The paper has established a daily sun duration dataset at 5 km×5 km resolution in China from 2016 to 2023. This dataset can provide a supplement for current solar radiation research and is of great importance for socio-economic application. 

I have two main questions on this paper. Firstly, why do you focus on sun duration instead of incident solar radiation at land surface? In which sector or subject, sun duration is indispensable not solar radiation? I think this should be highlighted in introduction and also should be discussed in detail in discussion. Secondly, is there any released sun duration dataset? If not, why researchers not establish such dataset? What is the bottleneck in theory and in technology? If yes, please compared your results with other similar datasets.

I do have several specific reviews and suggestions as follows.

1. Why do you choose Himawari AHI? Compared to other shortwave radiation product, what is superiority of this product?

2. Please give more information on the algorithm, accuracy and bias of Himawari AHI Level 3 shortwave radiation in 2.1 section. 

3. Lines 73-75: Do you want to describe the calculation for surface solar radiation? If yes, I do think it can be obtained by using TOA solar radiation minus solar radiation attenuated amount by atmosphere. 

4. Line 107: I do think the citation should be (Angstrom, 1924).

5. Why do you keep four decimals for R but one for RMSE?

6. Please add significance test when you showing R.

7. Why do you fix the parameter of a and b across the whole China. Is there no spatial variations in the two parameters?

8. Figure 3: Why there are many SD estimations with different values (close to 12) when observation is zero?

9. Line 178: Why estimated SD less than 0 happens? You should discuss the reason in detail in discussion.

10. Figure 9: I am not sure the correlation coefficients is for spatial pattern or for temporal variation. 

11. Figure 10: Why only choose 28 September 2016 as a case study? Once you have long-term MODIS AOD data, please try to analyze the relationship between simulation performance and AOD level for whole year or four seasons.

12. Once the value you showed have units, please give. Such as in Figure 11.

13. At last, you should show the spatial and temporal variations in SD with your datasets.

14. Please discuss the simulation bias in SD dataset. Please also discuss the limitation of the current SD dataset.

Sunshine data is generally used to estimate global radiation, rarely the other way round, so the purpose and significance of this paper is unclear.

Secondly, the author mentions the shortcomings of the Himawari-8 official radiation data in the introduction, so it is confusing to use such a flawed data as the basis for the SD estimation.