Assessing VIIRS constellation seasonal snow cover over the French mountains with Sentinel-2

Imperatore, Nicola; Gascoin, Simon; Lafaysse, Matthieu; Dumont, Marie; Mauss, Adrien; Guével, Stéphane; Hernandez, Jean-Baptiste

doi:10.5194/egusphere-2026-1122

Preprints

https://doi.org/10.5194/egusphere-2026-1122

Preprints

13 Mar 2026

| 13 Mar 2026

Assessing VIIRS constellation seasonal snow cover over the French mountains with Sentinel-2

Nicola Imperatore, Simon Gascoin, Matthieu Lafaysse, Marie Dumont, Adrien Mauss, Stéphane Guével, and Jean-Baptiste Hernandez

Abstract. Remote sensing observations of snow-covered area from moderate-resolution (100–1000 m) optical sensors are critical for water resource monitoring and climate-related applications. Many studies have relied on MODIS snow products, but NASA plans to stop science data collection from MODIS instruments in the next two years. The Visible Infrared Imaging Radiometer Suite (VIIRS), designed as follow-on instrument to MODIS, has similar characteristics and is currently onboard three satellite platforms with daily revisit. Several agencies now distribute operational snow cover products based on VIIRS data, yet independent evaluations of these products remain scarce, limiting their adoption by the snow science community. Here, we assess NASA VIIRS snow cover products over several mountain ranges in France using Sentinel-2 snow cover products as a high-resolution reference. We also evaluate a near-real-time VIIRS snow cover product developed by Météo-France over metropolitan France. The analysis covers an entire winter season and examines performance across contrasting topographic settings. Across the different products, we find an average bias close to zero and a root mean square error ranging from 10 to 15 % for snow cover fraction, while the corresponding binary snow classification achieves a F1-score of 92–93 %. However, uncertainties can reach 30 % for challenging observation conditions, including mixed pixels, forested areas, and north-facing slopes. Finally, we demonstrate that combining observations from multiple VIIRS platforms effectively reduces cloud cover without degrading snow cover retrieval quality.

Received: 26 Feb 2026 – Discussion started: 13 Mar 2026

Competing interests: At least one of the (co-)authors is a member of the editorial board of The Cryosphere.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Nicola Imperatore, Simon Gascoin, Matthieu Lafaysse, Marie Dumont, Adrien Mauss, Stéphane Guével, and Jean-Baptiste Hernandez

Status: final response (author comments only)

RC1: 'Comment on egusphere-2026-1122', Anonymous Referee #1, 07 Jun 2026

The authors analyze VIIRS snow cover product by leveraging a powerful approach of using coincidence Sentinel-2 snow cover products as a high-resolution reference dataset for error analysis. This approach is powerful because it enables a very large sample size for the analysis. The intent and overall architecture of the study is sound and a valuable contribution to the snow science community but, to recommend for publication I recommend major adjustments to the implementation details of the analysis and the structure of the paper.

More text needs to be included that defends the use of a NDSI based fractional snow cover truth data set over a physically based model. Evaluating a product that uses NDSI makes sense, but comparison to a NDSI based fractional snow cover "truth" product concerns me given the existence of physically based models, including open source versions, that outperform NDSI for calculating the fractional snow cover of multi spectral pixels. NDSI does not contain information regarding the spectral signature of snow and various non-snow mixed pixels can have the same NDSI values as snow-covered pixels. Various studies have shown that NDSI is less accurate than spectral unmixing techniques when estimating fractional snow cover. I’d prefer a physically based spectral unmixing truth data set for the highest fidelity truth set from Sentinel-2 and for the paper to have the most impact, but If the author wants to stick to a NDSI based truth dataset, they should better explain the choice. There are always trade offs with choice of validation data set, so I would like to see more discussion on the reason behind this choice and the known pros and cons (cons are omitted currently) so the reader can better contextualize the results.

Line 170 - Please explain and defend the choice to reproject and resample at the same spatial resolution as the original data set instead of coarsening to account for geolocation uncertainty. Coarsening is a typical approach to eliminate concern on geolocation accuracy that does not impact overall error analysis metrics (especially at the sample sizes of this paper) but can reduce the impact of geolocation errors on results. Also, in Section 4.4 I recommend only reprojecting the truth data from Sentinel-2 to the VIIRS projection. Please explain the reasoning to choose a third projection, a non-equal area projection to reproject all the data into for a fractional snow cover area analysis. The point is made that the errors (in terms of area size) introduced are small, but why not keep the VIIRS data in their original projection and only retroject the high spatial resolution truth data to the VIIRS projection for error analysis? This choice to reproject both datasets is error inducing. I strongly recommend keeping the dataset that is being evaluated in the same projection that it is delivered in, and coarsening prior to analysis.

Line 220 - “We binarize the FSC values into "snow" and "no-snow" classes using a threshold of 50% for both the reference and evaluation datasets.” I disagree with this choice. Pixels with substantive snow cover are classified as snow free pixels for analysis. Why was the binary snow cover threshold not set at FSC>0? It is possible to accurately detect snow in Sentinel data well below 50% snow cover and I’d expect the truth dataset to be a binary mask of all pixels that include snow. This is also a reason for such a high representation of TN’s in the overall dataset. It seems that the snow covered pixels are a significant portion of the TN dataset with the 50% cutoff. I disagree with this decision and would like to see the snow cover fraction threshold that binarizes between snow and now snow set at a FSC>0 or similar. This choice, alongside the choice to reproject both the truth and VIIRS data, concerns me on the quality of the output analysis. The results in the first few rows of Table 4 seem to invalidate the decision to use a threshold of 50% for snow / no snow cover as you see significantly worse performance in this range of valid snow covers from 1-99% FSC, which is an artifact of the choice to call snow covered pixels with less than 50% snow cover, snow free pixels, alongside the reprojecting and lack of coarsening to account for geolocation errors in the analysis.

Use of accuracy metirc - I disagree that this is a valuable measure of algorithm performance given the unbalanced nature of the data set and relative ease algorithms have in distinguishing snow free pixels from pixels with factional snow cover, FN and FP are far more challenging issues that are important for understanding the value of a snow cover product. I’d rather only see F1 score in the graphs in figure 4 and the detailed performance analysis.

Line 312 and use of VIIRS to validate MODIS - I disagree with this approach. The same truth data set (Sentinel) should be used to access MODIS and VIIRS and make determinations on the relative performance of the two. The current chained approach to error analysis introduces unnecessary error propagation into the analysis.

Discussion Section - Many results are presented directly in the discussion section. Recommend moving these important results into the results section and explaining the approach to these analysis in the methods section.

Line 411 - recommend removing accuracy as a main result metric.

Technical corrections:

- Suggest rewording sentence from line 18-20 for clarity.

- Line 35 - recommend changing “guaranteed” to “operationally planned” or similar, there is risk with any operational system that it does not meet expected mission lifetime and there can be a gap in coverage before a replacement is fielded.

- Line 67 - contradicts with Table 2 - reviewed and ASO snow depth data, not SWE was used for the validation. Suggest changing “which provides lidar based snow water equivalent (SWE) measure” to “which provides 3m spatial resolution lidar based snow depth measurements that were converted into binary snow masks.”

- Line 227 - after “we also report the F1 score “ suggest adding “which is the harmonic mean of omission and commission”.

Citation: https://doi.org/10.5194/egusphere-2026-1122-RC1

AC1: 'Reply on RC1', Nicola Imperatore, 24 Jul 2026

We thank the referee for their meticulous analysis of our work and the numerous pertinent suggestions they have provided. We have conducted several additional analyses to address their comments, as outlined in the point-by-point response below. We believe we can propose a revised manuscript that would largely resolve their concerns.

Note : During the revision process we identified and corrected two minor issues:

1. A coding error in the conversion from NDSI to FSC which only affected NDSI values below 0.1 in resampled NASA products.

2. Three corrupted products in our VNP10A1 dataset, corresponding to tile h18v04 from December 3–5, 2023. These products correspond to clear-sky acquisitions following a major snowfall event affecting a large number of relevant pixel-to-pixel correspondences

As a result, the updated metrics show minor differences, such as a slightly higher overall RMSE and a slightly more positive bias. We confirm that this update does not affect any of the conclusions of the work. Hereby, we compare Table 3 of the original manuscript to its updated version.

Tab. 3 of the original manuscript

Tab. 3 after corrections

VNP10A1	4.48+06	18.7	0.93	0.02	0.06	0.42	10.09
VJ110A1	4.45+06	17.9	0.92	0.02	0.06	0.81	10.88

We propose to update all tables and figures of the manuscript to integrate these two corrections. The numerical results reported in this reply will include these corrections.

Hereafter, the point-by-point response. In italic, we report the comments of the referee.

We agree that the choice of the retrieval algorithm for the reference dataset is relevant and needs to be further discussed. We specify that our goal is to use readily-available Sentinel-2 products as reference. This approach sacrifices some control on the reference data but allows for scalable, replicable, and straightforward ground truth generation. We apologize for the lack of clarity regarding this point. We propose to reformulate Section 6.1 to better explain our motivations and include some considerations on spectral unmixing and the choices of other authors.

We thank referee 1 for this comment. This choice is explained by our objective, which is to quantify the errors of VIIRS products for a snow modelling system that will assimilate VIIRS data at their native resolution. Our objective here is to evaluate a snow product (including all error sources) and not its snow retrieval algorithm. The finer resolution of VIIRS products (375 m) compared to MODIS products (463 m) is also a motivation for many users to move from MODIS to VIIRS, and, therefore, we believe that characterizing the errors at the original VIIRS resolution is also useful in this perspective. In the revised manuscript, we propose to clarify this motivation and make it clear that our assessment includes various sources of errors, including geolocation errors.

Also, in Section 4.4 I recommend only reprojecting the truth data from Sentinel-2 to the VIIRS projection. Please explain the reasoning to choose a third projection, a non-equal area projection to reproject all the data into for a fractional snow cover area analysis. The point is made that the errors (in terms of area size) introduced are small, but why not keep the VIIRS data in their original projection and only retroject the high spatial resolution truth data to the VIIRS projection for error analysis? This choice to reproject both datasets is error inducing. I strongly recommend keeping the dataset that is being evaluated in the same projection that it is delivered in, and coarsening prior to analysis.

We acknowledge the referee’s point that the least error-prone approach is to evaluate VIIRS data in their native projection. However, we evaluate here two VIIRS snow cover products with different coordinate systems: NASA (MODIS sinusoidal grid) and Météo-France (geographic coordinate system). We initially chose to (re)project VIIRS datasets to the same UTM spatial reference system as the Sentinel-2 products to minimize the number of reprojection operations. In fact, most users reproject moderate-resolution sensor data for their applications, with UTM being a common choice. Based on preliminary tests, we assumed that this reprojection step would not significantly affect the results. However, following the referee’s comment, we assessed this assumption by reprojecting the reference data to the VNP10A1 projection. We illustrate the differences by reporting Table 4 of the original manuscript in the two cases. Please note that the results have minor differences from the original manuscript for the reasons stated at the beginning of this document.

Table 4 in UTM projection

1-99 %	Forest	N	4.99e+04	39.58	0.71	0.20	0.28	-1.38	29.72
1-99 %	Forest	S	1.28e+04	43.83	0.80	0.15	0.20	-1.86	22.97
1-99 %	Open	N	6.58e+04	68.07	0.90	0.17	0.12	-1.74	23.06
1-99 %	Open	S	5.76e+04	65.63	0.93	0.17	0.05	3.31	18.01
0-100 %	Forest	N	3.18e+05	8.41	0.76	0.02	0.24	0.42	13.63
0-100 %	Forest	S	2.02e+05	4.32	0.87	0.01	0.14	-0.13	6.42
0-100 %	Open	N	2.44e+05	36.10	0.95	0.02	0.07	-0.64	13.00
0-100 %	Open	S	3.69e+05	18.27	0.96	0.01	0.03	0.55	7.43

Table 4 in VNP10A1 projection

1-99 %	Forest	N	4.61e+04	40.12	0.71	0.20	0.28	-2.01	30.72
1-99 %	Forest	S	1.23e+04	42.05	0.80	0.15	0.19	-2.01	23.79
1-99 %	Open	N	6.17e+04	69.00	0.90	0.18	0.12	-1.51	23.72
1-99 %	Open	S	5.38e+04	64.33	0.93	0.18	0.05	3.13	18.86
0-100 %	Forest	N	2.99e+05	8.50	0.77	0.02	0.23	0.31	13.97
0-100 %	Forest	S	1.91e+05	4.32	0.87	0.01	0.13	-0.16	6.66
0-100 %	Open	N	2.30e+05	37.36	0.95	0.03	0.06	-0.61	13.34
0-100 %	Open	S	3.49e+05	18.14	0.96	0.01	0.03	0.49	7.72

The differences are in the order of 0.5 to 1 % and do not alter the conclusions. Hence, we propose to keep the UTM results in the manuscript, and to add a comment in Section 4.1 to explain the motivations and the little impact.

Line 220 - “We binarize the FSC values into "snow" and "no-snow" classes using a threshold of 50% for both the reference and evaluation datasets.” I disagree with this choice. Pixels with substantive snow cover are classified as snow free pixels for analysis. Why was the binary snow cover threshold not set at FSC>0? It is possible to accurately detect snow in Sentinel data well below 50% snow cover and I’d expect the truth dataset to be a binary mask of all pixels that include snow. This is also a reason for such a high representation of TN’s in the overall dataset. It seems that the snow covered pixels are a significant portion of the TN dataset with the 50% cutoff. I disagree with this decision and would like to see the snow cover fraction threshold that binarizes between snow and not snow set at a FSC>0 or similar. This choice, alongside the choice to reproject both the truth and VIIRS data, concerns me on the quality of the output analysis. The results in the first few rows of Table 4 seem to invalidate the decision to use a threshold of 50% for snow / no snow cover as you see significantly worse performance in this range of valid snow covers from 1-99% FSC, which is an artifact of the choice to call snow covered pixels with less than 50% snow cover, snow free pixels, alongside the reprojecting and lack of coarsening to account for geolocation errors in the analysis

It is common practice to compute snow-covered area by applying a threshold of 0.4 to NDSI values (Härer et al., 2018), which roughly corresponds to 50% FSC. However, we understand the referee’s concern and acknowledge that, following the definition of the NDSI_Snow_Cover variable of V[NP|J1]10A1 (Riggs and Hall, 2023), 0 % is indeed a more appropriate threshold. Therefore, we recomputed the confusion matrix using a threshold of FSC>0 to binarize both target and reference datasets. The F1 score decreases by 2 to /3 % for V[NP|J1]10A1 products, with different commission and omission error estimates. Also, the confusion table becomes more balanced with a TN gof 73 % instead of 81 % (not shown as correctly guessed by the referee). We emphasize that the new binarization threshold only impacts the F1 score, commission and omission errors, while the RMSE and bias calculation remain unaffected. This means that the conclusions regarding the higher uncertainty found in mixed pixels remain unaltered. We propose to update the manuscript with these new methods and results. We hope that this revision will address the referee’s concerns regarding the quality of the analysis. Finally, the choices about reprojection and lack of coarsening and their impact have been motivated earlier in this response and will not be addressed in this paragraph.

Use of accuracy metric - I disagree that this is a valuable measure of algorithm performance given the unbalanced nature of the data set and relative ease algorithms have in distinguishing snow free pixels from pixels with factional snow cover, FN and FP are far more challenging issues that are important for understanding the value of a snow cover product. I’d rather only see F1 score in the graphs in figure 4 and the detailed performance analysis.

We thank the referee for this suggestion and will remove the accuracy metric from the manuscript.

We fully agree with the suggestion and confirm that the manuscript already implements what is proposed. We apologize for the lack of clarity in our original formulation. We propose to add a revised and clearer formulation of our methodology to assess MODIS in the manuscript.

We assume that the referee refers to the comparison of VNP10A1 vs. MOD10A1 performances (Figure 6) and possibly the comparison of VNP10A1 NDSI vs. Sentinel-2 FSC (Figure 7). We acknowledge that Figure 6 can be included in the results section. Hence, we propose to move the paragraph about the MODIS product evaluation to the results section and adjust the rest of the manuscript accordingly. We thank the referee for this suggestion that indeed allows us to highlight a result of interest for the community. We think that Figure 7 belongs to the Discussion, as we use it to investigate the reasons for the mixed performance of the products in mixed pixels, and it is more a perspective of our work.

Technical corrections:

- Suggest rewording sentence from line 18-20 for clarity.

We propose to reword this sentence.

- Line 35 - recommend changing “guaranteed” to “operationally planned” or similar, there is risk with any operational system that it does not meet expected mission lifetime and there can be a gap in coverage before a replacement is fielded.

We accept this remark and propose to include it in the manuscript.

- Line 67 - contradicts with Table 2 - reviewed and ASO snow depth data, not SWE was used for the validation. Suggest changing “which provides lidar based snow water equivalent (SWE) measure” to “which provides 3m spatial resolution lidar based snow depth measurements that were converted into binary snow masks.”

We propose to correct the manuscript and apologize for the oversight.

- Line 227 - after “we also report the F1 score “ suggest adding “which is the harmonic mean of omission and commission”.

We propose to include the referee’s suggestion and reorganize section 4.5 presenting first commission and omission errors and then F1 score in order to provide more meaningful insight on the proposed metrics.

References

Härer, S., Bernhardt, M., Siebers, M., and Schulz, K.: On the need for a time- and location-dependent estimation of the NDSI threshold value for reducing existing uncertainties in snow cover maps at different scales, The Cryosphere, 12, 1629–1642, https://doi.org/10.5194/tc-12-1629-2018, 2018.

Riggs, G. A. and Hall, D. K.: NASA VIIRS Snow Cover Products User Guide v3, 2023, 10.5194/tc-17-567-2023

Citation: https://doi.org/10.5194/egusphere-2026-1122-AC1

AC4: 'Reply on RC1', Nicola Imperatore, 24 Jul 2026

We realized that, in the reply we just posted, the table headers are not well visible. We apologize, there should have been an issue with formatting. Attached, you can find a version in pdf format of the answer, please refer to this for tables.

Citation: https://doi.org/10.5194/egusphere-2026-1122-AC4

RC2: 'Comment on egusphere-2026-1122', Anonymous Referee #2, 20 Jun 2026

This manuscript provides a useful and timely evaluation of operational VIIRS snow cover products against a high-resolution Sentinel-2 reference, with a well-designed stratification by land cover, topography, and sensor geometry. The work is relevant given the impending decommissioning of MODIS. However, there are several points that I believe warrant attention before the manuscript is published.

Section 5.1: Table 3 suggests that the choice of cloud mask has an effect on product performance that is comparable to that of the retrieval algorithm itself. The influence of the cloud mask on snow retrieval is therefore an important consideration, and it is closely tied to the well-known problem of cloud/snow discrimination. Since the VIIRS products evaluated here originate from different sources and apply different cloud masks, and since the present study does not explicitly isolate the effect of the cloud mask, this represents a limitation of the analysis. Where feasible, I would encourage the authors to compare and evaluate the accuracy of the cloud masks across the different product sources and to assess their influence on the computed snow cover area. At a minimum, this issue should be discussed.

Section 5.3: The multi-platform compositing clearly improves data availability, and the reported reductions in cloud cover are substantial. However, given that the three platforms share the same orbital plane with overpasses only about 50 minutes apart, it would be helpful to disentangle the mechanisms responsible for this improvement. Specifically, how much of the gain arises from cloud advection over the roughly 50-minute window, and how much from differing viewing geometries or other factors? A more important concern relates to Figure 5 and the comparability of the products shown. Section 3.1 states that NASA does not currently distribute a JPSS-2 product, yet Figure 5 presents single-platform series labelled MF-FSC-VNP-L3, MF-FSC-VJ1-L3, and MF-FSC-VJ2-L3. As I understand it, these single-platform series are all derived from the Meteo-France pipeline using one platform at a time, and are therefore not the same products (VNP10A1, VJ110A1) evaluated in Section 5.1. I suggest that the authors clarify explicitly how comparability is ensured between the single-platform and multi-platform composites, and between these and the NASA products evaluated elsewhere.

Section 6.1: The statement that Landsat-based products "are only available in the United States" is inaccurate, as Landsat provides global coverage. I suspect the authors intend to say that the specific high-resolution reference datasets or studies they cite are concentrated in the United States. The sentence should be reworded accordingly. In addition, since Landsat data are in fact globally available, the authors should explain why Landsat-based reference data were not used in this study.

Section 6.2: I have a substantive concern with the MOD10A1 versus VNP10A1 comparison. Terra (MOD10A1) is a morning-orbit platform (descending node around 10:30), whereas VIIRS daytime overpasses occur in the early afternoon (around 13:30), as does Aqua (MYD10A1). On the grounds of overpass-time matching, VNP10A1 is more appropriately compared with MYD10A1. The comparison with MOD10A1 remains informative but should be framed accordingly. This bears directly on the interpretation in the manuscript. The authors attribute the finding that MOD10A1 yields more snow than VNP10A1 to a less restrictive cloud mask. This attribution is likely incomplete. Because the morning overpass precedes much of the intraday accumulation of positive degree-hours, the morning platform may simply observe more snow than the early-afternoon platform owing to within-day melt, independent of cloud masking. Notably, the authors themselves observe that studies reporting the opposite result used MYD10A1, which indicates they are aware of the Terra/Aqua distinction. The same overpass-time effect is discussed in the China-based VIIRS evaluations they cite. I recommend that the authors add a comparison with MYD10A1.

Section 6.4: As the authors themselves demonstrate in Figure 7, the NDSI-FSC relationship differs markedly between forested and open areas. Yet the methodology (Section 4.1) applies a single linear relationship to convert NDSI to FSC everywhere. Applying one relationship across land cover types necessarily introduces a systematic error in forested pixels. I encourage the authors to strengthen the associated uncertainty analysis, for example by fitting separate NDSI-FSC relationships for forested and open terrain (or finer land-cover classes) and reporting the resulting change in accuracy. This would substantiate the discussion in Section 6.4.

Section 6.5: Topography exerts a strong influence on moderate-resolution snow retrievals, particularly in high mountains, and the discussion in this section is somewhat brief relative to the importance of the effect. The manuscript's own results, such as omissions on northern slopes and declining performance with increasing slope, are consistent with prior scale-effect studies of moderate-resolution snow retrieval (e.g., work over the Tibetan Plateau) that demonstrate a strong dependence of retrieval accuracy on slope and aspect. Importantly, such studies also show that the locally optimal NDSI threshold varies with slope and aspect because of illumination conditions, which directly challenges the use of a single, fixed NDSI-FSC relationship and threshold in this study. This connection, and its implications for the fixed-threshold approach, should at least be discussed.

Citation: https://doi.org/10.5194/egusphere-2026-1122-RC2

AC2: 'Reply on RC2', Nicola Imperatore, 24 Jul 2026

We would like to thank the referee for their careful examination of our work and the insightful suggestions they have provided. To address these comments, we have carried out several supplementary analyses, as described in the detailed response below. We believe the revised manuscript will satisfactorily address their concerns.

Note : During the revision process we identified and corrected two minor issues: 1. A coding error in the conversion from NDSI to FSC which only affected NDSI values below 0.1 in resampled NASA products. 2. Three corrupted products in our VNP10A1 dataset, corresponding to tile h18v04 from December 3–5, 2023. These products correspond to clear-sky acquisitions following a major snowfall event affecting a large number of relevant pixel-to-pixel correspondences

Tab. 3 of the original manuscript

Tab. 3 after corrections

F1-score	RMSE [%]
VNP10A1	4.48+06	18.7	0.93	0.02	0.06	0.42	10.09
VJ110A1	4.45+06	17.9	0.92	0.02	0.06	0.81	10.88

We propose to update all tables and figures of the manuscript to integrate these two corrections. The numerical results reported in this reply will include these corrections.

Hereafter, the point-by-point response. In italic, we report the comments of the referee.

Section 5.1: Table 3 suggests that the choice of cloud mask has an effect on product performance that is comparable to that of the retrieval algorithm itself. The influence of the cloud mask on snow retrieval is therefore an important consideration, and it is closely tied to the well-known problem of cloud/snow discrimination. Since the VIIRS products evaluated here originate from different sources and apply different cloud masks, and since the present study does not explicitly isolate the effect of the cloud mask, this represents a limitation of the analysis. Where feasible, I would encourage the authors to compare and evaluate the accuracy of the cloud masks across the different product sources and to assess their influence on the computed snow cover area. At a minimum, this issue should be discussed.

We agree with the referee that cloud masking is of paramount importance in a snow cover retrieval algorithm Here, we used the union of the cloud masks of reference and target datasets to limit the influence of cloud masking on the performance assessment. Cloud masking was considered during this study. For example, we tested some configurations of Météo-France cloud masking algorithm without finding relevant differences. However, a comprehensive investigation of cloud masking and how it affects snow detection accuracy require careful design and additional analysis that is well beyond the scope of this study. In particular, such a study would require a reference cloud mask that is not available. We can revise the manuscript to emphasize this limitation of our study.

Section 5.3: The multi-platform compositing clearly improves data availability, and the reported reductions in cloud cover are substantial. However, given that the three platforms share the same orbital plane with overpasses only about 50 minutes apart, it would be helpful to disentangle the mechanisms responsible for this improvement. Specifically, how much of the gain arises from cloud advection over the roughly 50-minute window, and how much from differing viewing geometries or other factors?

We thank the referee for this insightful comment. In the manuscript, we attributed the improvement to cloud advection but a dedicated analysis would be needed to disentangle the contributions of cloud advection, differing viewing geometries, and other potential mechanisms. For cloud advection, this would involve tracking cloud feature displacement between consecutive overpasses (50 minutes apart) using wind speed data to estimate expected movement. As shown by Li et al., 2025, wind speeds are typically on the order of 10 m/s, corresponding to a displacement of ~30 km between passes. For viewing geometry, this would require comparing cloud cover statistics across platforms while accounting for differences in solar and viewing angles. The VIIRS cloud shadow detection algorithm is based on the propagation of the line of sight (Huchinson et al., 2009); the shadow projected by a cloud is proportional to its height. We can suppose this to be of the same order of magnitude as cloud height (5–10 km), which is confirmed by Fig. 3 of (Hutchinson et al., 2009), where cloud shadows are on the order of 10 km. Based on this preliminary estimate, we hypothesize that cloud advection is the dominant mechanism. However, under certain conditions, the contribution due to viewing and solar geometry may be more significant. To conclude, a detailed attribution analysis would require additional data (e.g., high-temporal-resolution atmospheric fields) and modeling (e.g. radiative transfer) that falls beyond the scope of our current study. We can revise the manuscript to explicitly acknowledge this limitation and discuss how both mechanisms likely contribute to the observed improvements.

A more important concern relates to Figure 5 and the comparability of the products shown. Section 3.1 states that NASA does not currently distribute a JPSS-2 product, yet Figure 5 presents single-platform series labelled MF-FSC-VNP-L3, MF-FSC-VJ1-L3, and MF-FSC-VJ2-L3. As I understand it, these single-platform series are all derived from the Meteo-France pipeline using one platform at a time, and are therefore not the same products (VNP10A1, VJ110A1) evaluated in Section 5.1. I suggest that the authors clarify explicitly how comparability is ensured between the single-platform and multi-platform composites, and between these and the NASA products evaluated elsewhere.

We thank the reviewer for his comment and acknowledge that, given the multiple processing steps involved, the manuscript can be hard to follow when it comes to understand which products are compared and why. We propose to add a table in the revised manuscript to provide a synthetic overview of the assessed products and how they compare in terms of satellite platform and processing pipelines. We provide an example of such a table for one product (to be filled with each of the evaluated products).

Product ID	Sensor	Platform	Pipeline	Reference	Detailed analysis
VNP10A1	VIIRS	SNPP	NASA	(Riggs et al., 2023)	Yes

Section 6.1: The statement that Landsat-based products "are only available in the United States" is inaccurate, as Landsat provides global coverage. I suspect the authors intend to say that the specific high-resolution reference datasets or studies they cite are concentrated in the United States. The sentence should be reworded accordingly. In addition, since Landsat data are in fact globally available, the authors should explain why Landsat-based reference data were not used in this study

In Section 6.1, we state that using Sentinel-2 FSC operational products is a scalable and efficient method for generating reference data for moderate-resolution sensor assessment. We stress the fact that, unlike most studies, we use directly operational FSC products as reference rather than deriving snow cover from reflectance. This allows us to avoid an additional pipeline and quality evaluation for ground truth data processing. We apologize for any misunderstanding caused by the lack of clarity and propose to reformulate Section 6.1 to explain our position.

Section 6.2: I have a substantive concern with the MOD10A1 versus VNP10A1 comparison. Terra (MOD10A1) is a morning-orbit platform (descending node around 10:30), whereas VIIRS daytime overpasses occur in the early afternoon (around 13:30), as does Aqua (MYD10A1). On the grounds of overpass-time matching, VNP10A1 is more appropriately compared with MYD10A1. The comparison with MOD10A1 remains informative but should be framed accordingly. This bears directly on the interpretation in the manuscript. The authors attribute the finding that MOD10A1 yields more snow than VNP10A1 to a less restrictive cloud mask. This attribution is likely incomplete. Because the morning overpass precedes much of the intraday accumulation of positive degree-hours, the morning platform may simply observe more snow than the early-afternoon platform owing to within-day melt, independent of cloud masking. Notably, the authors themselves observe that studies reporting the opposite result used MYD10A1, which indicates they are aware ofthe Terra/Aqua distinction. The same overpass-time effect is discussed in the China-based VIIRS evaluations they cite. I recommend that the authors add a comparison with MYD10A1.

The referee's comment made us further investigate this point. We modified Figure 6 to show the evolution of the cloud-covered area, and we added the Aqua product MYD10A1 to every panel to allow a comparison with Terra’s MOD10A1. We attach the result of such a comparison, adding Aqua to Fig. 6 of the original manuscript (filename fig06_aqua.pdf).

We observe (panel b) that more clouds are detected in VNP10A1, except for the month of December where there is a notable difference in snow-covered area. This means that our interpretation that the larger snow cover area of MODIS (Terra) is due to a less restrictive cloud mask is indeed insufficient and will be removed from the manuscript. Furthermore, MYD10A1 (Aqua, 13:30) yields less snow and more cloud cover than both VNP10A1 (VIIRS SNPP, 13:30) and MOD10A1 (Terra, 10:30) from December to March (panels a and b). This comparison confirms the results of Thapa et al., 2019 and Liu et al., 2022 in North America and China, respectively. It also suggests that the overpass time is not the main driver of the differences in snow-covered area. Finally, in terms of bias and RMSE, the behavior of Aqua is more similar to Terra than to SNPP. This finding strengthens our discussion, confirming the need to examine systematic differences if MODIS and VIIRS products are to be merged into a long-term record, and showing that NASA’s VIIRS products slightly outperforms their MODIS counterparts in terms of RMSE. We propose to follow the referee’s suggestion, i.e., to add MYD10A1 in the comparison, and adjust the rest of the manuscript accordingly.

Section 6.4: As the authors themselves demonstrate in Figure 7, the NDSI-FSC relationship differs markedly between forested and open areas. Yet the methodology (Section 4.1) applies a single linear relationship to convert NDSI to FSC everywhere. Applying one relationship across land cover types necessarily introduces a systematic error in forested pixels. I encourage the authors to strengthen the associated uncertainty analysis, for example by fitting separate NDSI-FSC relationships for forested and open terrain (or finer land-cover classes) and reporting the resulting change in accuracy. This would substantiate the discussion in Section 6.4.

We thank the referee for this insightful remark and have performed a test to assess the proposition. We compared VNP10A1 FSC derived using the (Salomonson and Appel, 2006) fit (as used elsewhere in the manuscript) with a version obtained using the fit—different for open areas and forests—developed in this work (Fig. 7). We illustrate the differences by reporting Table 3 (for VNP10A1) and 4 of the manuscript in the two cases. Please note that the results have minor differences from the original manuscript for the reasons stated at the beginning of this document.

Tab. 3 (VNP10A1) and 4 of the manuscript, NDSI to FSC conversion with (Salomonson and Appel, 2006)

VNP10A1

4.48+06

18,7

0.93

0.02

0.06

0.42

10.09

1-99 %	Forest	N	4.99e+04	39.58	0.71	0.20	0.28	-1.38	29.72
1-99 %	Forest	S	1.28e+04	43.83	0.80	0.15	0.20	-1.86	22.97
1-99 %	open	N	6.58e+04	68.07	0.90	0.17	0.12	-1.74	23.06
1-99 %	open	S	5.76e+04	65.63	0.93	0.17	0.05	3.31	18.01
0-100 %	Forest	N	3.18e+05	8.41	0.76	0.02	0.24	0.42	13.63
0-100 %	Forest	S	2.02e+05	4.32	0.87	0.01	0.14	-0.13	6.42
0-100 %	open	N	2.44e+05	36.10	0.95	0.02	0.07	-0.64	13.00
0-100 %	open	S	3.69e+05	18.27	0.96	0.01	0.03	0.55	7.43

Tab. 3 (VNP10A1) and 4 of the manuscript, NDSI to FSC conversion with the fit found in this work:

Open areas : FSC = 1,43 NDSI - 0,04 Forest areas : FSC = 2,17 NDSI – 0,13

VNP10A1

4.48+06

18,7

0.93

0.02

0.06

0.51

10.81

1-99 \%	Forest	N	4.99e+04	39.58	0.71	0.27	0.23	4.89	33.93
1-99 \%	Forest	S	1.28e+04	43.83	0.81	0.20	0.14	4.88	26.14
1-99 \%	open	N	6.58e+04	68.07	0.90	0.15	0.13	-3.96	23.32
1-99 \%	open	S	5.76e+04	65.63	0.93	0.16	0.05	1.27	17.69
0-100 \%	Forest	N	3.18e+05	8.41	0.75	0.03	0.20	1.43	15.23
0-100 \%	Forest	S	2.02e+05	4.32	0.87	0.01	0.09	0.38	7.02
0-100 \%	open	N	2.44e+05	36.10	0.94	0.02	0.07	-1.41	13.19
0-100 \%	open	S	3.69e+05	18.27	0.96	0.01	0.03	0.16	7.32

These results indicate that our linear fit slightly degrades overall performance with respect to the universal linear fit. The RMSE obtained with the two formulas is comparable in open areas, while it degrades in the forest when using the linear model estimated in this work. The bias also increases in forested areas when applying our fit. This highlights the difficulty in developing a predictive model for NDSI-to-FSC conversion, which performs well over large areas and in different seasons. We propose to add these considerations to the manuscript.

Section 6.5: Topography exerts a strong influence on moderate-resolution snow retrievals, particularly in high mountains, and the discussion in this section is somewhat brief relative to the importance of the effect. The manuscript's own results, such as omissions on northern slopes and declining performance with increasing slope, are consistent with prior scale-effect studies of moderate-resolution snow retrieval (e.g., work over the Tibetan Plateau) that demonstrate a strong dependence of retrieval accuracy on slope and aspect. Importantly, such studies also show that the locally optimal NDSI threshold varies with slope and aspect because of illumination conditions, which directly challenges the use of a single, fixed NDSI-FSC relationship and threshold in this study. This connection, and its implications for the fixed-threshold approach, should at least be discussed

We agree with the referee on the importance of topography and the suboptimality of threshold approaches for snow cover retrieval over large and heterogenous areas. We propose to extend Section 6.5 to discuss this aspect.

References

Li, Jun, David Santek, Zhenglong Li, Agnes Lim, Di Di, Min Min, Christopher Velden, and W. Paul Menzel. "Tracking Atmospheric Motions for Obtaining Wind Estimates Using Satellite Observations—From 2D to 3D", Bulletin of the American Meteorological Society 106, 2 (2025): E344-E363, doi: https://doi.org/10.1175/BAMS-D-24-0027.1

Hutchison, K. D., R. L. Mahoney, E. F. Vermote, T. J. Kopp, J. M. Jackson, A. Sei, and B. D. Iisager, 2009: A Geometry-Based Approach to Identifying Cloud Shadows in the VIIRS Cloud Mask Algorithm for NPOESS. J. Atmos. Oceanic Technol., 26, 1388–1397, https://doi.org/10.1175/2009JTECHA1198.1.

Citation: https://doi.org/10.5194/egusphere-2026-1122-AC2

AC3: 'Reply on RC2', Nicola Imperatore, 24 Jul 2026

We realized that, in the reply we just posted, the table headers are not well visible. We apologize , there should have been an issue with formatting. Attached, you can find a version in pdf format of the answer, please refer to this one for tables.

Citation: https://doi.org/10.5194/egusphere-2026-1122-AC3

Nicola Imperatore, Simon Gascoin, Matthieu Lafaysse, Marie Dumont, Adrien Mauss, Stéphane Guével, and Jean-Baptiste Hernandez

Data sets

Data of the article "Assessing VIIRS constellation seasonal snow cover on French mountains with Sentinel-2" Nicola Imperatore, Simon Gascoin, Matthieu Lafaysse, Marie Dumont, Adrien Mauss, Stéphane Guével, and Jean-Baptiste Hernandez https://doi.org/10.5281/zenodo.18157029

Model code and software

Code of the article "Assessing VIIRS constellation seasonal snow cover on French mountains with Sentinel-2" N. Imperatore https://doi.org/10.5281/zenodo.18773106

Nicola Imperatore, Simon Gascoin, Matthieu Lafaysse, Marie Dumont, Adrien Mauss, Stéphane Guével, and Jean-Baptiste Hernandez

Viewed

Total article views: 1,434 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
713	647	74	1,434	59	92

HTML: 713
PDF: 647
XML: 74
Total: 1,434
BibTeX: 59
EndNote: 92

Views and downloads (calculated since 13 Mar 2026)

Month	HTML	PDF	XML	Total
Mar 2026	539	388	64	991
Apr 2026	90	89	1	180
May 2026	30	111	2	143
Jun 2026	11	12	2	25
Jul 2026	42	46	5	93
Aug 2026	1	1	0	2

Cumulative views and downloads (calculated since 13 Mar 2026)

Month	HTML	PDF	XML	Total
Mar 2026	539	388	64	991
Apr 2026	90	89	1	180
May 2026	30	111	2	143
Jun 2026	11	12	2	25
Jul 2026	42	46	5	93
Aug 2026	1	1	0	2

Viewed (geographical distribution)

Total article views: 1,386 (including HTML, PDF, and XML) Thereof 1,386 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 02 Aug 2026

Short summary

Daily satellite snow cover observations are beneficial for water management in mountain regions, but it's crucial to understand their uncertainties. We evaluate the next-generation NASA constellation in mountainous terrain using higher-resolution satellite imagery as reference. We find an overall consistent performance across the platforms and combine their data to reduce cloud cover. These results support a confident transition and open new opportunities in snowpack modeling.


Total:	0
HTML:	0
PDF:	0
XML:	0