Glacier Lake Outburst Floods (GLOFs)—Mapping the Hazard of a Threat to High Asia and Beyond - Impact of Global Changes on Mountains: Responses and Adaptation - page 339

separately. The fi rst and crucial step for the suggested scheme is the ordering
of the variables after the estimated hazard potential from the highest to the
lowest. Knowledge, e.g., from literature, about the general setting and past
GLOF events (Table 14.1) can be considered for this step. Then, the weights
are linearly distributed while the 2nd lowest weight is two times the lowest
weight, the 3rd lowest is the sum of the 2nd lowest plus the lowest weight
and so on. The sum of the weighting factor is by default set to 1 (Table 14.4).
The variables which are applied for each lake are then multiplied with
the weighting factor and subsequently added up. Thereafter, a total of nine
parameters are included and modelled in a GIS environment for the current
situation, and two additional parameters (glacier slope at terminus <5° and
stagnant terminus) are also included which indicate whether the glacial lake
may continue to grow in the near future (Bolch et al. 2011a).
In most cases the characteristics of the moraine dams (width, height,
freeboard) can only be addressed visually while other critical measures of
dam stability like material composition or piping can hardly be addressed
from remote sensing.
The fi nal classifi cation of PDGLs can be established by the defi nition
of qualitative threshold values ranging from very low potential danger
to high danger. A lake should have a very low hazard potential only if no
or only one factor with low weight applies to the lake. Hence, 0.1 may be
taken as the fi rst threshold. Bolch et al. (2011a) considered that a lake can
be of potentially high danger if the four most important factors apply to
the lake or a combination of several factors reaching the sum of the weights
of the four most important factors. The threshold between the low and
medium potentially dangerous glacial lakes is suggested to be the mean
value between class 1 and 4. The weighting and classifi cation scheme may
be evaluated based on visual interpretation of the morphometric variables
in satellite imagery of selected case studies, previous GLOF events and
possible knowledge from fi eld visits.
Discussion
An approach which is based on remote sensing analysis and modelling can
be successfully applied to identify potentially dangerous glacial lakes. It is
certainly suitable for a fi rst comprehensive assessment of PDGLs for a larger
area and addresses levels 1 and 2 of the approach suggested by Huggel et al.
(2002). Bolch et al. (2011a) combined the manifold conditioning parameters
which had not been addressed previously in such a comprehensive way.
Their approach is easily reproducible as it is based on well-developed
methods such as the detection of water bodies using multispectral imagery
(Huggel et al. 2002), the automated detection of glaciers (Paul et al.2002,
Bolch and Kamp 2006), and their velocities (Bolch et al. 2008, Kääb 2005) and