Assessment of glacier melt-model transferability: comparison of temperature-index and energy-balance models
Abstract. Transferability of glacier melt models is necessary for reliable projections of melt over large glacierized regions and over long time-scales. The transferability of such models has been examined for individual model types, but inter-comparison has been hindered by the diversity of validation statistics used to quantify transferability. We apply four common types of melt models – the classical degree-day model, an enhanced temperature-index model, a simplified energy-balance model and a full energy-balance model – to two glaciers in the same small mountain range. The transferability of each model is examined in space and over two melt seasons. We find that the full energy balance model is consistently the most transferable, with deviations in estimated glacier-wide surface ablation of ≤ 35% when the model is forced with parameters derived from the other glacier and/or melt season. The other three models have deviations in glacier-wide surface ablation of ≥ 100% under the same forcings. In addition, we find that there is no simple relationship between model complexity and model transferability.