The study investigates the land-atmosphere interactions during flash droughts using daily satellite products from 2000 to 2020 for the purpose of improving the S2S predictability of flash drought. The flash drought events are identified using surface soil moisture (ESA CCI Soil moisture combined active/passive microwave product) and land-atmosphere coupling processes for composites of flash drought events are analysed using standardised anomaly of net radiation at the surface from CERES, latent heat flux from GLEAM and sensible heat flux as the difference between land surface temperature (ESA CCI) and 2m air temperature (ERA5). The study demonstrates that flash droughts with stronger land-atmosphere coupling persistent surface energy budget perturbations months before and after onset. Further, the study shows that increased sensible heat flux during flash droughts feeds back to raise near-surface air temperatures, especially in semi-arid African regions.

The manuscript is generally well-written with comprehensive details on assumptions and limitations of the data. The study provides detailed investigation of land and near-surface atmospheric variables during the flash drought; however, the current work lacks substantial conclusions with respect to knowledge gaps in S2S predictability. I think the paper could be strengthened with additional investigations on evolution of variables modulating land-atmosphere interaction for other land cover classes (shown in Figure 2) in addition to rainfed croplands.

Therefore, I recommend major revisions before publication to enhance the robustness and significance of the findings.

Specific Comments:

Figure 2: Provide clarification on the timing of the drought event in the figure caption and discussion. The figure S2 mention the composites during peak growing season even though it is shown as accompanying figure of Figure 2.

Figure S3: The wind speed at 10m shows substantial difference for different quartile, which suggests wind speed is important for the sensible heat anomaly. The authors should add relevant discussion for the validity of sensible heat flux calculation in section 3.2.

Line 161-162: Can authors add more clarification on how DT is calculated at 0.01° spatial resolution? What is the spatial resolution of ERA5 2m used in the study?

Line 169-171: Provide clarity on ERA5 2m wind speed. How is it calculated?

Line 206-209: The negative latent heat flux anomaly for shrubland before the onset of flash drought has been explained as transiting to water limitation regime earlier than other land covers. However, the evolution of surface soil moisture is similar for all land cover classes. There should be other factor that may explain the early negative latent heat flux anomaly. I suggest investigating the evolution of variables for shrublands as done for rainfed cropland in Figure 3.

Line 303-305: Please rephrase for clarity or provide additional details on the regions.

Line 310-311: The three semi-arid regions have different land cover classes. I think land cover should be brought into the discussion as there is difference in the evolution of land-atmospheric coupling process for different land cover classes (Figure 2).

Line 351-355: These sentences suggests that study lacks substantial conclusion as per the objective set in the introduction. I suggest discussing the role of different land cover classes for non-robust relationship between precursor variable and 2m anomalies. Further, the role of VOD as precursor need to be assessed for other key regions to have robust conclusion.

Line 403: If VOD is closely linked to root zone soil moisture (RZSM) and serves as a precursor for 2m temperature anomalies during flash droughts, does identifying flash droughts based solely on surface soil moisture provide a reliable approach for flash drought monitoring? Further, I suggest using ESA-CCI-COM based root zone soil moisture dataset in addition to GLEAM RZSM and discuss its application for land atmosphere interaction during flash drought.

The authors present a study that uses remote sensing data (supplemented with reanalysis data) to characterize dynamics of the surface energy balance during flash drought events. They also look at other remotely sensed data, including vegetation optical depth, to investigate how they change with respect to drought strength as characterized by ΔT anomaly. Overall, there is a high need to assess drought globally and to improve predictability and the topic is well within the scope of HESS. 

Major comments

1. Overall, the manuscript is well written but could benefit from some reorganization: The results section contains considerable portions of methodology (e.g. Sections 3.2, 3.3). It would be good to explore whether this can be moved into the methodology section to improve readability of the manuscript.

2. Science questions: This study presents as a proof of concept for quantifying surface energy balance changes during drought. There are limited additional results (e.g. the relationship between VOD and drought). For me, this is OK, but the manuscript would benefit from additional justification for the choices of region and land-cover in results and additional discussion about approaches for applying this method in S2S forecasting or drought monitoring. 

3. Land-atmosphere interactions: I am not sure whether I agree with the author that this manuscript is primarily about land-atmosphere interactions as is indicated by the title. The manuscript mainly addresses surface energy balance, which is important enough. I suggest that the title is changed to something less broad. The main LA interaction discussed here is the relationship between Tair and Tsoil, which is part of the method (e.g. delta T), but since delta T is taken from reanalysis and delta T anomalies are discussed, it is not really explored in depth. I am also questioning the use of the word feedback (see specific comment)

 Specific comments: 

L215: "Although net radiation is decreasing, the land has entered a water-limited regime, so this radiation drives less evaporation" > I am not sure about the conjunction although here. Is that not something that would be expected. 

Fig 3: Provide explanation of variable abbreviations in figure caption

L176: "We investigate feedbacks from flash droughts to atmospheric temperatures using composites of ERA5 daily maximum

2m air temperature (T2m)" > I am not sure what is referred to here as feedback since the study looks at Ts -> T2m, which is not a feedback but maybe a forcing?