| Popis: |
The rapid development and instability of moraine-dammed proglacial lakes is increasing the potential for the occurrence of catastrophic Glacial Lake Outburst Floods (GLOFs) in high-mountain regions. Advanced, physically-based numerical dam-breach models represent an improvement over existing methods for the derivation of breach outflow hydrographs. However, significant uncertainty surrounds the initial parameterisation of such models, and remains largely unexplored. We use a unique combination of numerical dam-breach and two-dimensional hydrodynamic modelling, employed with a Generalised Likelihood Uncertainty Estimation (GLUE) framework to quantify the degree of equifinality in dam-breach model output for the reconstruction of the failure of Dig Tsho, Nepal. Monte Carlo analysis was used to sample the model parameter space, and morphological descriptors of the moraine breach were used to evaluate model performance. Equifinal breach morphologies were produced by parameter ensembles associated with differing breach initiation mechanisms, including overtopping waves and mechanical failure of the dam face. The material roughness coefficient was discovered to exert a dominant influence over model performance. Percentile breach hydrographs derived from cumulative distribution function hydrograph data under- or overestimated total hydrograph volume and were deemed to be inappropriate for input to hydrodynamic modelling. Our results support the use of a Total Variation Diminishing solver for outburst flood modelling, which was found to be largely free of numerical instability and flow oscillation. Routing of scenario-specific optimal breach hydrographs revealed prominent differences in the timing and extent of inundation. A GLUE-based method for constructing likelihood-weighted maps of GLOF inundation extent, flow depth, and hazard is presented, and represents an effective tool for communicating uncertainty and equifinality in GLOF hazard assessment. However, future research should focus on the utility of the approach for predictive, as opposed to reconstructive GLOF modelling. |
