Review of “Projected changes in forest fire season, number of fires and burnt area in Fennoscandia by 2100” by Kinnunen et al.

This is a land modelling study using JSBACH-SPITFIRE for projected changes to northern European forest fires, driven by bias-corrected climate from 3 ESMs from CMIP5 that have been downscaled to higher resolution (Euro CORDEX). The changes to several fire-related variables, such as the fire number, burned area, fire danger index, and length of fire season were all examined for 30 years near end-of-century, in the context of changing temperature, precipitation, winds, and fuel moisture. They found a large range in the results, depending heavily on the driving ESM, but generally finding an increase in both the number of fires and burned area towards the end of the century in the two RCP scenarios examined (RCP4.5 and RCP8.5). Understanding how wildland fires will change in a changing climate is a very important question, and I recommend this paper be published after the following minor revisions.

Minor comments:

Line 19: newer --> never

Line 68: rations --> ratios

Line 90: Can you add a line here that explains if anything is done to account for possibly overlapping fires? E.g. if 2 fires within a grid cell grow large enough, could they merge into one fire that may have less burned area than two distinct fires…?

Line 91-92: “taken into account” how?

Line 112: Since land cover changes were not accounted for over the ~150 year period simulated, can you comment on how that would affect your results? Is this region of Europe not expected to have large land cover/vegetation changes? Did it not become more agricultural than forested over time?

Lines 117-122: Is a 20% cloud-to-ground fraction applied everywhere in the model domain? If so, do you have a reference for that factor? At higher latitudes, the cloud-to-ground fraction is likely greater than 20%. It could also be a function of cloud-base height.  

Can you comment on how lightning frequency and distribution are expected to change in the RCP4.5 and RCP8.5 scenarios and how this might affect your results (e.g. lines 201-204) in this paper?

Line 120: ODT --> OTD

Line 121: Is the latitude-dependent correction factor applied to the LIS/OTD data a correction to the total lightning flashrate from that dataset? Or do you just mean a latitude-based cloud-to-ground fraction applied to this dataset? If the latter, I wouldn’t call it a “correction” since the total flashrate from this dataset is correct. If the former, I can’t find this correction in that Lasslop et al (2014) reference. Can you be more clear here about what was done?

Line 136: simulate --> simulated

Figure 1: the 6 locations are difficult to see (black font on dark background colour). Could you please put them on panel (c) instead? Or else, change the font colour to white to better see them.

Lines 208-209 (and repeated at lines 268-269): It says the simulations underestimate the number of fires, but then given the uncertainties, those numbers agree with each other. For example, 1355 is within the observed range of 1691 +/- 799, no?

Lines 218-219 (and repeated at lines 271-273): Similarly, the burnt area ranges overlap, so I don’t think it’s correct to say that the simulations overestimate, when they are within the range given (5.84 +/- 3.93 km^2).

Lines 241 & 266: newertheless --> nevertheless

Lines 259-260: this statement sounds too definitive for being based on one lightning parameterization being implemented in one model (EMAC). Given the high variability and uncertainties associated with both lightning and fire modelling, I suggest you change this sentence to: “The risk of lightning-ignited fires may vary from a 62% decrease to a 38% increase under RCP 6.0 in the polar regions from the 2010s to the 2090s, according to one study (Pérez-Invernón et al., 2023).

Table 1: this is a big table filled with numbers that is already well-represented by Figure 3. Therefore,  I suggest Table 1 be moved to the appendix or supplement.

Tables 2 & 4, and Figure A14: How come the smaller time frame (2011-2018) was not used for the models too? The different time ranges mean that you are not comparing the same thing from models to the observations. The additional years from the simulations may be responsible for the differences.

Figure A9: It would be helpful if the observations of these same variables were mapped up in a similar way for comparison to these multi-model avg results.

Figure A13: important to note the limitation of panel (c) that future lightning was kept the same as present lightning. Therefore, this panel only shows changes in the future due to the changing human ignition rate. So the title of panel (c) should be “human ignition rate change”, no?

Please see attached file. Thanks.