This is a very interesting and relevant study and yields important contributions to our understanding of lightning (and in particular, downward lightning behaviour).

The paper is well-organized, with a clear structure that follows the conventional format of academic papers. The use of figures to illustrate spatial distributions and properties of GSPs enhances the clarity and understanding of the research findings. However, a more detailed discussion on the limitations of the study and potential areas for future research could further strengthen the paper

I have a couple of comments for discussion:

1. I am interested in the choice of 500 m as the distance threshold used. As noted, the EUCLID network consistently is found to have a location accuracy below 250 m and, while there is not a significant increase in the accuracy of the algorithms, would it not be worth using to achieve more accurate GSP classification?

Similarly, while there is not a large accuracy increase between algorithms, why was the algorithm that does not take into account ellipse information chosen over the other algorithms?

1. Flash density is a key parameter for assessing lightning risk in lightning protection design and GSP density has an important implication for lightning risks. Recent recommendations to multiply Ng by a factor of 2 to approximate Nsg and include this in the IEC62305 approaches make this study particularly relevant.

It would be good to see some discussion of this in the manuscript. In particular, comment on whether a factor of 2 is appropriate?

1. The final sentence of Section 3.2 “The relationship between the average GSPF and the absolute peak current is noticeable on a monthly scale, but it is not as strong as it is at the daily level.”

Is this correct? The relationship seems clear at a daily level…

The paper is well-organized, well-written, and is an important contribution to the lightning detection community, especially in improving the lightning risk evaluation and related standards. Still, some aspects should be further discussed to emphasize the results.

Specific comments

The paper stresses that flash density (N_G) is a crucial parameter to evaluate lightning risk (e.g. IEC 62305). However, there’s no further mention of the norm neither in the discussion nor the conclusions. Results suggest the risk estimation of being struck by lightning should move to ground striking points (GSP) instead of N_G. This point is relevant and should be discussed, as lightning risk has been introduced in the paper as one of the possible applications of the obtained results, but not further mentioned.

Line 78. The method includes the CG stroke grouping (flash algorithm) as a first step before calculating the GSP. The flash algorithm parameters used are the classical ones introduced decades ago in the NLDN (Cummins et al., 1998). Since the stroke clustering has a bearing on the further GSP algorithm, I wonder if the EUCLID community has validated the Cummins criteria, since other studies (e.g. San Segundo et al., 2020) suggested that can be adjusted. Since the paper mentions the use of video and E-field records to estimate the EUCLID network detection efficiency, I think they should be used also to calibrate the flash algorithm, previous to the GSP calculation.

Line 57 “the network's configuration has undergone changes in both the past and during the investigation period, i.e., 2013-2022. However, these changes are not substantial enough to significantly affect the results presented in the following sections” I think this statement needs a supporting reference.

Line 122 to 126 It is mentioned that one of the aspects that leads to an overestimation of single-stroke flashes is the “misclassification of IC pulses as CG strokes.” Can you develop this statement? What is the relative contribution of this factor? A supporting reference is needed here.

The paper mainly discusses the ground strike points (GSP) spatial distribution over Europe based on cloud-to-ground (CG) stroke data from EUCLID network. It has a very short introduction that does not bring a clear motivation for the study. The description of the data and methods is fine, and the results are presented by nice figures and plots. However, the discussion is a little confusing, mixing different aspects of lightning detection that prevents from a good understanding.

1) My main concern is the actual motivation of the paper. Although the analyzes of the spatial distribution and the temporal characteristics of GSP are important for lightning protection, the paper does not discuss why we need of those results. In the Introduction (lines 26-32), it was said: “Among the various components that influence the risk estimation, the standard puts forward the flash density, NG, representing the number of lightning flashes per square kilometer per year, as one of the key parameters. However, by definition, the location of a flash is determined by the position of the first cloud-to-ground (CG) stroke within the flash. On the other hand, numerous studies, supported by high-speed camera observations (Rakov et al., 1994; Valine et al., 2002; Saraiva et al., 2010; Poelman et al., 2021a), have provided evidence that, on average, multiple ground strike points (GSPs) exist within multiple-stroke flashes. Hence, GSP densities should be given the pivotal role in lightning studies, particularly in the context of assessing lightning-related risks”. The authors shall discuss more comprehensively how this study can effectively improve the lightning protection standards “in the context of assessing lightning-related risks”. Remind that NHESS main scope is natural hazards, and the main topic of the paper must be connected to this subject.

2) In lines 147-180 there is a long discussion regarding the lightning location limitations on measuring the CG strokes. I’m not sure if all those details are necessary. Chi2 and SMA are quality solution parameters that are used to select “good” solutions, which were used in the analyzes. In my opinion, these parameters might only be relevant for the study if the authors describe more comprehensively how EUCLID detects and geolocates lightning. Even in this case, I was wondering if this discussion can be suppressed.

3) The paper also discusses the land-ocean peak-current contrast, which has already been described by several other publications cited by the authors: Cooray et al., 2014, Nag and Cummins, 2017 Poelman et al., 2016. I do not find any contribution of the GSP analysis to this topic. The same for the diurnal and seasonal variations (Figure 6). Is the intention of the authors only to present the GSP temporal behaviors? If yes, then you are please encouraged to discuss in more details how these characteristics impact on lightning protection and even on natural hazards.

4) Finally, the correlations of the GSP with multiplicity and peak current are discussed based on Figures 7 and 8. Again I do not find any relevance of these results in terms of GSP analysis, lightning protection, or natural hazards. I’d like to see a more comprehensive discussion on how these parameters affects the GSP results which will consequently impact on the lightning protection standards.