Thanks a lot for these positive and constructive comments.

Comment 1: Indeed, micro-meteorological phenomena can have some influence on UHI creation. Wind only intervenes directly in the large scale evapotranspiration assessment by Surface Energy Balance, and more precisely in the sensible heat flux estimated by scintillometer. To understand the variables that most impact the iterative process of 𝑄h calculation (Eq. 10, 11, 12), a sensitivity analysis through a Latin Hypercube Sampling (LHS) was conducted (see Castellanos, 2022). The Pearson’s Correlation coefficient used as a sensitivity index in this analysis showed that the wind speed 𝑈 was moderately correlated to 𝑄h (0.35). Moreover, for these particular cases, the monitoring campaigns occurred in relative no-wind conditions, for which we can assume very few differences between local data and Meteo France data. Note that, since these campaigns occurred, some anemometers have been implemented on the ENPC campus to have local wind measurments. These details will be added in the revised manuscript.

Commetn 2: As mentioned in the guidelines for authors, “all material required to understand the essential aspects of the paper such as experimental methods, data, and interpretation should preferably be included in the main text”. In consequence the description of the python scripts could be transferred in an Appendix as advised by HESSD. These descriptions contain some elements to explain the provided scripts to read and analyze the data, and how the equations presented in the manuscript are used. In return, the comparison of the obtained ET values can be extended (by presenting the spatial variability captured by the water content sensors for instance).

Thanks a lot for your constructive comments and suggestions (italic in the following). Here are our answers and proposals.

Reviewer's Remarks: While the Python scripts could be handy to some extent to get a quick impression of the provided data, they do not work in my environment (Python 3.11, Anaconda). In each script I get error messages. So, to the very least, it should be stated in which environment these scripts are expected to work free of errors. Please add the acronyms for each variable in the glossary in the annex.

Indeed, the Python scripts have been written and developed in Python 2.6. It will be indicated as the acronyms in the revised version of the manuscript.

Detailed reviewer's remarks:

p. 2 l. 87: EC is useful but only applicable to very large green roofs with sufficient fetch

We completely agree. As explained, EC can used for agricultural purposes on large crops. Some similar uses on green roofs require a very large asset. It will be indicated.

p. 9 l. 281: the term is usually called “zero-plane displacement height”. And you might add that the velocity at this height + roughness length is zero according to the logarithmic wind profile.

Ok, “zero-displacement heigh” will be replaced by “zero-plane displacement height”. Additional information about the consequences related to the assumption of the wind profile will also be added.

p. 9 l. 305: As it seems QE was estimated as a residual term. Therefore, QG should be calculated as close to the surface as possible, i.e. with z1 and z2.

Indeed. We had chosen z1 and z4 to capture the temperature gradient in the substrate layer profile. It is possible to compute Qg closest to the surface by using z1 and z2. The Python script will be modify to have the possibility to choose the thermocouples in order to compute Qg

p. 10 l. 311: which value was k set to? I suggest to implement a variable k in dependence on VWC according to Sailor and Hagos 2011 e.g. since k can vary by a factor of 2.

For now, we have proposed only two values for k, corresponding to dry condition (0.15 W/mK) and wet condition (0.85 W/mK) regarding Vera et al. (2017). This range is similar to that illustrated in Sailor and Hagos (2011). Nevertheless, it can be modified to better take into account this variability.

p. 11 l. 372: I presume it was interpolated to match the time step of the scintillometer measurements? Which method was used for that?

This wind measurement a clearly a weak point in our study (see answer to reviewer 1). In the absence of a local measure, we have used the wind data from the Orly Airport weather station, located 50 km from the BGW. As this dataset is characterized by a resolution of 3 hours, we used constant values in this interval to match with the scintillometer time step. Nevertheless, the Python script offers the possibility to implement a downscaling method.

Figure 10: I recommend to use lines (no smoothing) here as well for easier comparison.

Ok, it will be modified for a question of coherency.

p. 23 l. 800: “line” should be “column” I guess. 

Indeed! Sorry for this mistake.