A number of methods exist to model the stages of a GLOF event. Dam-breach models can be categorised as being empirical, analytical or numerical in nature, with each method having significant advantages and shortcomings. Empirical relationships that produce estimates of peak discharge and time to peak are straightforward to implement, but the applicability of these models is often limited by the nature of the case study data from which they are derived. Furthermore, empirical models neglect the inclusion of basic hydraulic principles that describe the mechanics of breach formation. Analytical or parametric models simulate breach development using simplified versions of the physically based equations that describe breach enlargement, whilst complex, physically-based codes represent the state-of-the-art in numerical dam-breach modelling. To date, few of the latter have been applied to investigate the moraine-dam failure problem.
Despite significant advances in the physical complexity and availability of higher-order hydrodynamic solvers, the majority of published accounts that have attempted to reconstruct or predict GLOF characteristics have been limited, often by necessity, to the use of relatively simplistic models. This is in part attributable to the unavailability of terrain models of many high-mountain catchments at the fine spatial resolutions required for the effective application of numerically-sophisticated codes, and their proprietary (and often cost-prohibitive) nature. However, advanced models are experiencing increasing use in the glacial hazards literature. In particular, the suitability of emerging mesh-free, particle-based methods for simulating dam-breach and GLOF routing may represent a solution to many of the challenges associated with modelling this complex phenomenon.
Sources of uncertainty in the GLOF modelling chain have been identified by various workers. However, to date their significance for the robustness of reconstructive and predictive modelling efforts have been largely unexplored and quantified in detail. These sources include the geometric and material characterisation of moraine dam complexes, including lake bathymetry and the presence and extent of buried ice, initial conditions (freeboard, precise spillway dimensions), spatial discretisation of the down-valley domain, hydrodynamic model dimensionality and the dynamic coupling of successive components in the GLOF model cascade.