Item

Abstract

Glaciers supply freshwater to almost two billion people worldwide, and they are major contributors to twenty-first century sea level rise, it is therefore critical to project how much they may retreat under possible scenarios of future climate change. In recent years, various glacier evolution models have been used to generate these projections at a global-scale. This development has been facilitated by the use of simple mass balance models that rely heavily on parameterisations, often with poorly constrained parameters. These models are typically calibrated by tuning parameters in order to find the best agreement between modelled and observed mass balance. Though intuitive, this calibration procedure risks an ‘over-tuning’ model parameters, ultimately resulting in an underestimation of parametric uncertainty in future projections of glacier mass change. In this study, we use a simple mass balance model combined with volume-area scaling to generate projections of global glacier mass change under different shared socioeconomic pathways. For model calibration, we use geodetic mass balance observations for each Randolph Glacier Inventory (RGI) region. Rather than tuning parameters to obtain a single ‘best’ set, we use a maximin Latin Hypercube design to generate an ensemble of parameter combinations within multi-dimensional parameter space. Parameter space is constrained by filtering ensemble members that are inconsistent with observations within defined limits of plausibility. For each RGI region, plausible parameter sets are used to generate a perturbed-parameter ensemble of glacier mass change. These ensembles are used to quantify the sensitivity of projections to total parametric uncertainty as well as individual model parameters.