The described surface solar radiation climate data records (CDR) product SARAH is important and helpful for understanding the climate system, model evaluation, renewable energy application, etc. This preprint introduces a new version, SARAH-3, which updates the  CDRs and motivates and explains new updates in the applied methods and their impact. The preprint is very well written and illustrated, but I have some questions and suggest some minor improvements to be made.

- Abstract: the reader would like to see the covered region here already (SARAH is not global)

- Abstract: Name the seven parameters of Climate Data Records explicitly here for clarity, as you mention Interim CDRS later in the abstract. 

- Introduction: for some applications, it might be helpful to use a fitting top-of-atmosphere CDRs dataset. Can the authors suggest a data set?

- Line 78: -> "SARAH-3 paramters, abbr & units". It sounds strange that "units are included" in the dataset.

- Sec. 2.1: Is there no reference available for HelSnow? 

- Line 114: Why no units? What is 160? I guess pixels, but guessing is risky. So, is the displacement speed more than 160 pixels/30 min in the case of MVIRI?

- Figure 3: units? What is optical flow (a term from image processing?)? Displacement? 

- Sec. 2.1.: Is snow ageing and thus the change of snow albedo of relevance? Even considered?

- Line 146: How much would the results degrade if ERA5 snow-cover were used (after interpolation)? In other words, what is the quantitative added value of HelSnow?

- Line 187: SID = ??? How derived?

- Do you use ERA5-Land snow cover? The ERA5_Land snow cover does not assimilate snow observations! ERA5 does, but not in complex terrain. 

- Sec. 2.5.4: Is a change in aerosol concentration over time considered? MACC does not cover the entire SARAH period? How can you discuss trends without including AOD change? A reference for MACC?

- Table 3: Strassburg -> Strasbourg?

- Line 379: "and its functions"?

- Figure 11: Absolute Bias is called MAD elsewhere? 

- The references list needs to be sorted and, therefore, difficult to check.

This article presents a concise and well-written overview of the theoretical basis and updates of the latest release of the SARAH climate data record. The article is a very useful source of reference information for potential users of this dataset and falls well within the scope of ESSD. There are a few minor points (in addition to/partly overlapping with the points mentioned by the other reviewers), which I ask the authors to address prior to publication.

List of minor points:

* Abstract: language- and information-wise, I think that the abstract can still be improved. Specifically, I suggest to mention all parameters included in SARAH-3 (or at least the newly added ones?) and to explicitely mention the start data of the dataset. I do also think that the term SARAH-3 is used to frequently. The mention of „ICDR / near-real time processing“ directly before the statement „enabling climate monitoring applications“ seems mis-leading, as the near-real time extension of the data record is not really crucial for climate applications. The phrase „The SARAH.3 climate analysis reveals“ seems la strange start of this senetence (is this a „standardized“ analysis?). „good accuracy and stability“: can you give quantitative numbers here, e.g. for SSR?

* ICDR: I think the distinction between the CDR and ICDR needs to be described in more detail, including giving some guidance for users of this data record. While the start of the ICDR period is mentioned twice, this is still somewhat hidden in the description of meta-data, and I recommend to dedicate a paragraph to this at the start of Section 2. What is the plan/time-line for updating periods now covered by the ICDR to CDR-quality processing? If one is interested in climate trends, when should one wait for the availability of the CDR instead of calculating trends based on the spliced CDR/ICDR?

* L55: „All CM SAF data records are freely available without restrictions.“ While I am not a lawyer, I do not think this statement is true. Specifically, the clause to acknowledge EUMETSAT in the CMSAF license is a restriction as far as I can see (even if a very weak one), or not?! Please check/correct this aspect. (see https://cds.climate.copernicus.eu/api/v2/terms/static/eumetsat-cm-saf.pdf )

* As mentioned in the introduction, homogeneity is a key goal for CDRs. Do you see any impact of transitioning to MSG? Can you also discuss aspects where sensor limitiation affect the algorithm choices. I guess you could have used the 1.6um channel for snow detection from SEVIRI. Does the difference in spatial resolutions between Metesat first and second generation have an impact?

* I found it surprising that optical flow is used to identify snow in HelSnow, by an exclusion logic (i.e. if it moves, it cannot be snow). Did you consider using a constraint on temporal constancy instead (if reflectivity does not change significantly in time, it must be snow)? In addition, what window size and other parameter settings are used for inferring the optical flow?

* While I know PAR, I did not know the term DAL before. Can you please add some additional explanations what these parameters are used for/what spectral weighting is used?

* Figure 9: I suggest to replace the SARAH2 image by a difference image (or add this as 3rd panel). I found it hard to quickly identify differences and think a difference image would help to highlight regions with significant changes.

* Larger context and outlook: I missed a reference to the comprehensive review of Huang et al., 2019 (*). For readers, it could provide additonal context, in particular with respect to your method and their section on „Current problems“. Can you add some thoughts on future improvements of SARAH? How will SARAH-3 be affected by the transition to MTG? (*) https://doi.org/10.1016/j.rse.2019.111371