The paper regards the evaluation of heat stress indicators for various cities in in Europe. 

Quantification of heat stress under global warming is performed employing state-of-the-art GCMs and RCMs, and the authors correlates the various indicators to latitude, longitude, and proximity to the sea.

The statistical analysis is well performed, clearly presenting results and with robustness. From my point of view, the current work deserves a publication, after these minor points, referring in particular to 1) strengthen some sentence with reference to literature 2) Explaining more in detail the method employed for the statistical analysis:

LINE 32: ...such as in Canada in summer 2021 or in China and Europe in summer 2022. I would add a reference to this sentence, and/or a quantification through te registered anomaly

LINES 47-51. Here you present HWMId. However, there are more heat stress indicator, such as MRT or UTCI. I would extend the introduction about heat stress indicators, stressing why you employ HWMId.

LINES 66-67: Analyses of climate and climate change in cities face the challenge of delivering results on spatial resolutions that are high enough to be relevant for cities while robustly estimating the risk of extreme events. I would add a reference to this sentence

LINES 67: "Urban models" is too general in your topic. I would call them "Urban canopy parameterizations"

LINE 83: ...simulations at a resolution of  0.11° (EUR-11, ~12.5 km), which is fine enough to analyze climate conditions in major European cities at the city-level.  In my opinion this resolution is too coarse. So to convince me, you should justify your sentence.

LINES 144-145: we use the grid cell closest to the city  centre for our analysis. 

How do you define the city center? You should say it since you use it as  the center even for the subsequent averaging over surrounding cells.

METHODOLOGY in general: I would Add a subsection-paragraph saying if or which model employ a urban canopy parameterization. I know it is so relevant for your work, but since some of the models cited later could use a UCM, I would say which member use it, and which model with appropriate reference to literature.

LINE 177: You should say what GSAT is (it appears the first time here)

LINE 220: How is T calculated in your models? Is it a diagnostic 2m temperature,  the temperature of the first layer of the model or what? You should clarify it.

LINE 229: However, we do not apply bias adjustment here due to the lack of reliable reference data, as urban areas are not specifically represented in the reference datasets ERA5-Land and E-OBS. 

I don't get the point here. If E-OBS is observation, I guess cities are present, at least influencing the observation. Could you clarify this point?

LINE 234: ...is calculated for each grid cell in a box of 5x5 grid  cells around the centre of each city in the reference period 1981-2010. Ok, I think I don't agree with this methodology. In fact, cities taken into account in this work varies substantially in size, so I don't get why, for example, a "small" city such as Lisbon should cover more than 50x50 km. I would strengthen this methodology justifying why you use this method.

LINE 304: Change "depending" with depends

LINE 311: Here you present the validation comparing the model with observation.. Did you find any trend in the systematic error, for example depending on latitude/longitude/altitude/distance from sea? It can be really helpful tu interpret the results.

LINE 325: How do you choose wether a city is close to the sea or not? You should mention it somewhere in the paper.

LINES 338-345: HEre you say that there is a strong correlation between city and the metric employed, reporting some example. Can you justify someway those behaviours?

LINE 450: ... box of 3x3 grid cells around the center. Here again the issue of 1) city center and 2) city dimension as noted in the previous comments

LINE 522:...showing an increase in heatwave risk in southern Europe along with substantial increases in coastal regions in northern Europe.

Could you explain why there is this substantial increase in particular in coastal regions in Northen Europe?

LINE 579:...UHI is projected to only intensify gradually due to global warming...

The paper presents a thorough analysis of ambient heat projections in major European cities using the EURO-CORDEX ensemble, comparing them with data from E-OBS, ERA5-Land, and weather stations. The study evaluates temperature biases, uncertainties, and factors influencing spatial patterns. It highlights variations in biases across cities and emphasizes the role of downscaling by regional climate models in shaping temperature estimates. The paper introduces a novel examination of nighttime ambient heat and compares projections with CMIP5 and CMIP6 ensembles. While the paper offers valuable insights, providing additional methodological details and discussing the limitations of the EURO-CORDEX data in the context of cities and urban areas could enhance its clarity and impact.



Urban Processes:

One noteworthy concern I have regarding the paper is the potential limitation arising from the EURO-CORDEX dataset's lack of representation of urban processes. The absence of urban-specific factors like the urban heat island (UHI) effect in the EURO-CORDEX models might lead to an incomplete understanding of how local urban conditions could influence temperature distributions. Addressing this limitation explicitly and discussing its potential implications for the reliability of the findings could benefit the readership.



Clarity of Methodology:

While the section on "Identifying factors influencing spatial patterns" is intriguing, the exact statistical methods used to establish the relationships between climate and location factors and heat metrics and the limitations of these methods should be explicitly stated.



Comparative Analysis:

The comparison between EURO-CORDEX, CMIP5, and CMIP6 ensembles in Section 3.5 is a valuable addition. However, the paper could provide more insights into the potential reasons behind the differences in projections. Elaborating on the distinct characteristics of the GCMs and RCMs, such as spatial resolution and physical parameterizations, could enhance the understanding of the results.