLILM) method, a new water mapping approach is presented. The new method utilizes log-transformed spectral data and a normalized difference water index, NDWIblue, for delineating the boundaries of lakes within shadowed regions. The application of this approach is explored within the context of mapping lakes across space and time using Landsat images in the glacially dominated Tianshan mountainous of Central Asia. The results demonstrate that glacial lakes, both in sunlit and in shaded areas, can be mapped reliably, and that the results are useful for lake change analysis studies.
Acknowledgments
This study was supported in part by National Key R&D Program of China (Grant No. 2017YFB0504204), National Natural Science Foundation of China (Grant No. 41671034, 41101041, U1178302), Youth council for the promotion of innovation of the Chinese Academy of Sciences (Grant No. 2015357) and China Scholarship Council (CSC No. 201604910024).
Disclosure statement
No potential conflict of interest was reported by the authors.