Abstract
As Himalayan glaciers recede in response to climate amelioration, the size and number of moraine-dammed lakes is increasing dramatically. Many lakes are currently developing behind unstable moraines and have the potential to failcatastrophically, producing devastating Glacial Lake Outburst Floods (GLOFs). There is an urgent need to assess the various factors which control styles of moraine dam-breach and to link failure scenarios to their downstream impacts.
This study focuses on the reconstruction of palaeoGLOFs from the Dig Tsho and Chukhung moraine-dammed lake complexes (Khumbu Himal, Nepal Himalaya). This research includes the pioneering application of an effective, low-cost
photogrammetric method, termed ‘Structure-from-Motion’, for reconstructing moraine and valley floor topography. Benchmarking of the technique revealed it to be capable of reconstructing complex topography at spatial resolutions and
accuracies comparable to terrestrial laser scanning. It was subsequently implemented successfully at the two Himalayan field sites. Metric data pertaining to moraine dam and lake basin geometry were extracted from high-resolution
digital terrain models and served to constrain the initial boundary conditions of an advanced numerical dam-breach model. Generalised Likelihood Uncertainty Estimation (GLUE) was used to quantify the extent to which uncertainties surrounding dam-breach model parameterisation are translated into equifinal model output. Observed breach geometries were replicated by a range of unique parameter ensembles, associated with different modes of breach initiation and reflecting a range of output hydrographs. Material roughness was discovered to exert the dominant influence over hydrograph form. Factors influencing GLOF inundation extents and wetting front travel times were investigated using 2-D hydrodynamic modelling, with inundation patterns found to be highly sensitive to grid resolution. A GLUE-based method for probabilistic GLOF hazard mapping was developed, which represents a significant improvement over traditional, deterministic approaches and can be utilised within a unified framework for cascading parameter uncertainty through the GLOF modelling chain.
| Date of Award | 2013 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Neil Glasser (Supervisor), James Brasington (Supervisor) & Michael Hambrey (Supervisor) |