Dear authors,

This work provides a hierarchical classification system of surface and canopy fuels adapted to continental conditions, covering 85 fuel types, and grouped into six major fuel categories. The final European fuel map contains 20 fuel types, obtaining a high estimated overall accuracy (88%-81%) for all mapped fuel types. 

I reaffirm the words of the authors and it is the first step towards a mapping that takes into account the risk of risk throughout Europe, which can help in decision making from a prevention point of view that in years like 2022 in the Mediterranean regions a pre-season fire risk assessment is necessary. 

It only remains for me to congratulate the authors since the article brings a novelty in the state of the art, it has great quality both in the content and in the presentation of the same. 

To mention it, I would only detail the processing time required and the work process followed. 

In this paper the authors propose a new fuel classification system for Europe and release a fuel map at 1 km spatial resolution. I accepted to review this paper because of the title and the abstract promise to fill one of the main gap in forest fire research. However, I read this paper several times, but it is really difficult to understand how this work fills the gap of the lack of appropriate fuel mapping in Europe.

The paper is very well written and full of suggestions and sharable concepts. The authors elaborate huge information and datasets but the results are not completely satisfactory. Unfortunately, at this stage the main innovation seems to be related with the integration of bioclimatic models to estimate the fuel depth in shrubland and grassland. However, because of the high variability of this kind of vegetation cover and the disturbances it is really difficult to validate the results of these models.

Starting from the classification, the considered forest classes are very similar to existing vegetation cover maps already available. Using a threshold of 15% of trees does not allow to discriminate fire behaviour if there is no information on which kind of fuel is present in the 85% of the area. In addition, it is evident that some species are more flammable than others and the considered classes does not allow to discriminate among these species. 

As a general consideration, I suggest to remove the paragraph 4.2 and define the association of fuel models parameters in a second stage of this work when most of the presented issues will be solved using other technologies such as Lidar, AI, etc…

The paper can be accepted after major revision and a better description of the objective of this work starting from the title, the abstract and the conclusions. This could help also to make the paper shorter, improving its readability. 

Please, find specific comments on the attached pdf.

Dear authors,

I was carefully reading the manuscript several times and I found the manuscript well structured and written, with very interesting findings. Indeed, you used very updated and deep literature on the topic and clearly, you added new inputs adapting these fuel maps approach in a European context, being the main gap. However, there are some points that I would like to stress, which could improve the quality of the manuscript. See the comments below:

General comments

Introduction

The introduction demonstrated a deep state of the art regarding fuel classification systems, globally and continentally. The review about fuel classification is a great input for other researchers, where the authors put a lot of effort, being a very valuable contribution for future investigations. Well done!

Even though the authors demonstrate a great knowledge and handling of the literature on the subject. The introduction is extremely long and repetitive. From what I understood, subchapters 2.1 and 2.2 are also part of the introduction, or? I would suggest sending these two subchapters as supplementary material, since they are part of a technical report rather than part of a scientific article.

Material and methods

The methodology in general is well addressed. However, I have the following concerns:

• The European fuel classification map that you created to which year belong 2016, 2018, 2019 or 2020? Because it is important to define to which year the map belong independently of the year of publication (article). I would suggest use the oldest layer input to define the year of the European fuel map classification.
• Why the spatial resolution of the European fuel map classification is 1 km. Considering that most of the input layers used are between 100 and 300 m, I’m not sure why you reprojected the spatial resolution to 1 km. Maybe a reason could be the climate data that you were using from WorldClim2, or? In my opinion, if the main goal is to develop strategies on how handle forest fires at Pan European level the resolution of 1 km is quite wide for this goal.
• Regarding the validation, I’m not sure about the validation with another remote sensing input and not through ground truth data for the six main fuel types. However, if you are able to explain well the limitations of this work regarding this gap of information, it can be a valid approach for the two-validation process proposed.

Results

The results are well addressed, but as I mentioned in the methodology section, it is difficult to validate a remote sensing product as a European fuel map classification with other remote sensing inputs. So, my main concern is the lack of field data in the validation. I’m aware of the limitations of the work at this scale. Therefore, these limitations should be properly addressed in the discussion.

Discussion

The discussion needs to be addressed more realistic with the results obtained. Again, as I mentioned previously, I have my doubts if this fuel map describes also canopy fuels. Basically, there are two main concerns: i) fuel depth is just one variable (basically, the height of the vetegation), but other important attributes as the height at the crown interception and the crown bulk density are missing and ii) I have some doubts about the models that you were using in the fuel depth estimation. It is important to clarify this, if the main novelty in this work is the comparison from the previous one at global and European levels, is that these maps only dealt with surface fuels. In my opinion, the canopy fuels are not solved at all through the manuscript. Therefore, I would suggest addressing the discussion about the limitations of this work, that the canopy fuel still is a pending task, but the fuel depth is a contribution on it. Also, you can address the discussion on some recommendations on how canopy fuels can be produced in the future through remote sensing.

Conclusions

The conclusions should be rewrite according to the previous suggestion and clearly define the novelty of the work compering to the previous one according to the limitations implied by a map at the European level.

In short, the manuscript is a valuable contribution, but need to be improved before to be published according to the comments and suggestions provided.

Specific comments

L20 In the abstract, you mentioned that the main fuel categories were grouped into six land uses, but I’m not sure why you included urban category, are you expecting forest fires into urban parks, or?

L110 Into the objectives, the first one is “generating a European advanced fuel classification system to facilitate the integration of continental wildfire risk assessment, including both surface and canopy fuels” I would like to highlight the last point. In my opinion, the canopy fuels are not really addressed within the methodology. It is true that you added the depth fuel (or height of the shrubland, grassland and so on). I have the feeling that canopy fuels is too ambitious according to the methodology and results. The depth fuel can partially be an input in the canopy fuels. However, other important metrics to understand better the canopy fuels are missing, such as crown base height, crown bulk density, among others. Therefore, I would suggest being more focused on surface fuels, which is most of the work done through the manuscript.

L145 You defined the vegetation height as fuel depth, but I’m not sure if this term is correct, maybe “fuel height”?

L160 Garrigue and maquis are just terms used in different countries, but with the same meaning “shrubland”. I would suggest deleting these terms because is repetitive or explain that is a specific term according to a specific country.

L195 Since the article probably will be published in 2023, it could be important also highlight the recent forest fire between the Czech Republic and Germany in 2022 and how the forest fires are going more up in latitude.

L230 It is not clear at all to me how the fuel height was determined and how accurate it was considering that the units are in centimetres, which is hard to handle from earth observation data (and models). Are some ground truth data available to validate the fuel depth?

L245 Considering that the map layers are from different years (e.g., Copernicus GLC map 2019, CCI LC map 2020, etc.) I would suggest adding the years when the layers were created in the workflow of the figure 3.

L290 Where the biomass data is coming from to feed the depth (m) response variable? I have my doubts about the models, especially in its validation (see comment in L230). The same concern is in equation 3.

L420 The histogram showed shrubs up to 6 m. I believe that there is a misunderstanding in the definition between what is a tree and what is a shrub. Probably, in the distribution that the histogram is showing, there are some trees included.

L455 The table 4 is quite large and the same information is repeated in the text. Maybe, it would be better to replace the table by pie graphs or send as supplementary material.

L525 I would suggest the same that in L455, the table 8 is quite large and could be more suitable transform in a pie graph only the % values.

L545 Figure 7 is connected to figure 4? Because the results are completely different but is the same attributed measured. Or?