the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 May 2024
Quantifying radiative effects of light–absorbing particles deposition on snow at the SnowMIP sites
Abstract. The deposition of light-absorbing particles (LAPs) leads to a decrease of surface albedo over snow covered surfaces. This effect, by increasing the energy absorbed by the snowpack, enhances snow melt and accelerates snow aging, process which in turn is responsible for further decreasing the snow albedo. Capturing this combined process is important in land surface modelling, as the change in surface reflectivity connected with the deposition of LAPs can modulate time and magnitude of snowmelt and runoff. These processes impact regional water resources, and can also lead to relevant feedbacks to the global climate system. We have recently developed a new numerical snowpack model for the GFDL land model (A Global Land Snow Scheme, or GLASS). GLASS provides a detailed description of snow mass and energy balance, as well as the evolution of snow microphysical properties (grain shape and size). We now extend this model to account for the presence of light-absorbing impurities, modelling their dry and wet deposition in the snowpack, the evolution of their vertical distribution in the snow due to precipitation and snow melt, and the effect of their concentration on snow optical properties. To test the effects of the resulting snow scheme, we force the GFDL land model with deposition of black carbon, mineral dust and organic carbon obtained from a general circulation model (GFDL AM4.0). We evaluate the new model configuration at a set of instrumented sites, including an alpine site (Col de Porte, France) where in-situ observations of snow (including spectral measurements of snow reflectivity and concentration of LAPs) allow for a comprehensive model evaluation. For the Col de Porte site, we show that GLASS reproduces the observed magnitudes of impurities concentration in the snowpack throughout a winter season. The seasonal evolution of the snow optical diameter is also qualitatively reproduced by the model, although the increase in snow grain diameter during the melt season appears to be underestimated. For a set of instrumented sites spanning a range of climates and LAP deposition rates (the `SnowMIP' sites) we then evaluate the number of snow-days lost due to the deposition of dust and carbonaceous aerosols. We find that this loss ranges between 5 and 24 days depending on the site. The resulting snow model with LAP-aware snow reflectivity show a good agreement with measurements of broadband albedo and seasonal SWE over the study sites.
- Preprint
(4388 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1258', Anonymous Referee #1, 20 Jun 2024
This study describes the implementation of a new snow albedo scheme in the GLASS land surface component of the GFDL climate model, accounting for the effects of light-absorbing particles (LAPs). It then evaluates the simulation of snow mass, snow depth, and surface albedo at several sites contributing to the SnowMIP effort. Overall, this study represents an advance in scientific capabilities of the GFDL climate model. I found it particularly useful that the model was run in "single point" mode for comparisons with the SnowMIP sites. The quantification of the reduction in number of snow cover days due to the presence of LAPs was also useful, though it would be helpful to also include an evaluation of LAP concentrations in snow, compared with observations, to help understand how realistic the simulated LAP-induced albedo effect is. The manuscript is generally well-written and well-organized. Aside from one major suggestion, I have only minor comments.
Major comments:The new model prognoses the mixing ratios of dust and black carbon in surface snow, and I believe that concentrations of these particles have been measured at some of the SnowMIP sites used in the evaluation, such as Senator Beck and Col du Port. It would be quite helpful to know how the simulated mixing ratios compare with observations, as this would inform on potential sources of bias in the simulated SWE evolution throughout the seasons. Such an evaluation could suggest, for example, that biases in impurity concentrations are responsible for SWE biases, or conversely if the particle mixing ratios appear realistic, that there are other problems with the snow model.
Section 2.6: It was not apparent to me how/if the albedo of the ground underlying snow and snow thickness affect the snow albedo calculation. Is the snow assumed to be optically "semi-infinite" regardless of snowpack thickness? If so, this would cause a high bias in the albedo of thin snowpack, and it should be acknowledged.
Minor comments:
line 55: "Black carbon has the largest absorption..." -> Black carbon has the largest absorption *per unit mass* ..."
line 119: "sol-snow" -> "soil-snow"
Lines 132-146, section 2.2: Overall, this is a helpful summary. Briefly, though, could you please also list the maximum number of snow layers allowed in this model, along with maximum/minimum layer thicknesses (especially near the top)? This info is probably available in the companion paper, but it would be helpful to include it here, too.
lines 158-164: Would it be accurate to state that the mixing ratio of LAPs within precipitation is held constant throughout the month? Also, is there any interpolation between the months, or does the mixing ratio change abruptly on the first day of the month?
lines 158-168: Related, is there an interactive, coupled version of the atmosphere and land models, where prognosed aerosol deposition is coupled with GLASS each timestep? (Or, are there plans to extend this modeling framework to the coupled model?)
line 193: Is the same scavenging ratio assumed for hydrophilic and hydrophobic BC? It seems that the scavenging ratio should be larger for hydrophilic BC, as specified by Flanner et al (2007).
line 196-197: "... snow properties are averaged over a near surface layer of thickness set equal to up to 3cm." - Is this simply the top thermodynamic snow layer in GLASS, or is this a weighted average of multiple snow layers? When is it less than 3cm? Please elaborate a bit on this scheme.
line 197: "... snow albedo is expressed..." -> For clarification and to avoid confusion with broadband albedo, I suggest "... band-specific snow albedo is expressed..."
lines 215-232 (Section 2.7): Do LAPs influence albedo in both spectral bands, or only the visible band? Line 221 mentions "as a function of spectral band", but line 230 lists only single absorption cross-sections for each type of LAP. Are these absorption cross-sections for the visible band only, and if so, what is assumed for the near-IR band?
Also, what are the spectral intervals of the two bands used in this model? (Often they are separated at 700nm)
lines 290-295: The assumption of constant LAP mixing ratios within precipitation throughout the month could also explain some of this discrepancy.
Figure 3: Please explain this "retrieval parameter" for the different observational curves, perhaps in the text.
Please include a table describing the acronyms of the SnowMIP sites (clp, snb, etc), including the long names and locations of the sites.
Section 4.2: As mentioned under Major Comments, this analysis would be augmented with an evaluation of the impurity amount (mixing ratio) in snow.
Figures 4-6: Which "single year" was simulated and observed at each site? (And were they the same?) Please explain and/or include this information in the table of sites requested two comments above
Overall, the grammar is good, but there are numerous instances of minor issues that should be fixed prior to publication. Lines 79-80 demonstrate just one example of a sentence that needs to be cleaned up.
Citation: https://doi.org/10.5194/egusphere-2024-1258-RC1 - AC2: 'Reply on RC1', Enrico Zorzetto, 07 Oct 2024
-
RC2: 'Review Comment on egusphere-2024-1258', Anonymous Referee #2, 19 Jul 2024
The authors expanded the GFDL LM4.1 snowpack processes to include LAPs effects and evaluate the simulated snow albedo across different measurement sites. They found that the resulting snow model with LAP-aware snow reflectivity show a good agreement with measurements of broadband albedo and seasonal SWE over the study sites. They further evaluated the number of snow-days lost due to the deposition of dust and carbonaceous aerosols, which ranges between 5 and 24 days depending on locations. This work is an important improvement for the GFDL snowpack model, which could provide better simulations for future coupled climate runs. Overall, the manuscript is well organized. I have a few specific comments/suggestions for the authors to consider.
Specific comments:
- Is there any canopy snow process included in LM4.1? A brief description would be useful.
- How is the snowpack liquid water treated? Is there a liquid water holding capacity parameter prescribed? How important is it to assign different inter-layer snowmelt water scavenging coefficients for IM vs EM LAPs?
- How did the authors assign dry and wet deposited LAPs to IM or EM within the snow?
- How many dust size bins are considered and what are they?
- Does melt-freeze of snow change the IM or EM status of LAPs?
- Is internal heating within the snowpack column due to light absorption considered?
- The snow albedo parameterization from Dang et al. 2015 and He et al. 2018 is for semi-infinite snowpack, which may lead to uncertainties in the albedo calculation here. It will be good to clarify this and briefly discuss this.
- How did the authors use the snow grain shape parameters to compute the optical diameter?
- How is the alpha_b in equation (5) computed? What is the physical meaning of this parameter?
- For daily average of albedo, did the authors use downward solar radiation as the weights?
- Figure 2A: Why does the model have a consistently high snow albedo than observations after 2014 May (which seems to be a snow-free period)? Also, since there is no snow after May 2014, why is the snow albedo from both observation and model not zero?
- Section 4: the description of the results is too qualitative. Please include some quantitative numbers when presenting the results. Also, more physical explanations/insights could be added to the results. For example, why does the model not capture the variation of the daily albedo variation over most sites (Figure 4) and why does the model results show systematic overestimate/underestimate in some sites (Figure 5).
- Figure 7: It seems that the model SWE bias is dominated by other model snow or forcing processes instead of LAP effects. This may be worth some discussion.
- I would suggest adding a subsection for uncertainty discussion. Some of the uncertainties involved in the model are mentioned in my earlier comments. A few key uncertainty factors that are worth discussing: (1) snow grain shape and size evolution, (2) using Eq.8 to combine different LAPs, (3) aerosol deposition flux, (4) missing snowpack processes, (5) LAP meltwater scavenging, etc.
Citation: https://doi.org/10.5194/egusphere-2024-1258-RC2 - AC1: 'Reply on RC2', Enrico Zorzetto, 07 Oct 2024
Model code and software
GLASS snow model including LAPs deposition, model input and output data Enrico Zorzetto, Elena Shevliakova, Sergey Malyshev, and Paul Ginoux https://zenodo.org/records/10901373
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 301 | 99 | 29 | 429 | 25 | 20 |
- HTML: 301
- PDF: 99
- XML: 29
- Total: 429
- BibTeX: 25
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1