<p>Projections of future climate implicate the thickening of debris-covered glaciers. Thus it becomes<br>increasingly important to consider the influences of the debris layer on glacier melt especially<br>because of the strong non-linear relationship between debris thickness and ablation rate. This<br>relationship is known as the Østrem curve. For very thin debris layers ablation is increased due to a<br>higher albedo where as thicker debris layers reduce ablation due to insulation. Some local scale<br>models based on satellite images already partly account for this effect. They take the average debris<br>thickness per decametre large grid cells and scale the ablation based on a representative Østrem<br>curve. Unfortunately, supraglacial debris thickness has been shown to vary substantially over short<br>length scales on a glacier surface. The debris thickness frequency distribution is also non-linear, with<br>varying skew and kurtosis. Therefore sub-debris ice ablation is a result of the combined non-<br>linearities of the particular Østrem curve with the particular frequency distribution of the debris<br>thickness.<br>In this study, we investigate the error of the ablation rate introduced by neglecting debris frequency<br>distribution. We obtained the Østrem curves from field measurements and different numerical<br>models. A gamma distribution with different rate and shape parameters was used to represent likely<br>debris thickness frequency distribution. The analysis shows that the debris frequency distribution<br>has a strong effect on calculated glacier melt. For some cases, this can result in a doubling or<br>tripling of ablation values in comparison to that calculated using the mean debris thickness.<br>Therefore neglecting this effect in numerical models has the potential to drastically underestimate<br>glacier melt.<br>Within the framework of this study, an interactive application was developed that allows one to<br>investigate the error on ice ablation if not considering the debris layer's thickness frequency<br>distribution.</p>