The present study is concerned with the multidecadal satellite-derived Portuguese Burn Severity Atlas. The authors offer valuable insights into the topic. The authors present the following observations for consideration.

(1) It is my contention that the role of small fires in the context of wildfire management is a significant yet understudied area. In contrast, the current study has a narrow focus on large fires (>100 ha).  

(2) It is anticipated that the fire atlas will have an identical resolution, resulting in comparable outcomes at either 30 or 500 meters. It is recommended that the ALI sensor be used for the 2012 data set.    

(3) It is my contention that a burn severity mapping system based solely on these indices with a fixed threshold will not yield optimal results. The resulting burn severity map is an inaccurate representation of the landscape due to the influence of environmental conditions, the diversity of objects, and the impact of climate.      

(4) It is anticipated that the reliability of the atlas dataset will be enhanced by the incorporation of a more robust validation dataset.

(5) It would be beneficial to examine the influence of seasonality and the climate dataset on the severity map.

(6) It would be advantageous to include the severity of different objects and trees during the fire period.

Minor comments:

(1) It is recommended that all figures be enhanced in terms of quality.

(2) It would be helpful to include a map indicating the frequency of burn for different areas.

(3) It would be beneficial to analyze the impact of the threshold value for different sensors on the same fire. 

The data article provides a new dataset to potentially evaluate fire severity in Portugal in the period 1984-2022.  This new dateset is a contribution to the analysis of fire severity in Portugal in the last decades. However, there are a series of limitations in the dataset that must be addressed prior to its publication:

The authors select fires larger than 500 ha for the period 1984-2000 to assess fire severity and fires larger than 100 ha for the remaining period, until 2022. The authors must justify the selection of these fire sizes and express in the methodoly the percentage that these fires represent for Portugal in terms of number of fires as compared to the total number of fires in the country, and in terms of burnt area, as compared to the total burnt area in the country for all the years of the analysis. Fires larger than 100 ha are probably below 1% of the total number of fires, although they may have a larger contribution in terms of burnt area, maybe about 30-40%? These basic statistics are needed to know how representative the fire severity dataset is for Portugal.

On the basis of the above and the availability of satellite imagery from Landsat sensors, the authors must justify the selection of the 500 ha and 100 ha thresholds. In principle, and given that the imagery is at 30 m spatial resolution, there is no obvious justification for the selection of these thresholds.  Please, explain also why MODIS was used for 2012, which was a critical year of wildfires in Portugal and the potential inter-calibration (comparability) of the fire severity series for the rest of the years with the data obtained for 2012.

The statement on the use of the DNBR-EVI regarding the potentila use of this index based on the analysis of 3 fires, only, should be excluded from the article as the representativity of this analysis is not acceptable by any scientific standards. The authors are encouraged to continue the testing of this index with a sufficiently large/representative dataset.