The Paper has been improved and I think the paper could be published after a few more improvements. I am not fully convinced by the authors’ responses and I still have some requests about the analysis, and I believe that the manuscript quality could be easily improved:
1. I definitely think that correlations, RMSD or Nash-Sutcliffe Efficiencies between modelled daily values would better support the conclusions of the paper than Figure 5, 6, 7 and 8. Indeed, these figures show a good agreement between models rather than big differences. Even if I partially agree with the fact that “Bringing the melt rates predicted by these highly parameterized models in agreement with the observed ones seems to be rather a curve adjustment exercise than an indicator of correct physics. » I still think that the authors need to present a Table with statistics before reaching this conclusion. Figures with hourly values would also help. Perhaps these figures could be included in the appendix (see below).
2. I still believe that COSIMA calibration has been done “quickly” on Mocho glacier and is not robust. For instance, the mean modelled SWnet values (from COSIMA model) significantly differ from the measured ones (even on Mocho glacier), whereas the mean values obtained with the very simple assumption of the EB-model (a constant albedo value) correctly fit with mean observations (see Table A1). This suggests that model calibration has not been performed in order to allow transferability of the parameters. A calibration based on a Monte-Carlo approach and an optimization of scores (e.g., the Nash-Sutcliffe efficiency) using observations from different sites would remove this. This aspect could be improved. I understand that the authors will not make any further full calibration in the present study, but I believe that this is a clear deficiency of the study, because this impedes concluding whether modelling inconsistencies result from the lack of optimization or from model itself.
As a consequence, I suggest that the authors inform the reader about this lack. After writing that “Openly shared codes are the best way to improve physical models, since everyone can test the individual parametrizations against his data and adjust or improve them accordingly. This is the preferred way to obtain physical parametrizations as opposed to large chains of models which supposedly model physical processes whose individual performance, however, is not validated and the final results rather come out of a black box”, I suggest to add that “Moreover, an accurate model optimization is required to allow transferability of the parameters at large scales. This could be done using a Monte Carlo approach (e.g., Mölg et al., 2012) and calculation of scores between measured and modelled surface height changes and/or energy fluxes. Optimization strategies could be defined using field records from the different studied regions. Finally, the parameter transferability in space and time may be tested using a leave-one-out cross validation approach (e.g., Hofer et al., 2010).”
=> Concerning Authors responses:
>Authors: “What do you mean with “fully” used? To our understanding the bulk aerodynamic approach is a way of quantifying the turbulent fluxes by making three important assumption (which are stated in the manuscript). We are happy to get feedback from you in the case you are thinking that we forget to mention another important assumption of this approach.”
My response: My concern was that the authors did not use stability corrections in the reference and in the EB-model. These corrections are crucial to accurately compute the turbulent heat fluxes. In the present version it is still unclear whether the authors include corrections during unstable conditions. Indeed, Page 11, line 15, the authors write “The same stability correction based on the bulk Richardson number Ri describe in section 3.2 is applied here to account for the reduced vertical exchange of air masses in stable conditions (Braithwaite, 1995b).”. Do the authors apply any corrections when Ri < 0?
>Authors: “You are right. This formula was implemented in the first version of COSIMA that we downloaded and was indicated in Huintjes et al. 2015a. But this seemed to be a bug which was changed now.”
My response: Please remove the red curve in figure 3 or clearly write in the figure caption that the initial COSIMA relationship was not used and that you used Bolton curve instead.
>Authors: “Ok. Since this work is about Chilean Glaciers we restricted the literature review mainly to studies on Chilean Glaciers. We are happy to receive additional recommendations for studies of relevance for our work to include
them in the literature review.”
My response: I did not make an exhaustive review, but only for Chile and only during the last 4 years, several papers were published:
Weidemann SS, Sauter T, Malz P, Jaña R, Arigony-Neto J, Casassa G and Schneider C (2018) Glacier Mass Changes of Lake-Terminating Grey and Tyndall Glaciers at the Southern Patagonia Icefield Derived From Geodetic Observations and Energy and Mass Balance Modeling. Front. Earth Sci. 6:81. doi: 10.3389/feart.2018.00081
Réveillet, M., MacDonell, S., Gascoin, S., Kinnard, C., Lhermitte, S., and Schaffer, N.: Impact of forcing on sublimation simulations for a high mountain catchment in the semiarid Andes, The Cryosphere, 14, 147–163, https://doi.org/10.5194/tc-14-147-2020, 2020.
AYALA, A., PELLICCIOTTI, F., PELEG, N., & BURLANDO, P. (2017). Melt and surface sublimation across a glacier in a dry environment: Distributed energy-balance modelling of Juncal Norte Glacier, Chile. Journal of Glaciology, 63(241), 803-822. doi:10.1017/jog.2017.46
Bravo, C., Quincey, D. J., Ross, A. N., Rivera, A., Brock, B., Miles, E., & Silva, A.(2019). Air temperature characteristics, distribution, and impact on modelled ablation for the South Patagonia Icefield. Journal of Geophysical Research: Atmospheres,124, 907–925. https://doi.org/10.1029/2018JD028857
Bravo, C., Loriaux, T., Rivera, A., and Brock, B. W.: Assessing glacier melt contribution to streamflow at Universidad Glacier, central Andes of Chile, Hydrol. Earth Syst. Sci., 21, 3249–3266, https://doi.org/10.5194/hess-21-3249-2017, 2017.
Other studies were published in areas near Chile, but with a clear interest in the present paepr (in particular for turbulent heat fluxes estimations, or for glacier-wide calculations):
Litt, M., Sicart, J., Helgason, W.D. et al. Turbulence Characteristics in the Atmospheric Surface Layer for Different Wind Regimes over the Tropical Zongo Glacier (Bolivia, 16∘S). Boundary-Layer Meteorol 154, 471–495 (2015).
Maussion, F., Gurgiser, W., Großhauser, M., Kaser, G., and Marzeion, B.: ENSO influence on surface energy and mass balance at Shallap Glacier, Cordillera Blanca, Peru, The Cryosphere, 9, 1663–1683, https://doi.org/10.5194/tc-9-1663-2015, 2015.
Finally several studies using the SEB at large scale (including Chile), have been published. In these studies, estimates are not clearly validated with field data, but these studies could be discussed (at least Mernild et al. study is frequently cited):
Mernild, S. & Wilson, R. The Andes Cordillera. Part III: glacier surface mass balance and contribution to sea level rise (1979-2014). Int. J. Climatol. 37, 3154–3174 (2016).
L.J. Vargo, J. Galewsky, S. Rupper, D.J. Ward, Sensitivity of glaciation in the arid subtropical Andes to changes in temperature, precipitation, and solar radiation, Global and Planetary Change, Volume 163, 2018, Pages 86-96, https://doi.org/10.1016/j.gloplacha.2018.02.006.
Sagredo, E., Rupper, S., & Lowell, T. (2014). Sensitivities of the equilibrium line altitude to temperature and precipitation changes along the Andes. Quaternary Research, 81(2), 355-366. doi:10.1016/j.yqres.2014.01.008
>“Page 10 Line 6: please cite : U.S. Army Corps of Engineers (1956).
>Authors: We do not have access to this piece of work!”
My response: The Snow Hydrology book is (at least) available in google books (even if the plate 5.2, presenting albedo variations has not been scanned): https://books.google.fr
>Authors: “Regarding the great similarity of the course of the melt rates observed in the Figures 6,7,8 we do not think that additional statistics (like correlations and standard deviations) would add some crucial new insights.”
My response: I still believe that a table with correlations, RMSE or Nash-Sutcliffe Efficiency values between daily values of SWnet, albedo, LWnet, SH and LH, between every models would help (in particular, to interpret the modelling of the albedo). Please see other comments below.
=> “Comparison with previous version:
>A quick comparison of the results of present and previous versions of the paper demonstrates that assuming a constant albedo value or Tsurf = 0°C (see Table A1) induce very large biases in the modelling results. The use of stability corrections is crucial in the calculation of the turbulent heat fluxes. This demonstrates that results from simple models (as EB-model) are largely erroneous. This could be commented in the text.
>In the models (e.g. in the reference model), I understand that melt occurs when Tsurf = 0°C and SEB = R + SH + LH > 0. However, how do the authors consider the refreezing and the frigories stored during the night or when the SEB is negative or very close from 0 (at the end of the melt season (see figure 6,7,8)? Melt can occur at the surface when Tsurf = 0°C, while the subsurface temperature is still below 0°C. But, in that case, the surface melting is reduced because incoming shortwave radiation penetrates in the ice and heat condition warms the snow/ice layers below the surface. This induces that Melting amount is less than R+ SH + LH until frigories are removed from the subsurface layers. How is this process considered in the simple reference and EB-models? Please comment this point in the text.
=> Minor remarks:
>In the abstract, the authors write that “The influence of the stability correction and the roughness length on the magnitude of the turbulent fluxes in the different climate settings was examined.” => This conclusion is almost not developed in the text and relies on Table A1, which is found in the appendix. Please move Table A1 in the main text and develop the discussion on this point.
>Page 2, Line 35 : Sensibel => sensible
Please remove other typos
>Page 4 Last line : “The projections of future changes in climate depend on the different climatological/glaciological zones. This is why a detailed analysis of the processes that determine the energy exchange at the surface of the glaciers in the different climatological zones is necessary, to be able to make reliable predictions of future surface mass balance and melt water discharge of Chilean glaciers.” => please, move this sentence at the beginning of this section.
>Page 6, line 15: “Hourly data were generated using a matlab interpolation scheme.” => What is "a matlab interpolation scheme" ? what kind of scheme did you use?
>Page 7 line 2: “incoming and outgoing longwave radiative fluxes (table 3)” => Table 3 refers to surface roughness lengths.
>Page 8, line 6 :” Assumption 1. is normally fullfilled for a neutral atmosphere, but, over a glacier surface, the temperature gradient is often inverted (especially during summer). This stable layering of air masses reduces the vertical exchange specially for low wind speeds” => what do the authors mean with an inverted gradient? The surface is warmer than above? Do they mean unstable conditions? If it were the case, there would be a contradiction in the sentence. Please explain and refer directly to stable or unstable conditions. Moreover, did they apply corrections when Ri<0?
>Page 13, line 5 : Figure Cs4?
>Page 13, Line 11: radiation show => radiation shows
>Page 14, line 15: “The predicted melt rates for the specific study points (locations of the AWS) in the ablation area of the glaciers are higher for the Patagonian Glaciers as compared to the glaciers of the Central Andes” and Page 22, line 1: “The inferred melt rates in the ablation area of the glaciers were higher for the Patagonian Andes than for the Central Andes.” => as written in my first review, I don’t agree with this sentence: it depends on elevation. The only way to conclude would be to plot the mean melting vs. mean temperature, or Melting vs. elevation difference between the AWS and the ELA. Actually, if the authors display this figure, they will observe that there is no possible comparison between sites: for instance, melting at the Bello Glacier was similar to the Tyndall Glacier when temperature was 5°C lower at The Bello Glacier. This figure could be presented in the appendix.
>Section 5.2 : Parametrizations of the surface energy fluxes :
Again, I propose to include table A1 directly in the main text. I also propose to include a table with statistics in the appendix, and to discuss this new table in the text. These tables are important to understand the discussion. For instance: Section 5.5 page 21, line 2: “The albedo aging effect implemented in COSIMA is a big improvement regarding to constant albedo parametrizations for snow, firn and ice surfaces. » => after analysing SWnet values in Table A1, this sentence looks strange. It seems that a constant albedo gives better results that a varying albedo. A quick statistical analysis, using a RMSD or a Nash–Sutcliffe efficiency calculations done on SWnet values would help to characterize whether the albedo model improves the quality of the modelling (i.e. using the albedo model gives better results than assuming a constant albedo value).
Actually, I already asked information on albedo calibration (and more generally on calibration of the model parameters) in my first review, but it seems that my question was not clear.
>The authors answered:
>“By comparison with the measured (daily) albedo.”
How did the authors obtain the parameters given in the caption of Figure 9 ? Did they perform a monte carlo approach with data from Mocho glacier? Or did they realize a quick estimation of parameters allowing the mean modelled albedo to approximately equal the mean measured values at Mocho ? Finally, please give references to discuss the values given in the caption of Figure 9 (These values are likely representative for an area and type of glacier elsewhere).
>Section 5.5, Page 21, Line 1: “The capacity of the models to reproduce the measured radiative fluxes is still improvable.” => The authors never compare the fluxes at an hourly timescale, but such a figure would clearly help to see model discrepancies. In particular, if the measured hourly (uncorrected and corrected) values of LWout were displayed and compared with those from the COSIMA model, it would help to see whether the "constant " correction on LWout (and LWin) data is accurate or not. A comparison over a 5-day time period for one (or every glaciers) could be presented in the appendix.
>Page 19, line 3 : paramtrizations => parametrizations
>Page 20, line6 : CR4 => CNR4 ?
>Page 20, Line 8 : this in in very good agreement => this is in good agreement
>Page 20, line 30 : higer => higher
>Page 20, lines 10-14 : I don’t understand the comparison between the Grey glacier and the Exploradores glacier. Elevation of the sites is very different and the glaciers are from two different icefields. Please remove this and use data from NPI instead. Why did the authors use these data rather than those used in Figure 11 from Schaefer, et al., (2013)?
>Line 1 page 18 : “A drawback of this comparison is certainly that we do not know the exact amount of snow falling on the glacier, but deduce it from the liquid precipitation measured at a automatic weather station in the valley.“ => I propose to include that “Moreover, a more detailed modelling of albedo accounting for snow metamorphism could be tested to see whether it reproduces more accurately the short timescale variations”.
=> References:
Hofer, M., Mölg, T., Marzeion, B., and Kaser, G.: Empirical statistical downscaling of reanalysis data to high resolution air temperature and specific humidity above a glacier surface (Cordillera Blanca, Peru), J. Geophys. Res., 115, D12120, doi:10.1029/2009JD012556, 2010.
Mölg, T., Maussion, F., Yang, W., and Scherer, D.: The footprint of Asian monsoon dynamics in the mass and energy balance of a Tibetan glacier, The Cryosphere, 6, 1445–1461, doi:10.5194/tc-6-1445-2012, 2012.
Schaefer, M., H. Machguth, M. Falvey, and G. Casassa (2013), Modeling past and future surface mass balance of the Northern Patagonia Icefield, J. Geophys. Res. Earth Surf., 118, 571–588, doi:10.1002/jgrf.20038. |
- Printer-friendly version
- Supplement
