Title: Melt pond fractions on Arctic summer sea ice retrieved from Sentinel-3 satellite data with a constrained physical forward model.

Overall:  Melt ponds play a critical role in assessing the reflectance of sea ice and thus the surface energy balance.  Determination of areal melt pond coverage is challenging, due to the variability in individual pond reflectance, and to the flat spectral reflectance of snow-covered ice and open ocean.  The MDP2 technique presented in this paper utilizes physical parameterizations to determine individual pond reflectance rather than a single fixed reflectance curve, as well as 4 surface types to more accurately estimate pond coverage from a high-resolution satellite-based sensor (Sentinel-2).  The results presented show promise to improve pond coverage parameterizations in sea ice models, as well as regional studies of the evolution of the sea ice surface.

The MDP2 technique is sound and presented clearly.  The paper is well-written and the figures are important to explain intermediate and final results.

I recommend accepting this paper after addressing a couple comments and the corrections that follow:

Specific comments:

Line #    Comment

95    Should provide a citation for OSI-SAF drift data, and mention its tracking error

124    should read “from 2012 to 2022”

166      The assumption of clear melt water may lead to occasional error in actual pond reflectance, with dust and other aerosols present in or near the bottom of the pond.  This is particularly true near coastal areas but have been shown to be transported from regions with high industrial activity.  This would be difficult to address in this pond estimation technique but should be mentioned as a possible impact on pond reflectance.

230    Should read “This additional information…”

286    Should read “no longer considered…”

297    Should read “are NOT appropriate…”

372    Should read “both differences…”

456    End of sentence continues after Fig 14 – hard to find

525    Does Webster or others have MOSAiC aerial melt pond coverage, that can be directly compared to the derived pond fraction from MPD2?

559    need brackets about citations

The authors build on previous work on deriving melt pond fraction from Sentinel-3. The main improvement is including open ocean as additional surface type classification. The improvements to the algorithm are impressive when compared to the old algorithm and a better pan arctic melt pond fraction data is really good news! The methodology is sound and detailed, and the authors have done a good job of explaining it all in a logical order. Additionally, the figures are clear and contain a lot of information. I would have liked to see a more thorough assessment of errors (i.e. do they vary based on melt pond fraction), but given the lack of data for the and length of the paper it would be good to see it in future work.

I recommend accepting this paper.

L43: Satellites don’t cover all of the arctic.

L175: ?

L373: bothdifferences

In Figure 15, it would be nice to see a few more of the example scenes (and perhaps the S2 image?)