the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Oct 2024
Regional and seasonal evolution of melt ponds on Arctic sea ice
Abstract. Melt ponds on sea ice significantly modify the absorption of solar radiation by the sea ice-ocean system in the Arctic, thereby influencing the energy budget and sea ice mass balance. Consequently, melt ponds are crucial to the positive sea ice-albedo feedback mechanism, which is a major factor to the enhanced Arctic warming observed in recent decades, with implications for the global climate. Given the high seasonal and interannual variability of melt ponds, understanding the mechanisms behind their evolution and their impact on the sea ice state is essential for improving sea ice and global climate models. This analysis must also take into account regional differences in melt pond evolution.
Here we present seven years (2017–2023) of melt pond fraction data produced with the Melt Pond Detection 2 (MPD2) algorithm from optical Sentinel-3 satellite observations. We demonstrate strong regional differences in the melt pond evolution process as well as high seasonal and interannual variability. The study shows that the variability is lower in the Central Arctic than in the marginal Arctic seas, which is in compliance with the more stable sea ice coverage in the Central Arctic. Hence this region also shows the highest potential of using melt pond fractions at the beginning of summer as an indicator for the summer surface energy budgets and thus the progress of melt season. Between the nine regions for the marginal seas, strong differences in melt pond variability are observed.
Sea ice surface topography and air temperature are investigated as primary factors to influence melt pond formation and evolution. Air temperature shows an immediate impact on the melt pond coverage, whose short-lived changes can be well resolved with the new MPD2 melt pond fraction product. A higher sea ice surface roughness leads to lower melt pond fractions in the beginning of the season. Later in the melt season, different behavior of melt pond drainage leads to a reversal of that relationship and hence lower melt pond fractions are observed on the level, flatter sea ice.
- Preprint
(29435 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3127', Anonymous Referee #1, 13 Dec 2024
Review of ”Regional and seasonal evolution of melt ponds on Arctic sea ice”
Tracking of melt pond coverage with satellite remote sensing remains the only viable method for full Arctic coverage across the melting seasons and through years. Given the importance of melt ponds to the surface energy balance of the Arctic Ocean, efforts to improve accuracy and coverage in said application are welcomed. Here, the authors present an effort to compile a 7-yr dataset of melt pond coverage from Sentinel-3 OLCI, representing an important step forward from prior studies involving MODIS as the EO instrument, as MODIS is now nearing the end of its life.
In the analysis, I particularly noted the results from surface roughness (ridge) impacts on melt pond fractions, which were interesting and novel. Overall, I feel that the paper carries sufficient weight to be considered for publication in TC. Some parts of the analysis could use improvement, with details listed in the comments below.
Major comments:
Surface topography effects on MPF retrieval: Not only clouds but also surface topography features such as large and small-scale ridges and other ice deformities can cause shading of the surface at low Sun elevations. Have you considered if this can play any noticeable effect on the retrieval of MPF, which relies on the variability of the surface reflectance as the principal input? What are the satellite/illumination geometry limits you apply?
Also, what is the general MPF retrieval uncertainty in MPD2, I assume you have an order of magnitude number available somewhere but this manuscript did not seem to contain it? Considering the significance of the results w.r.t. observational uncertainties is of course always advisable in remote sensing studies, but perhaps that was treated in more detail by Niehaus et al. (2023)?
Temporal & spatial inhomogeneity filter: To be clear – all data over the 7-yr period is removed over the areas where the 0.25 threshold is crossed in the temporal side? The text is a bit ambiguous on this point. On the spatial filtering side, a similar question arose – you removed a full 7-day period if less than 20% of the region was covered by valid retrievals? Is the 20% condition thus different from the spatial homogeneity condition, where a threshold of 0.1 seems to apply (but was that value only for Central Arctic)? Please consider rewriting this section for clarity, I had quite some trouble following along.
Figure 7: Whereas Fig 6 has almost too much content to properly keep track of, this is a very good and informative figure, thanks for including it!
Section 5 – the analysis is interesting and references to earlier such studies are appropriate. I was missing a more direct quantitative comparison to Schröder’s and Liu’s studies – could you not compute the correlation coefficients in a consistent manner with them to perform a S3-simulation and S3-MODIS intercomparison here?
Minor comments (line):
195: “It would then fall into the third group that is otherwise contains” – I could not follow this sentence, please revise.
218: “little deformed” -> meaning what?
379-380: Are there definite thresholds applied to classify large and small obstacle spacing in the text?
Citation: https://doi.org/10.5194/egusphere-2024-3127-RC1 -
RC2: 'Comment on egusphere-2024-3127', Anonymous Referee #2, 05 Jan 2025
General Comments:
This study uses the Melt Pond Detection 2 algorithm to examine the seasonal evolution and regional differences in Arctic melt pond fractions. It also examines the drivers of melt pond formation and evolution and the implications of melt ponds on the sea ice minimum. Overall, the manuscript is well-written, has a thorough methodology, and has sufficient results to support its conclusions. I particularly appreciate the comprehensive discussion of regional differences, which includes summaries that highlight the differences. The study effectively demonstrates the potential of the MPD2 product to analyze the seasonal evolution of melt ponds at various scales and their influence on sea ice variability and trends. As highlighted in the manuscript, analyzing changes in melt pond fraction during moist intrusion events is another valuable application of this product. Below, I provide several comments aimed at improving clarity and suggestions for enhancing the Figures.
Specific Comments:
Figure 4: Does a similar figure appear somewhere on NSIDC? If so, consider referencing that figure, moving Figure 4 to a supplement file, or adding the region outlines to Figure 2. The closest I could find is Figure 3 of “Special Report 25” at https://nsidc.org/data/nsidc-0780/versions/1, although it is less accessible and not as visually appealing.
Figures 5–7:
- While it is clear that Figures 5–6 show the median melt pond fraction, it is unclear if Figure 7 shows the spatial mean or average melt pond fraction. “Average melt pond fractions” also appear in several places; does “average” consistently refer to the median, or is the spatial mean sometimes used? Clarification of this distinction would enhance reader understanding.
- Is the median used because it is more representative of the regional melt pond fraction than the mean given the skewed distributions and data filtering/masking?
- Regional differences in melt pond fractions are somewhat challenging to discern in Figures 5–6. In contrast, regional differences are more apparent in Figure 7 and A1, and the summary in Table 1 is helpful. Have the authors considered plotting the 7-year average (mean or median) seasonal melt pond fraction? This could be added as an additional subplot in Figures 5–6 or an extra row in Figure 7. A 7-year average might further highlight regional differences.
Line 274: Is the uncertainty of 8% estimated using data over the entire Arctic, and is there evidence that certain regions experience substantially different retrieval uncertainties? It would also be helpful to cite Niehaus et al. (2024) here (or in the Figure 7 caption).
Surface energy budget: Have the authors considered explicitly illustrating the impact of melt pond fraction on surface fluxes? While I appreciate the albedo comparisons in Section 4.1 and the correlations between accumulated melt pond-covered area and sea ice minimum extent, a visual comparison would be insightful. For instance, an additional row of subplots in Figure 9 showing albedo could provide a clearer link.
Data availability: Thank you for including links to all the datasets used in the study, particularly the gridded MPD2 melt pond fraction data. While the methodology is thorough, making some code publicly available would improve transparency and facilitate reproducibility.
Technical Corrections:
Line 56: Change “temporal and and spatial” to “temporal and spatial.”
Line 190: Change “the the Barents Sea” to “the Barents Sea.”
Line 470: Change “surface surface energy budget” to “surface energy budget.”
Citation: https://doi.org/10.5194/egusphere-2024-3127-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 211 | 81 | 7 | 299 | 2 | 4 |
- HTML: 211
- PDF: 81
- XML: 7
- Total: 299
- BibTeX: 2
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1