This manuscript produced a long-term and high-resolution global gridded PAR product based on the latest ISCCP-HXG cloud product and reanalysis data. PAR data is required for researches in the ecological, agricultural, and global change fields. The algorithm used to estimate PAR in this study is the physical scheme that used in the previous study of Tang et al. (2017), and was proven to be a more accurate algorithm than previous ones. The produced gridded PAR product was evaluated against surface observations collected at more than 80 experimental stations worldwide. Compared with the well-known CERES PAR product, the PAR product produced in this study was found to be a more accurate dataset with higher resolution. The topic is highly interesting and appropriated for ESSD. The paper is clear and well written. Therefore, I recommend its publishing on the ESSD after answering the following several minor issues.

1. In this study, aerosol data from the reanalysis data MERRA-2 was used? Why not use the satellite-based aerosol products with higher accuracy?
2. Cloud top temperature was used to discriminate the water and ice cloud in this study. Since MODIS has a cloud top temperature product, and why not use this product?
3. In section of in-situ measurements, seven experimental stations from SURFRAD, 42 experimental stations from NEON, and 38 experimental stations from CERN were used to evaluate the performance of the estimated PAR. Did you do quality control on these observations and what are the criteria for control?

The paper by Tang et al. produced a 35-year (1984‒2018) high-resolution (3 h, 10 km) global gridded PAR dataset using ISCCP, MERRA-2, ERA5, MODIS and CLARRA-2 products as inputs in a physical-based model. In the paper, authors compared their instantaneous and daily PAR products against surface experimental stations, including SURFRAD, NEON, CERN networks, and CERES products. The results prove that the PAR product was found to be a more accurate dataset with higher resolution. This study could be a good supplement to meet refined analysis and understand climate variations. I believe this study can be valuable to the relevant community.

Overall, the study describes the background and introduction, and methods in a comprehensive way. However, the method in the paper contains inadequate innovations. Therefore, I would encourage the authors to submit a revised manuscript by addressing my specific comments below:

• First of all, to ensure the accuracy of surface network data, Does the surface network data used in the paper undergo the data preprocessing and the removal process of invalid values？
• How to compare ground-based stations with satellite pixels, the comparison process could introduce errors in the results.
• To some extent, the accuracy of the parameterized method used in the paper depends on the accuracy of the input data. The descriptions of the accuracy of the input data were insufficient.
• The algorithm used to map global gridded PAR in this study was the parameterization method developed by Tang et al. (2017), so the method used in the paper lack some innovations, more descriptions of Highlights are need in the paper.
• Is the band range of the PAR observations derived from the ground station consistent with the estimated PAR?