General comments:

The researchers accomplished SIF retrieval from the Sentinel-5P TROPOMI mission and provided data with high quality, which expands the application of TROPOMI data in vegetation monitoring. The methods and materials used in the manuscript are reasonable and described in detail, which can support the publication of the dataset. The dataset is accessible and complete, the quality value and retrieval error of the data were fully evaluated. The TROPOSIF product has a high consistency compared with previous SIF products, and the results are reliable. There are only a few questions to be discussed.

Specific comments:

1. Line 37 to 42: Since Caltech's TROPOMI SIF product has been proven to be effective, the gaps in previous research and the purpose of this research should be more clearly stated.
2. Line 88 to 90: To what extent can the influence of the atmosphere be considered negligible? Can you add a comparison to compare the retrievals using calculated effective atmospheric transmittance and transmittance set as 1?
3. Figure 8: Previous SIF products have shown a tendency that SIF magnitudes decrease with narrower fitting windows toward longer wavelengths near the far‐red fluorescence peak and in fitting windows with less water vapor absorption (Parazoo, 2019), which is inconsistent with the results shown in Figure 8, how do you account for this?

Parazoo, N. C., Frankenberg, C., Köhler, P., Joiner, J., Yoshida, Y., Magney, T., Sun, Y., and Yadav, V.: Towards a Harmonized LongTerm Spaceborne Record of Far-Red Solar-Induced Fluorescence, Journal of Geophysical Research: Biogeosciences, 124, 2518–2539.

1. Line 393 to 398: The use of the 735-758nm fitting window is a feature of this research, but the limitations of this window were also stated. Is it possible to select one of the retrievals from the two fitting windows for each observation according to several indicators (e.g. cloud fraction threshold) to merge the retrievals from the two fitting windows and maintain the advantages of both, rather than providing two separate datasets?

Technical corrections:

1. Figure 1: The “FT” in the figure note is inconsistent with the abbreviation “FW” in the figure.
2. Figure 2: Only the weights of the first 8 singular vectors are shown in the figure, which is inconsistent with the figure note.

The paper describes a global dataset of Sun-Induced chlorophyll Fluorescence (SIF) obtained by the Sentinel 5P-TROPOMI mission. I believe this dataset is of great value to the community, and I want to express my gratitude to the authors for their diligent work.

L28: In addition to the GPP related studies, I suggest mentioning also the value of SIF for hydrological studies, e.g., on ecosystem transpiration and water limitation (see, e.g., Maes et al. Remote Sensing of Env. 2020; the review of Jonard et al. Agric. Forest Meteo. 2020; Gonsamo et al., Remote Sensing of Env. 2019).

L50-55: It might be interesting to mention the Fluorescence Correction Vegetation Index by Yang et al. Remote Sensing of Env. 2020 here. This is, similar to NIRv, an index to assess the effect of canopy structure. I’m not entirely sure whether their framework is applicable to TROPOMI data, but it’s certainly worth mentioning.

L54-56: I have a few annotations to make on the sentence about SIF yield, used for monitoring vegetation stress. First, it is important to highlight the difference between SIF yield and fluorescence yield. SIF yield is a canopy-scale variable, affected by both canopy structural and leaf biochemical aspects. SIF yield is relatively easy to retrieve from satellites. The variable that is however directly impacted by a plant’s stress status is the photosystem scale fluorescence quantum yield_. This variable is fundamentally different from SIF yield as it does not depend on variables such as chlorophyll content, leaf area index, or leaf orientation.  Measuring the latter variable from satellite imagery is not self-evident. Celesti et al. Remote Sensing of Env. 2018 provides a framework for retrieving the fluorescence quantum yield at the canopy scale. Please clarify how you define the SIF yield and how it can be compared to photosynthetic efficiency?

L56: I understand the point that you want to make here, but I suggest being more careful in the wording. While Dechant et al. indeed showed that NIRv is a better predictor for GPP compared to SIF, I am not sure to which extent this idea holds in case of a stress situation. SIF can be decomposed into the following factors:  SIF = PAR ⋅ fPAR ⋅ Phi_f ⋅ f_esc

NIRvP only takes into account PAR⋅fPAR⋅f_esc. This does not consider variations in Phi_f despite the latter’s sensitivity to short-term stresses. This is why Dechant et al. raises the point that NIRvP might show significant disagreements with SIF in case of short-term stresses, such as droughts or heatwaves. We believe that NIRvP bears the potential to serve as a ‘potential SIF emission’ or ‘reference SIF emission’ (similar to reference evapotranspiration using the Penman-Monteith method). The ratio between NIRvP and SIF could then serve as a stress factor, as it more or less isolates Phi_f. It is however worth noting that there is not yet any experimental evidence supporting this claim, despite the seemingly simple logic behind it.

L80: It would be good if the authors could provide a brief explanation of the meaning of a and α.

L279: Section 3.3 provides an interesting comparison between SIF products. It describes a difference in the SIF signal depending on which bands have been used for the retrieval. It describes that OCO-2 and Sentinel 5P-TROPOMI data are comparable. I wonder to which extent the SIF data from the dataset here will be comparable to FLEX-derived SIF data, as it will not be based on the exact same bands. Do you have an idea on that?

E.g. Figure 9 and text below: Information on the interpretation of the negative SIF values is missing. Could you please explain the reason why there are negative SIF values and what their physical meaning is? It would be good to know for future users whether these negative SIF data should be considered or not?