Geoscience Reference
In-Depth Information
Tien Shan as an example. According to this paper and in part adapted on the
basis of ICIMOD (2011), in the following subsections the above mentioned
determining parameters will be described.
Glacier lake mapping
Water bodies can be in principle be identifi ed well using remote sensing
imagery. For this purpose several methods like ratioing and the Normalised
Differential Water Index (NDWI) with different band combinations (e.g.,
Blue, Green, NIR, SWIR) may be used (Huggel et al. 2002, Bolch et al.
2008, 2011a). One of the biggest challenges when mapping glacial lakes is
to obtain a highly accurate delineation of lakes with (partial) ice coverage
and of turbid lakes with minimum classifi cation error. Optimum results
can be obtained using the NDWI approach by employing NIR and Green
bands (Green - NIR/Green + NIR). The water index using Blue mostly
performs better in the shaded areas while the index with Green shows
fewer problems with ice on the water bodies. The applied thresholds have,
however, to be adapted for each individual scene (Table 14.1). One major
Green Band
NIR Band
NDWI
Threshold
Raw Lake Pixels
Shadow Mask
Slope Mask
Improved Lake Pixels
Vectorization
z z
Manual
Improvement
Final
Lakes
Figure 14.4. Proposed optimised procedure for lake mapping based on multi-spectral remote
sensing data (Bolch et al. 2011a).
Search WWH ::




Custom Search