**5. Conclusions**

This study investigated the potential of using S-2 data to assess moderate-resolution EUMETSAT H-SAF product of snow extent (H10) and FSC (H12) in Finland, the Italian Alps and Turkey. Snow masks derived from S-2 imagery have revealed a significant consistency with both ground-based snow measurements (POD = 0.82, FAR = 0.08,) and in-situ webcam photography, revealing a RMSE of about 12%, in terms of FSC. Hence the reliability of assuming this high-resolution dataset as a reference for intercomparison purposes. The results obtained in this study reveal that S-2 data can be properly used to continuously assess these medium resolution satellite snow products, which have been commonly validated against in-situ data so far [53,56]. However, it is noteworthy to consider that under specific conditions the snow mapping derived from S-2 data can be affected by critical flaws. Indeed, the analysis of camera images in the Italian Alps has shown that dense cloud cover can undermine the reliability of S-2 snow masks, mainly when patchy snow cover is present. Furthermore, during melting period the widespread presence of meltwater over flat areas may lead to an overestimation of snow cover.

The results of the cross-sensor comparison prove that the analyzed H-SAF snow products are highly consistent with S-2 imagery in detecting snow, also in terms of FSC, generally in compliance with the products requirements [104,105]. Nevertheless, the analyzed satellite datasets generally reveal a higher agreemen<sup>t</sup> over flat/forested areas (PODH10 and RMSEH12 equal to 0.98 and 0.15, respectively) than in mountainous regions (over Italian Alps, PODH10 and RMSEH12 equal to 0.78 and 0.33, respectively). Indeed, the local complex topography is likely to significantly hinder snow detection over mountain sides at coarser satellite spatial resolution. Conversely, the vegetation cover has turned out to have a less relevant impact on the consistency among remotely-sensed observations, even in the presence of dense evergreen forest. In Finland the long-lasting snow interception on vegetation canopy is expected to contribute to strengthening the agreemen<sup>t</sup> between S-2 snow maps and H12 images during the winter season. However, further key issues need to be addressed in the future. Primarily, a comparative study on different retrieval algorithms would allow an assessment of the reliability of the snow mapping derived from S-2 imagery. Secondly, the impact of cloudiness on the consistency among remotely-sensed observations should be investigated in more details through the analysis of scenes with different cloud cover percentages.

Nevertheless, these promising results currently encourage the effective use of the analyzed H-SAF snow products for hydrological and climatological studies, since they provide reliable snow-related information at large scale. Furthermore, thanks to their free availability at a daily scale, both H10 and H12 products are recommended as particularly suited for operational applications.

**Author Contributions:** All co-authors contributed to the design of this study. Validation algorithms were developed by C.M.T., G.P., A.N.A., Z.A., S.K., B.S., S.G., and M.T. Collection and validation of satellite data were performed by C.M.T. and B.S. in Finland, G.P. and C.M.T. in Italy, Z.A. and S.K. in Turkey. G.P. wrote the paper with contributions of C.M.T., S.K. The manuscript was reviewed by A.N.A., Z.A., C.M.T., S.K., S.G., B.S.

**Funding:** This research received no external funding.

**Acknowledgments:** The research has been conducted in the framework of the EUMETSAT H-SAF Project, thanks to the collaboration among several partner institutes of the validation cluster of snow products. We wish to thank all the institutions for supporting and promoting these research activities. We are primarily grateful to the European Cooperation in Science and Technology (COST), which has funded a Short Term Scientific Mission (STSM) of key importance for the success of this collaboration through the ES1404 Harmosnow Action. The work has been also supported by the Italian National Department of Civil Protection. The authors would like to acknowledge the Environmental Protection Agency of Aosta Valley (Italy) for providing Italian webcam datasets used in this study. We would like to address special thanks to Edoardo Cremonese for the fruitful discussions on this matter.

**Conflicts of Interest:** The authors declare no conflict of interest.
