the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Enhanced MOIDS-derived ice physical properties within CoLM revealing bare ice-snow-albedo feedback over Greenland
Abstract. Under global warming, the Greenland Ice Sheet (GrIS) is experiencing unprecedented mass loss. A key factor closely associated with this loss is the change of snow and ice albedo, which is directly influenced by the firn metamorphism. To investigate the impact of bare ice microstructure changes on the regional warming of the GrIS ablation zone, SNICAR-ADv4 (Snow, Ice and Aerosol Radiation model Adding-Doubling Version 4), a physically based radiative transfer model, is incorporated in Common Land Model version 2024 (CoLM2024). It allows the land surface model represent the ice albedo with changes in ice properties rather than using a constant ice albedo value. Meanwhile, quality control was performed on the bare ice physical property dataset input into CoLM, with multiple MODIS products combined. Using SNICAR-ADv4 reduced the overestimation of shortwave broadband albedo by 60 %, with a bias of only 0.053. Further sensitivity experiments indicate that the albedo in the bare ice region is reduced by 0.032 during the summer due to the bare ice metamorphism, producing a 2-m temperature forcing of 0.071 °C and a snow cover change of -0.011. The contraction of snow cover exposes more bare ice and will further decrease albedo and increase the ground's absorption of solar radiation, suggesting a feedback mechanism involving bare ice, snow, and albedo. This highlights the indispensable role of bare ice physical properties in the bare ice-snow-albedo feedback for amplifying melt, and more significant feedback is expected to be produced by land-atmosphere coupling model.
- Preprint
(2988 KB) - Metadata XML
-
Supplement
(681 KB) - BibTeX
- EndNote
Status: open (until 09 Apr 2025)
-
RC1: 'Comment on egusphere-2025-230 : Review of “Enhanced MODIS-derived physical properties within CoLM revealing bare-ice-snow albedo feedback over Greenland”, by S. Guo et al.', Anonymous Referee #1, 18 Mar 2025
reply
Review of “Enhanced MODIS-derived physical properties within CoLM revealing bare-ice-snow albedo feedback over Greenland”, by S. Guo et al.
This paper examines the extent to which accounting for the physical properties of ice in areas of exposed bare ice on the Greenland ice sheet affects the albedo and surface air temperature as well as the extent of snow cover via what they call the ice-snow-albedo feedback. The work is based on the use of a SNICAR-ADv4 radiative transfer model (which explicitly represents the optical properties of snow and ice, taking into account several species of light-absorbing constituents) implemented in the CoLM surface model. It also takes advantage of MODIS products combined with data quality-indices to provide more reliable physical properties of bare ice that are used as inputs to SNICAR-ADv4. The simulation results are compared with those from an earlier version of SNICAR (SNICAR-AD), which uses constant ice albedo values. Comparison of the results from the two SNICAR versions allows to assess the importance of changes in ice properties (i.e. bare ice metamorphism) on the albedo and the surface climate.
The method does not appear to be novel as it is similar to that proposed by Wicker-Clarke et al. (2024), albeit with the Energy Exascale Earth System Model (E3ESM) rather than CoLM. A similar study has also been conducted by Antwerpen et al. (2022). You mention that you added quality-information regarding MODIS products. However, both Wicker-Clarke et al. (2024) and Antwerpen et al. (2022) excluded some pixels from the analysis and filtered data. Wouldn't this be a way of adding quality information? However, I acknowledge that the evolution of Greenland is a growing matter of concern with increasing mass losses now dominated by changes in surface mass balance (SMB). SMB is strongly dependent on surface albedo which is expected to decrease in response to surface meting and increase in the extent of darken areas. It is therefore of primary importance to investigate the response of a variety of models to surface processes including bare ice metamorphism. This is why, I recommend the publication of this paper after major (and minor) comments (see below) have been addressed.
Major comments:
1/ First of all, I found that the methodology is not sufficiently explained. This is detrimental for the overall understanding of the paper. I had to read the paper several times. I had to read Section 2 several times to understand the whole procedure. In my opinion, part of the problem is that the description of the method closely resembles that described in Wicker-Clarke et al (2024) but with the removal of a certain amount of information that would have been necessary for a full understanding of the method.
More details should also be given about the different models used in this study. A number of things are not very clear:
i/ Which variables are simulated by CoLM and and for this study? It seems to me that this is not clearly stated anywhere. Is it albedo, but I thought that the albedo was calculated by SNICAR?
ii/ Why is the BATS scheme mentioned (at the same level as SNICAR) whereas you never refer to in the rest of the paper? Mentioning the BATS scheme adds to the confusion.
My recommendation is therefore to clearly explain the functionalities of each of the models used in this study: SNICAR-AD, SNICAR-ADv4 and CoLM. In fact, in the current version of the paper, I have the feeling that the information is diluted in various places or that it arrives too late. To make things clearer, a scheme similar to that of Figure 10 could be incorporated in Section 2 (obviously without the panel illustrating the ice-snow-albedo feedback).
This would also offer the opportunity to briefly present the physical processes associated with the evolution of the snowpack, such as compaction and refreezing, among others (see for example Flanner and Zender, 2005). This aspect is important because it is involved in the ice-snow-albedo feedback highlighted in the present paper.
In the introduction, you also mention that the vertical profile of snow grain size as well as snow thickness are considered as input variables of the SNICAR model. If so, where do these input data come from? On the other hand, it seems to me that the snowpack model should be able to simulate these variables itself. They should therefore be considered as output variables. Can you clarify or comment please?
Overall, I suggest to reorganize Section 2 (while addressing the above comments) as follows:
Section 2.1: Snow and ice albedo schemes / Section 2.2: Data / Section 2.3: Method / Section 2.4: CoLM simulation
L130 (and also L154 and L203): You mention that ice albedo is 0.80 and 0.55 for VIS and NIR spectra. These values correspond more to the albedo values for fresh snow than to the albedo values for bare ice. Do SNICAR-AD-CoLM simulations actually use these values for bare ice? If so, it is not surprising that the use of SNICAR-ADv4 leads to a significant reduction in albedo. If this is the case, you should redo a SNICAR-AD-CoLM simulation with values more characteristic of those for bare ice.
Furthermore, Figure 10 shows that the ice albedo values for SNICAR-AD are 0.6 and 0.4 for VIS and NIR respectively (which seems more realistic to me). Please clarify
2/ My second comment is related to the effect of bare ice metamorphism on the surface ai temperature (+ 0.071°C) and on the reduction of ~1% of the snow cover. This does not seem very significant. To be more convincing, I recommend to provide additional diagnostics. As SNICAR (AD and Adv4) includes a snow scheme, I guess that all the elements are available for computing the surface mass balance and the runoff coming. This should help better quantify the actual impact of a more realistic calculation of the ice albedo.
3/ The Discussion section lacks a detailed comparison with the results of Antwerpen et al. (2022) and Wicker-Clarke etal. (2024).
Other comments:
Title: MOIDS à MODIS
Section 1:
L63-65 The sentence is too long. Please, split in two parts.
L63: extend → extent
L64: Surface melt is also associated with a reduction of snowpack thickness and is not only due to bare ice exposure. However, I agree with the fact that surface melting over bare ice surfaces contributes to GrIS mass loss. This should be better explained.
L66: was → is
L73: in → over
L130 (and also L154 and L203): You mention that ice albedo is 0.80 and 0.55 for VIS and NIR spectra. These values correspond more to the albedo values for fresh snow than to the albedo values for bare ice.
L136: in ablation season à during the ablation season
L145: properties
Section 2:
L151: features enhancements à” includes improvements in the representation of…” sounds better?
L152: What are the improvements in the anthropogenic disturbances processes? Which kind of processes are you referring to?
L155: Please remove “is”
L177: “include snow non sphericity” à include non-spherical snow grains? Maybe, you should precise (here or in Section 2.4) that you only consider spherical grains (i.e. the SSA formulation in Eq. 1 is only valuable for spherical grains).
L182: Accounting for ice layers in SNICAR (AD and/or ADv4) comes too late
L188: land-only CoLM simulations? In your results section, you do not mention any land-only CoLM simulations but SNICAR-AD-CoLM simulations or SNICAR-ADv4-CoLM simulations. I guess that in this study, you do not consider ESM simulations, but the formulation “land-only CoLM simulation” seems to be a bit confusing as you consider the snow/ice albedo schemes embedded in CoLM. Please clarify.
L209: grids → grid cells
L231: Start a new sentence after “September”
L235: I guess that the quality index is provided in the MODIS database? Maybe, you could add a sentence like “the quality helps to identify regions with cloud cover contamination, detrimental atmospheric conditions or insufficient observation”? Or something equivalent…
L247: “and SZA” → What do you mean?
L248: fist → first
L248: Is the cloud mask determined with the use of the quality index?
L253: Use the same wavelength units everywhere: nm or µm.
L253: Do you mean that for reflectance values below 0.6, pixels are considered to be bare ice? This does not sound very clear for me. How has this threshold been defined ?
L257-258: Explain why pixels with elevations exceeding the mean equilibrium line altitude are excluded (I guess that above a certain elevation, snow is not completely melted and therefore there ice no exposed bare ice? But this could be specified).
L267: “running offline SNICAR-ADv4 simulations” → running offline the SNICAR-ADv4 model
L267-272: This sentence is too long and is not very clear. Please better explain why do you need to adjust the input parameters. On which basis? Split the sentence in 2 or 3 parts.
L274: ari→ air
L279-280 the non-unicity of the relationship between SSA, ice density and air bubble radius should be justified/explained. It seems to me that, to the first order, pixels with larger SSA values correspond to pixels of lower density, larger Vair and smaller Reff. This defines the unicity of the relationship(?) But, maybe I missed something. In any case, this should be clarified.
L294-297: Too long sentence. Please, split in two parts.
L296: Figure → Figures
L301-302 : Is it due to poor-quality data?
L318: within ice → within the ice
L318-321: Could you explain the link between Reff and the scattering/reflection efficiency?
Section 3:
L329: Figure → Figures (same for L333, L339, L370, L374)
L329: demonstrate → display
L341: southwest: not really convincing. Do you mean southeast?
L344-346: This sounds a bit subliminal. Could you be more synthetic?
L346: illustrated → illustrates
L348: southwestern and northeastern à Rather: everywhere in the peripheral areas of the ice sheet except in the southeastern part.
L346-349: Please, rephrase. I suggest the following (or something equivalent): Figure 5a shows the spatial distribution of land ice underlying the snowpack. The areas where land ice is the main type of land cover are located in the periphery of the the GrIS with the exception of the southeastern edge. Values of land ice fraction below 1 implies that the corresponding grid cells contain etc…
L349-350: Removing “In tandem… enabled CoLM” would make the sentence clearer
L354: “bare ice fraction frequency”: I don't understand what this means at all. Please, explain. How this frequency is determined? (same thing for Fig. 5c caption).
L354: Figure 3d→ Figure 5c
L365: region → regions
L376-377: I would expect a single RMSE value for the whole GrIS as RSME is defined as a sum. Please explain (give the mathematical formula) how you compute the RMSE; Moreover, I am not sure I have the right idea of what you mean by “linear trend”; Linear trend of what? Please clarify in the main text and in the figure captions of the Supplement.
L384-387: Not clear. I suggest a new formulation: The decrease in the positive bias of CoLM SNICAR-ADv4 can also be clearly seen in the shortwave, visible and near-infrared albedo time series, with the area-weighted mean albedo of the GrIS bare ice regions steadily decreasing throughout the summer period from 2000 to 2020, compared with CoLM SNICAR-AD.
L390: SNICAR-ADv4 enabled simulations à CoLM SNICAR-ADv4 simulations
L391 and L392: MCD43C4 à MCD43C3
L410: from → compared to
L425: has significantly reduced
L440: This could be confirmed or infirmed with new diagnostics (e.g. surface mass balance and/or runoff à See Major comments)
L443: northeast ablation zone: Rather northwestern and western?
L455: control experiment: this is the first time you use this term. Please explain what is your control experiment.
L456: commence → starts/begins
L465: region → regional
L468: effect → affect
L476: speciafic →specific
L489: 2021 → 2020
L492-493: strong climate response: This sounds like an overstatement
L497: in ablation zone → in the ablation zone
L513-514: impact of the glacier calving (dynamic process) and submarine melting: Add a reference and/or develop your arguments.
L526: coupling → coupled
The reference Wicker et al. (2024) should be changed in Wicker-Clarke et al. (2024) both in the main text and in the reference list
Figures: Avoid using pastel colours in some figures (e.g. Fig. 4, Fig. 5c, Fig. 6-8).
Flanner, M. G., and C. S. Zender (2005), Snowpack radiative heating: Influence on Tibetan Plateau climate, Geophys. Res. Lett., 32, L06501, doi:10.1029/2004GL022076.
Citation: https://doi.org/10.5194/egusphere-2025-230-RC1
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
84 | 13 | 7 | 104 | 11 | 1 | 2 |
- HTML: 84
- PDF: 13
- XML: 7
- Total: 104
- Supplement: 11
- BibTeX: 1
- EndNote: 2
Viewed (geographical distribution)
Country | # | Views | % |
---|---|---|---|
United States of America | 1 | 30 | 30 |
China | 2 | 25 | 25 |
France | 3 | 11 | 11 |
United Kingdom | 4 | 7 | 7 |
Germany | 5 | 5 | 5 |
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
- 30