By estimating the effect the zone of clear-to-cloud transition has on longwave radiation, the study addresses an important yet little known issue. By showing that these effects are quite substantial, the results provide information that could be of great interest to the community. The methodology is suitable for the task and the presentation quality is high. Even so, I believe that the manuscript needs some significant improvements. My recommendations for substantial improvements and for some minor refinements are listed below.

Main issues:

1.

It is important to clarify the sentence in Lines 90-91, which says “Indeed, for these pixels neither aerosol nor cloud optical property retrievals exist, yet they are classified as containing a cloud (Lost A), a non-cloud obstruction (Lost B), or were not processed at all in the cloud masking (Lost C).” Specifically, it should be clarified whether the study considers 1 km-size MODIS pixels as “cloudy” or “Lost” (most likely “Lost A”) if the MYD06 cloud product does not include a positive retrieved value in the Scientific Data Set (SDS) “Cloud_Optical_Thickness”, but includes a positive retrieved value in the SDS named “Cloud_Optical_Thickness_PCL”. This occurs for partly cloudy 1 km-size pixels, in which clouds were detected for some, but not all 250 m-size subpixels. 

Clarifying this would be important because if such pixels were considered “Lost”, CERES footprints containing many small clouds could be included in the transition zone statistics even if their total cloud fraction was well above 10% and their longwave effects came from cloud elements for which the MODIS cloud product did provide cloud property estimates. 

2.

Line 250 explains that the low-level transition zone effect of 0.8 W/m2 was calculated using the first four temperature difference (dT) bins in Fig. 4. However, it is not clear why four bins were used, rather than three, five, or more than five. This is a significant issue because Figure 4 suggests that the number of dT bins included into the low-level category can affect the results. It would help to explain why using the first four dT bins is a good choice. For example, could it be linked to a certain altitude range? It would also help to mention how the definition or the extent of the low-level category compares to the definition or extent in Eytan et al. (2020), which provided the radiative effect estimate of 0.75 W/m2 that was close to the 0.8 W/m2 in this paper. 

3.

The transition zone statistics include CERES footprints where up to 10% of MODIS pixels have neither aerosol nor cloud data. This criterion is very reasonable, but it allows including footprints where the cloud fraction can reach 10% (or much higher, depending on the treatment of partly cloudy MODIS pixels, as discussed in Point #1 above). Therefore, it could be interesting to discuss whether the transition zone radiative effect shows any statistical relationship to cloud fraction within the CERES footprint. This could be done either for all dT bins combined or for selected dT bins only.

Suggestions for minor refinements:

Line 18: I suggest adding “the” between “onboard” and “Aqua”.

Line 28: I suggest replacing “regardless of” by “without considering”.

Line 39: I suggest adding “the” between “that” and “transition”.

Line 81: I suggest adding “the” between “for” and “identification”.

Lines 83-84: It would help to clarify whether the 3 km or 10 km resolution MYD04 aerosol product was used.

Lines 96-97: I suggest rewording “number of ocean MODIS pixels more than or equal to 75% of the expected ≈400 pixels falling within CERES field of view (FOV)” to something like “the number of MODIS ocean pixels equals or exceeds 75% of the ≈400 pixels expected to fall within the CERES field of view (FOV)”. 

This paper presents a novel method to estimate the impact of the transition zone on outgoing longwave radiation, an under explored area, using CERES observations with help from MODIS measurements. The authors, as well as other recent studies, showed that the impact of the transition zone on longwave radiation is quite significant compared to clear atmospheric conditions. The paper is well written and the method is sound. The paper could provide useful information to the community. However, I do have some questions and suggestions. I hope that could help the authors improve the manuscript before it could be published.

Major comments:

1. The “conventional” transition zone study focused on low level clouds such as Eytan et al. (2020) referenced in line 55. However, this paper does not mention if they looked at the transition zone near low level clouds. This is important because high thin cirrus could have similar LW effects but the cloud processes of thin cirrus clouds are completely different from low level clouds. This needs to be clarified.
2. The classification of undefined pixels (Lost A, Lost B, Lost C) is useful. However, the paper lacks the description of how to match MODIS pixels to CERES footprints. Since this paper mainly presents a method to estimate the longwave effects of the transition zone, matching MODIS pixels to CERES footprints is a critical step, and it should be described.
3. CERES products provide both radiances and fluxes. The authors used LW radiance without no mentioning the reason not using the LW in the product. Is the sub-footprint cloud variability that makes the radiance-to-flux conversion difficult? Some discussions are necessary.
4. The definition of temperature dT is not clear. It seems dT is the difference between surface air temperature and cloud stop temperature (lines 227-235). It is hard for me to comprehend very small values of dT. What is the physical meaning when dT is very small? Is it because clouds are very low? Is it because of sub-pixel clouds in MODIS observations that makes cloud top temperature appears low? Some discussions are necessary.

Minor:

Line 34: “a phase called transition zone”. “phase” has been used several times for the transition zone (e.g., line 40, line 217: “a phase of particles between the cloudy and so-called cloud-free skies..”, line 289: “an important phase of particle suspensions..”, line 293: “intermediate phase of particle suspension..”.) To me the transition zone is not another phase of matters (e.g., solid, liquid, vapor). Even clouds contain liquid drops, ice crystals, and water vapor. I would use “a special region” to distinguish from clouds and cloud-free areas.

Line 81: “homogenous” -> homogeneous

Line 85: “These products were obtained for all MODIS-Aqua granules that contain data in the region 0° E – 15° E and 10° S –30° S during August 2010, which their data spreads over the area between 21° W – 21° E and 10° N –50° S.” Not understand.

Line 85: “MODIS-Aqua”. I would use Aqua MODIS (e.g, Minnis 2011)

Line 275: “3783 cases have been found…” I would change it to 3783 CERES footprints.

Not sure if the boxplot inset of Figure 3 is necessary since all information is already available from the cumulative distribution and median and mean values indicated. If it do not provide additional information, it would be better to remove it.

In the bar chart in Figure 3, should we expect to the sum of LOST A, B, and C to be one? I might have missed something. It would be nice to add some description in the caption so that the potential reader could see it immediately.