Monte Carlo experiments for uncertainty investigation of glacier melt discharge predictions through surface energy balance analysis

The spatial representativeness of point records is a concern in glacier discharge predictions. A Monte Carlo-based global sensitivity approach is used to investigate the predictive uncertainty in the net radiation (R _n) as the major component driving glacier melt in the Bolivian Andes. The R _n is inferred through the Surface and Energy Balance Algorithm, calibrated with point dry-season records monitored on a glacier's ablation area. High uncertainties are expected in the vicinity of the monitoring station (surface albedo (α) between 0.81 and 0.79, specific melt discharge (SMD) between 72 and 88 L s ^-1 km ^-2); smaller uncertainties are expected on the glacier boundaries (α between 0.10 and 0.08, SMD between 128 and 143 L s ^-1 km ^-2). Thus, with the incoming long wave radiation (R _L↓) as the most sensitive model parameter, the spatial variability in α determines the spatial variability in the SMD predictive uncertainties.