See attached file

The authors present an analysis of a novel dataset of mid-infrared measurements of the lunar surface. These data were acquired by the MAJIS instrument during a gravity assist maneuver of the JUICE spacecraft in August 2024 and represent a rare opportunity to observe the Moon in the mid-infrared wavelength range. The paper integrates recent developments in thermal modeling and lunar science and serves as a preparatory step for the upcoming MAJIS observations at Europa, Ganymede, and Callisto.

While the dataset is valuable and provides a challenging test case for thermal models, there are several significant issues regarding the line of argumentation, technical implementation, and scientific evaluation. In its current form, both the study and the manuscript fall short of the standards expected for publication. Substantial revision and additional work will be required to address these issues before the manuscript is ready for publication.

Overall evaluation

• Does the paper contain new data or new ideas or both of them?
  
  The paper contains new data of the Moon acquired by the MAJIS instrument in the mid-infrared. This dataset is of general interest to the community.
• Are these up to international standards?  
  
  The dataset is up to international standards. The data-processing needs substantial improvement. The interpretation has several major issues (see my comments below).
• Is the presentation clear?
  
  No (see my comments below).

• Does the author reach substantial conclusions?
  
  Especially the conclusion section is confusingly written and contains many errors (see my comments below).

• Is the length of the paper adequate?
  
  Sometimes the paper reads a little verbose. I recommend shortening it.

• Is the language fluent and precise?
  
  The language is fluent. Unfortunately, many formulations are too vague or misleading (see my comments below).

• Are the title and the abstract pertinent and understandable?
  
  The title is good. The abstract does not adequately reflect what was done in the study (please see my comments below)

• Is the size of each figure adequate to the quantity of data it contains?
  
  Yes – all good.
• Does the author give proper credit to related work and do they indicate clearly their own contribution?
  
  Yes. However, some works are missing that focus on the same wavelength region.

Overall rating

• may have potential after additional work and resubmission

Revise the scope of the work:

Several claims in the abstract regarding validation and preparation are not supported by the analysis and are not consistently addressed in the conclusion. I recommend revising the stated scope and improving the overall coherence of the manuscript to ensure alignment between abstract, analysis, and conclusions. Please consider the following points:

Validation of MAJIS capabilities: The abstract states that the study aims to validate the capabilities of MAJIS (line 15), and the term “validation” is used repeatedly throughout the manuscript. However, it is unclear what specific type of validation is actually performed. The paper does not explicitly address, for example, absolute calibration, pointing performance, thermal stability, or instrument-level performance metrics. Instrument calibration and validation appear to be primarily covered by Langevin et al. (this issue). In contrast, the present manuscript does not read as a systematic validation of MAJIS performance, more like a plausibility check. Please revise.

Application of methods to JUICE’s main mission: At several points (e.g. line 93-95), the study claims to prepare for the main mission. “can help validate techniques for interpreting thermal data from Ganymede, Callisto and Europa, where similar sunlit observations are planned. This is a relevant and interesting activity. However, I do not explicitly see how the results from this paper will be transferred to the Jovian moons.

Are similar thermal models/thermal retrievals required? Can they be used right away, or must they be adapted? If so, how?

Comparison of different thermal models/retrieval approaches: Using three different models/approaches makes the study more robust and also provides interesting insights on how the models compare. The investigations have close ties to lunar hydration analysis, as the thermal correction approaches of Li and Milliken 2017 and Wohlfarth et al. 2023 have been primarily developed for this purpose. If not already covered in the companion paper Langevin et al. 2026 (this issue), the manuscript could become more interesting to a wider audience by discussing the implications of the three different thermal approaches/methods for lunar hydration analysis (See also Schörghofer et al 2021).

Reframe the scope of the work (recommendation): In my view, the main contribution of this study lies in the exploration of a novel dataset in particularly challenging wavelength region, as well as in the development and testing of thermal modeling and retrieval approaches in preparation for future studies of the Jovian moons. In my view, the analysis therefore appears to serve as a scientific plausibility check, based on comparisons with well-established lunar properties, rather than as a formal instrument validation.

I therefore recommend revising the stated scope of the work and adjusting the terminology accordingly, for example by:

• Framing the study as an exploratory investigation of the challenging lunar mid-infrared wavelength region using a new dataset.
• Describing the analysis as a scientific plausibility check for the MAJIS instrument (useful for assessing instrument performance), rather than a formal validation.
• Emphasizing the role of this work in preparing thermal modeling and retrieval methods for application to the Jovian moons.

Revise the Bayesian approach:

The authors state Kirchhoff’s law as simply “emissivity = 1 – reflectance,” which is not accurate. Kirchhoff’s law is derived from energy conservation and properly states:



e_d = 1-r_hd,

where is the directional spectral emissivity and is the hemispherical-directional spectral reflectance. Using an incorrect formulation can introduce significant errors in the crossover region, as demonstrated by Myhrvold et al. (2018) in response to thermal modeling in the NeoWISE mission, which erroneously simplified Kirchhoff’s law to .

It is important to note that Kirchhoff’s law applies to directional spectral emissivity, not bolometric emissivity. This distinction is particularly critical when making graybody assumptions. The authors should ensure that the correct definition is applied, energy conservation is properly satisfied, and the models are re-run, if any changes had to be made.

Empirical thermal correction:       

No comments on this section. 

Revise the thermal roughness model approach:

Quantifying surface roughness on planetary surfaces is inherently challenging due to its scale dependence and statistical complexity. For example, Figure 1(a) in Rubanenko et al. (2020) shows lunar surface roughness at multiple scales, indicating that the RMS slope generally increases as the spatial scale decreases. As discussed by Bandfield et al. (2015, Figure 11), the spatial scales most relevant for thermal modeling are those at which thermal isolation becomes significant, typically in the millimeter-to-centimeter range. At these scales, common lunar roughness estimates are approximately 20° near nadir (Bandfield et al. 2015; Wohlfarth et al. 2023) and 30–35° in the TIR (Bandfield et al. 2015; Rubanenko et al. 2020; Müller et al. 2021; Wohlfarth et al. 2023).

In contrast, the authors derive surface roughness from LOLA measurements, which sample much larger spatial scales (meters) and therefore likely underestimate roughness at the scales controlling thermal emission. Additionally, the manuscript does not provide specific roughness estimates in terms of RMS slope or Hapke θ.

To improve the study, the authors should:

1. Include relevant literature on lunar surface roughness at thermal-relevant scales.
2. Revise the roughness implementation to reflect the appropriate spatial scales.
3. Investigate the effects of surface roughness on their results and discuss the implications accordingly. 

Line 494: The statement that the model “explicitly accounts for local topography through a fractal surface formulation” is misleading. Typically, “explicitly accounting for topography” implies the use of actual topographic information, such as a digital elevation model. Please clarify what is meant here.

The thermal model uses albedo, but the terminology in the manuscript is inconsistent: the authors refer to it as directional-hemispherical albedo (Lines 435, 443) and bolometric albedo (Line 395). This should be made consistent, and the derivation of this quantity should be explained. Since albedo varies between mare and highland regions, it likely has a significant effect on the energy balance. How does this parameter influence the results?

The model assumes an effective emissivity of 0.95. Is this based on a graybody assumption? I checked the original publication and Wohlfarth et al. (2023) use directional spectral emissivity derived from Apollo data. Please compare the graybody assumption using a single effective emissivity with an approach that combines a bolometric emissivity of 0.95 and spectral emissivity in the relevant wavelength range. Please revise the manuscript accordingly.

The authors also derive maps of macroscopic surface roughness. It is unclear whether this parameter is meaningful, given that thermal modelling typically requires roughness at millimeter-to-centimeter scales. Furthermore, there is a concern that the roughness map may partially absorb variations in bolometric albedo. Please analyze this interaction and discuss its implications.

Lines 493–494: The phrase “and thus thermal beaming” is misleading. Thermal beaming occurs primarily when the vectors pointing to the observer and the Sun are roughly aligned. Is this truly the effect being referred to?

Figure 11: The numbers in the legend are unclear. Roughness is typically expressed in terms of RMS slope or Hapke θ, so the manuscript should clarify what these values represent.

Line 496: The points listed here (sub-pixel topography, surface composition, and wavelength-dependent emissivity) are in fact the critical factors that drive differences in thermal emission. Their importance should be highlighted and discussed. 

Comparison with Diviner:

No comments on this section.          

Temperature analysis:

I do not agree with the conclusion of this section. I do not see complementary strengths in the approaches themselves. It is more due to implementation choices in this particular study, that some approaches perform better than others. Please revise.           

Revise the emissivity analysis:

This section requires substantial revision.

Emissivity definitions: It is unclear whether the emissivity quantities are directly comparable. Several different emissivity definitions exist, with subtle but important differences. The manuscript does not specify which emissivity the Bayesian approach uses; if Kirchhoff’s law was applied correctly, it should be the spectral directional emissivity . What type of emissivity does the empirical approach use? The roughness model assumes an effective emissivity, while the original publication used a spectral emissivity. Please ensure that all emissivities compared in Figure 14 follow the same definition.

Comparison with other lunar mid-infrared measurements: Wu et al. (2018) present disk-resolved infrared measurements of the lunar nearside at 3.8 µm, acquired by the Gaofen-4 weather satellite. These observations cover most regions targeted by MAJIS and fall within its spectral range, making the Gaofen-4 dataset well suited for comparison. Including such a comparison would allow a clear examination of:

• Emissivity differences between mare and highland terrains,
• Maturity-related effects, and
• The consistency of emissivity estimates across instruments.

Comparison with laboratory samples: Reflectance and emissivity spectra of Apollo return samples are available and could provide a direct spectral comparison. I suggest adding these spectra to Figure 15. Notably, the highland emissivity spectrum in Figure 15 appears generally higher than the mare spectrum around 4 µm, which contradicts the spectral shapes of Apollo samples (e.g., Müller et al., 2021, Figure 6). Please discuss this discrepancy.

Diviner comparison: The comparison with Diviner appears incomplete and potentially misleading. Diviner channel 3 measures emissivity at 7.8 µm, near the Christiansen feature (CF) of highland materials, whereas the CF of mare materials occurs at longer wavelengths (Greenhagen et al., 2010). This naturally produces higher emissivity for highlands at the specific wavelength of channel 3, but does not imply that highlands are generally more emissive than mare regions across the mid-infrared. Including additional channels around 8 µm would provide a more complete picture. Please clarify the meaning of this comparison.      

Comparison with IIRS data: A comparison with data from the IIRS instrument onboard Chandrayaan-3 would, in general, be insightful. However, I think there is not yet a definitive dataset suitable for comparison.

Lunar hydration: Lunar hydration has been extensively studied, focusing on the 3 µm absorption band in reflectance spectra. How does the bump around 3 µm in Figure 15 relate to this hydration feature? Even though the companion paper (Langevin et al., this issue) addresses lunar hydration in MAJIS data, a brief discussion here would be valuable.

Retrieved reflectance spectrum: Figure 15 currently shows only retrieved emissivities. For evaluating the Bayesian retrieval algorithm, it would be helpful to also plot the retrieved reflectance spectrum, which could be directly compared with existing laboratory samples for consistency. How do the other methods treat reflectance spectra? Please add this analysis.

Lines 634–636: The statement that “The spectral crossover, governed by vibrational properties of silicate minerals and by particle-scale roughness, reconciles the MAJIS and Diviner observations and highlights their complementarity” is confusing and scientifically inaccurate.

• For lunar silicates, vibrational effects are primarily observed in the thermal infrared (TIR). To my knowledge, the spectral crossover region of lunar silicates is not significantly influenced by vibrational effects. While CO₃ or organic absorption bands exist in this range, they are not relevant for lunar materials. I recommend examining Apollo sample reflectance spectra available from the RELAB database for guidance.
• The term “particle-scale roughness” is unclear. For thermal emission, the relevant effect is macroscopic roughness at the thermal isolation scale (millimeter to centimeter scale), not individual particle properties.
• It is unclear what is meant by “reconciles”. Was there a specific discrepancy between MAJIS and Diviner observations? This should be clearly stated and justified.

The entire paragraph requires revision for scientific clarity and accuracy.

Lines 636–643: This paragraph currently reads more like an advertisement and does not add scientific value. Also, it is quite vague. Please revise to focus on evidence, analysis, and interpretation rather than general statements.

Spectral feature around 3.5 µm: This feature is unusual and requires further investigation:

• How does the radiance spectrum appear? Is the feature present in the raw radiance, or only in the retrieved emissivity spectrum?
• If it is present in the radiance, it could indicate an instrumental effect.
• If only in the retrieved emissivity, it may point to a retrieval artifact.
• How does the retrieved reflectance spectrum compare?
• What emissivity spectra are obtained from the roughness model and the empirical model in this region?

Addressing these questions will clarify the origin of this feature and improve confidence in the retrievals.

Line 679: No, it’s the Christiansen feature, not the Reststrahlen bands.              

Lines 679 – 683: Is this really what was shown? Please revise!

Revise the conclusion:         

The conclusion requires substantial revision. It appears to have been written in haste and contains multiple incorrect statements, unsupported logical inferences, and vague or confusing formulations. Please carefully check consistency with the results presented in the manuscript and revise accordingly. In particular, the following points should be addressed:

Line 694: The phrase “where reflected and thermal contributions overlap” is misleading. Reflected and thermally emitted radiance are superimposed across the entire scene, independent of incidence angle. Please revise this statement.

Line 698: The term “anisothermal regions” is not defined. Please clarify what is meant or remove the term.

Lines 697–698: I do not agree with this conclusion, as it is not explicitly demonstrated in the manuscript.

Line 701: The phrase “key outcome” is vague and informal. Please restate more precisely.

Lines 702–706: This paragraph is incorrect on multiple levels and must be revised. The authors attribute emissivity differences between mare and highland regions primarily to porosity, which contradicts the established literature. In the visible and near-infrared, mare basalts appear darker than feldspathic highlands primarily due to composition, notably higher abundances of pyroxene, olivine, and ilmenite, with maturity effects also playing a significant role. The wavelength range up to appx. 6-8 µm is  affected by nanophase-iron–related space-weathering effects. Porosity may contribute, but only as a secondary effect. Please revise.

Line 705: The statement “display higher thermal inertia” is misplaced. Thermal inertia cannot be determined in the present observational scenario, as it requires nighttime measurements.

Line 706: Similarly, enhanced thermal conductivity cannot be inferred without nighttime data and should not be claimed.

Line 707: The statement “The comparison with Diviner further refines this picture” is vague. There is no refinement shown; only an additional data point. Please revise.

Lines 707–711: The argument in this paragraph is unclear and potentially misleading. Diviner Channel 3 measures emissivity at 7.8 µm, near the Christiansen feature (CF) of lunar highland materials, while mare CFs occur at longer wavelengths (Greenhagen et al., 2010). Due to the characteristic shape of silicate emissivity spectra, this naturally results in higher emissivity values for highlands at this specific wavelength. This does not imply that highlands are generally more emissive than mare regions across the mid-infrared, where mare materials typically exhibit higher emissivity due to compositional differences. The wording “with highlands appearing more emissive” should therefore be revised to clarify that this is a narrow-band effect driven by CF position, not a general emissivity property.

Moreover, the purpose of this comparison remains unclear. As stated before, a more meaningful discussion would involve comparisons with Gaofen-4 observations in the same wavelength range and with mid-infrared laboratory measurements of Apollo samples and analog materials.

Line 727: The statement “display higher emissivity and warmer daytime temperatures, consistent with denser and less weathered regolith” is incorrect. As discussed above, composition is the dominant factor.

Line 734: The claim that the study captures “regolith properties that control emissivity and roughness” is not supported by the presented analysis.

Lines 733–736: This paragraph is vague and lacks a clear scientific argument. In particular, the phrase “inversion across the Christiansen feature” does not constitute a meaningful analysis in this context.

Line 739: The statement regarding “how mineralogy, grain size and roughness jointly shape the thermal environment” partly repeats earlier claims and introduces grain size, which has not been discussed elsewhere in the manuscript. This does not align with the presented work.

Lines 744–750: Terms such as “unique,” “critical benchmark,” and “pivotal” read more like promotional language than scientific conclusions. Several claims are also unclear:

• Why does this constitute a robust framework for MAJIS exploration?
• Why is Callisto only mentioned at the very end, despite being a major MAJIS target?

Please rework the entire conclusion, ensuring that all statements are supported by the analysis, that interpretations are physically justified, and that limitations and shortcomings of the emissivity retrieval are clearly acknowledged. The revised conclusion should also explicitly reflect and integrate the concerns raised throughout the earlier sections of the review.

Improve the presentation, especially the overall line of argumentation:

Unfortunately, the study contains several loose ends and remains too vague in multiple places. The manuscript would benefit from improved internal coherence, a clearer articulation of its central aims, and a careful revision to ensure consistency across all sections. I recommend performing a thorough consistency check after addressing the points raised above.

References

Bandfield et al. 2015 DOI: 10.1016/j.icarus.2014.11.009

Greenhagen et al. 2010 DOI: 10.1126/science.1192196

Li and Milliken 2018   DOI: 10.1126/sciadv.1701471

Müller et al. 2021 DOI: 10.1051/0004-6361/202039946

Myhrvold et al. 2018 DOI: 10.1016/j.icarus.2017.12.024

Rubanenko et al. 2020 DOI: 10.1029/2020JE006377         



Wohlfarth et al. 2023 DOI: 10.1051/0004-6361/202245343

Wu et al. 2021 DOI: 10.1029/2020GL088393