The manuscript “Reconciling different approaches to quantifying land surface temperature impacts of afforestation using satellite observations” by Wang et al presented thoughtful analyses regarding three different types of temperature effects of forestation that appeared in the literature (potential vs actual), and trying to explain the causes of the different magnitudes. The research is a nice addition to the literature on this topic as it is helpful to clarify the interpretation of different results.

Major comments:

First, I disagree with the authors' interpretation of these results and the claim that the causes of the different estimates are unknown. On the contrary, spatial scale or fractions of forest change matters for interpreting the temperature impact, which has been considered in previous studies. Taking the influential work cited by the authors as an example: 

In Alkama 2016, the fraction of forest cover change is explicitly taken into account, and the results clearly indicated that the temperature effect depended on the fraction of change.  

In Li 2016, the fractional dependency has been reported: “It should be noted that the estimated impacts also depend on the thresholds used to define forest cover change, as discussed in section 2.2. The sensitivity analysis shows that a higher threshold to define forest change leads to stronger impacts on temperature.”

In Duveiller 2018, they used the temperature effect of 100% conversion to avoid the influence of fractional changes. 

The strength of this work is that it explicitly addressed this question. Perhaps the authors could consider an alternative title better reflecting this point.  

Second, the main finding is that the fraction of forestation (complete vs incomplete) explains the different magnitude of temperature effects. Fraction could indeed have a strong influence on the temperature signal. But it is not the only one. Other factors such as the timing of land cover change, length of the study period, and the spatial extent of forest cover change impact may also contribute. (1) Taking the timing of de-/forestation as an example, if the change happened in the different years of the two periods of 2002–2004 (t1) and 2010–2014 (t2) (L277), changes in 2002 and 2010 would produce a larger temperature change compared to changes in 2004 and 2014, depending on whether the change signals lasted full three years or just the last year. (2) More importantly, the space-for-time assumption is acceptable but it is not strictly true in reality. The adjacent two sites did not share the same climate condition (see Chen 2016). This also contributes to the different temperature effects. (3) When the spatial extent of forest change is large, the local and nonlocal temperature effect appear with heterogeneity which confounds the estimation of the local temperature. (4) The consistency between the actual and potential effect is also scale-dependent. At small scales (e.g., 10m resolution), it would be easier to achieve full change compared to large scales (1km). Therefore, the differences in fractional change alone cannot fully reconcile the observed differences. 

Third, I feel the language of this manuscript should be improved and polished. 

Specific comments:

L102-103 They may not assume 100$ complete ground coverage. They used the defined forest and nonforest in the paper. Of course, due to inherent scaling and the mixed pixel issue in remote sensing, the defined pixels cannot be 100% pure at a given scale. I think many studies were aware of this issue but they did not explicitly address it. 

L161-162: How are the afforestation and adjacent control pixels defined?

L518: What do you mean “extensive variable”?

L549 to 551: For this fractional dependency, it has been reported in such as Li 2016 

L572-573: The actual and potential effect is also scale-dependent, and so is the feasibility of full afforestation in reality. Fully afforested could be easily achieved for a small pixel 30m. And for this pixel, the potential and actual could be similar following the findings of this work. At larger scales, it is more difficult to become “fully” afforested, which leads to larger differences between potential and actual impacts. Therefore, whether “achieving the full cooling potential” is scale-dependent. 

L581-583: I disagree with the authors on this. The potential effect is useful as it measures the possible outcome of full conversion or mostly afforested (depending on resolution and scale), and whether it is realized depends on the fraction of the change. One can take into account the fractional change to convert the potential effect to more reasonable estimation. At least for this reason, it is not misleading. It is about different interpretation and clarification is needed. 

L602 to 605: I don’t agree this statement because both the actual and potential effects are scale dependent. Without mentioning the scale, it is incorrect. 

References 

Alkama, R., & Cescatti, A. (2016). Biophysical climate impacts of recent changes in global forest cover. Science, 351(6273), 600–604. https://doi.org/10.1126/science.aac8083

Chen, L., & Dirmeyer, P. A. (2016). Adapting observationally based metrics of biogeophysical feedbacks from land cover/land use change to climate modeling. Environmental Research Letters, 11(3), 034002. https://doi.org/10.1088/1748-9326/11/3/034002

Duveiller, G., Hooker, J., & Cescatti, A. (2018). The mark of vegetation change on Earth’s surface energy balance. Nature Communications, 9(2018). https://doi.org/10.1038/s41467-017-02810

Li, Y., Zhao, M., Mildrexler, D. J., Motesharrei, S., Mu, Q., Kalnay, E., … Wang, K. (2016). Potential and Actual impacts of deforestation and afforestation on land surface temperature. Journal of Geophysical Research: Atmospheres, 121(24), 14372–14386. https://doi.org/10.1002/2016JD024969

The biophysical effects of deforestation/afforestation have drawn a lot of attention in the past few years. However, the results are not very consistent among different studies using different products and methods. The authors revealed the methodological differences among different studies and summarized them into one actual and two potential temperature effects. They also used afforestation in China as a test case to quantify the differences in biophysical effects using the three approaches and verify their hypotheses. The manuscript is well-structured, and the results are clearly represented. I would recommend the publication of this manuscript after minor revisions.

Some minor comments:

Language needs to be further polished throughout the text. Some long sentences are difficult to understand.

L30, “and that it … explained”, Not clear

In Methods, need to clarify how gridded effects were aggregated into the country mean for comparison among the three approaches, because different LC / LST data may have different coverage. How tis the overlapped region representative for the whole country?

L275-277, afforestation from GFC is not consistent with the inventory data, so can the results based on GFC be considered as the real biophysical effects of afforestation in China? I think this key message is important for policy makers.

L391, that’s what I meant, the afforestation area is much smaller than the national inventory.

Fig. 4, better to show the latitudes on the left axis of (a)

Fig. 5, did you consider the spatial distribution of each bin? Whether the regions with higher Faff happen to be in the tropics with larger cooling effects?

Fig. 8, I guess the differences for changes in seasonal fluxes would be much larger between the partial and full coverage of each pixel, especially in the snowing regions in winter.

L742, should be Nature Communications