The authors have provided a well-formulated and complete revision of their manuscript. Most of the comments raised by the reviewers were addressed and new analysis has been added to the paper. The article presents the method (glacier retreat scheme for lumped hydrological models) in a clear and reproducible way and will be helpful to the community. Nevertheless, there are some (mostly minor) issues with the present version of the article that should be addressed before final acceptance.
- Page 5, line 15: The reasoning is somewhat unclear here. Obviously, the authors have a complete snow redistribution model available (is this described in Freudiger et al, 2017? – the text seems to provide this hint) but decide not to use it and go for an extremely simple approach. I would avoid mentioning the “other” approach for removing “snow towers” and just describe what has been done.
- Page 5, line 19: Please shortly explain how the parameters were chosen and if there was any possibility to validate them.
- Page 5, line 25: The “lookup table” seems to be one of the major modifications in comparison to the original dh-parameterisation. It is mentioned for the first time here but unfortunately it remains highly unclear what this table (1) actually is and does, and (2) how it is derived. Regarding (1): This is clarified later in the paper (with the aid of the figure). However, it should be outlined here. Regarding (2): The formulation calls for better explanation. “… a lookup table can be derived from any glaciological model”. So, another model (a glaciological model) is needed to generate this lookup table? Which model was used by the authors? Apparently, this might quite drastically limit the applicability of the authors’ proposed model implementation as not all hydrologists might have a glaciological model at hand to produce this lookup table. Maybe it is just a misunderstanding that can be clarified with a better formulation?
- Page 10, line 28: The area change 1900-2006 is correctly calculated by the full model. However, the two individual periods 1900-1940, and 1940-1973 are completely off. This should also be mentioned. I suggest providing the results for relative area change (observed / simulated) for all four experiments in a table. This would better allow tracking how well the model implementations work.
- Page 11, line 9: A general comment on “glacier advance”. The authors directly link increases in glacier mass to advances of the glacier front. They do not consider a temporal delay. This strongly contradicts observations: Several years or even decades of mass gain are required (depending on glacier size and shape) until ice flow dynamics have changed in a way to make the glacier snout advance. This is completely neglected in the presented approach. I fully understand that this is not feasible, and probably also not necessary (!), for such a model, but this effect should be critically discussed by the authors and formulations should be adapted throughout the paper to not imply equality of glacier mass gain and glacier advance.
- Page 12, line 22: This statement should be corrected. Modern remote sensing data provide elevation changes for virtually all glaciers in the world (see e.g. Brun et al., 2017, NGEO). Glacier-specific elevation changes are also available for all glaciers in Switzerland, for example (Fischer et al., 2015, The Cryosphere).
- Figure 1: Nice! However, I strongly recommend enlarging text size (and reduce unfilled white spaces) to make the figure readable more easily.
- Figure 3c: Maybe show relative instead of absolute errors? This would be more informative. Why is there a difference for the 1900 surface?
- Figure 4a: Why do not all runs start with the same area? Also 1900 should have a dot for an observed area. The dots should have error bars (!) and might be linked with lines to make the figure clearer.
- Figure 4b: This is not observed volume! A well-constrained bedrock topography from the dataset of M. Huss is available for 1973 and 2010 but the volumes for 1940 and 1900 are derived from volume-area scaling, i.e. an extremely simple and highly uncertain model. So, it cannot be used for model validation! As already suggested in my last review, there are actual measurements of ice volume change for the respective glaciers (Fischer et al., 2015). They do not refer to the entire study period but nevertheless would allow validation over several decades.
- Figure 4c: Please change y-axis label. Glacier runoff is normally interpreted as the water leaving the glacier. These water volumes are much higher as they also comprise seasonal snow melt on the glacier surfaces. The authors show cumulative glacier mass loss here, and not runoff. The legend of Figure 4 should be positioned more prominently – I first didn’t find it. Furthermore, I would suggest to not use model abbreviations but more intuitive names.
- Page 26, line 18: Combining volume-area scaling with the ice thickness distribution data set derived by a different methodology seems to be inconsistent, leading to a non-continuous evolution of glacier volume. In any case, the derived volumes for 1900 and 1940 using this method will not be suitable to validate the model regarding glacier mass change (Fig. 4b)! A better discussion is needed. |