Early recognition of glacial lake hazards in the Himalaya using remote sensing datasets

摘要

Glacier recession in high-Himalayan catchments leads to the formation of moraine-dammed lakes on many debris-covered glacier tongues. Such lakes are hazardous to communities and infrastructure downstream because of their potential to breach catastrophically, and their early recognition is required if remedial efforts are to be timely and cost-effective. Whilst the development of supraglacial lakes is known to begin as a series of ponds that subsequently coalesce into a larger lake, the relationship between glacier dynamics and lake formation is not well understood. Using ERS-1 and ERS-2 Synthetic Aperture Radar (SAR) data, SPOT-5 optical imagery and historical aerial photography, information is presented on the dynamics and structure of glaciers in Tibet (China) and Nepal that drain the southern side of the Himalaya. Glacier velocity data derived from interferometry show that where lakes are developing on debris-covered tongues the ice is virtually stagnant (displacements < 5 m a~(-1)). Furthermore, elevation data from Digital Elevation Models (DEMs) derived from aerial photography and SPOT-5 HRS data reveal that supraglacial lake formation is prevalent where glacier surface gradients are less than 2° from the glacier terminus, supporting empirical observations from previous work. The resolution offered by the DEMs and SAR data allows variations in transverse glacier elevations and velocities to be detected, such that the pattern of lake development on an individual glacier can be identified. Whilst the glacier surface gradient provides the boundary conditions favourable for lake formation, local variations in glacier velocity and surface morphology between flow units control the precise location of lake growth. Integrating the surface gradient and velocity information into a single analysis highlights those glaciers that are particularly vulnerable to lake development over an expected decadal timescale. The wider application of these techniques, based on remote sensing data, is particularly suitable for 'first-pass' hazard assessments and for regions where field access is difficult due to severe terrain, political sensitivity or financial constraints.