A Wildfire Detection Algorithm Based on the Dynamic Brightness Temperature Threshold

Round 1

Reviewer 1 Report

Considerated article is devoted to a very important problem - the identification of areas affected by forest fires based on remote sensing data. The authors well illustrate previous studies in this direction and propose their own algorithm based on the analysis of global databases. However, a number of questions remain:

1. There are no examples of the application of the proposed algorithm in the work, which reduces the understanding of the possibility of its use in practice.

2. The period for which the authors collect data remains unclear. Despite the fact that the elevation remains constant for weather stations, on the basis of which the authors build the model, global climate changes occur. A number of authors distinguish periods of climatic changes, indicating that the averaging of the values of meteorological parameters must be carried out based on the boundaries of the identified periods.

3. How were the latitudinal zonality, altitudinal zonality and geographic (insolation, circulation) position, distance from the coastline taken into account in the model? These factors are defining at the global level in the redistribution of meteorological parameters.

4. It seems that despite the large coverage of databases, the model is built on the basis of data from one particular fire. It would be good to present the results of model testing in various types of forest landscapes.

Author Response

We have carefully studied your revisions and have made careful changes. The manuscript of the paper has also been touched up by a professional organization (MDPI). Thank you for your dedication in this review process. Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The goal of the study is noble. However, there are too many faults and the article is poorly organized. The selection of the data and methodologies utilized in this paper should be supported by a justification. Furthermore, it is strongly recommended to arrange the paper to present novel findings or demonstrate the proposed strategy with excellent performance.

 

1.     Does “remotion” mean “remote sensing” in the abstract? The authors need to choose the correct term.

2.     L97: An explanation of the CV full name appears in L100, but the full name for the abbreviation should be written at the first time. Be sure to check that there are no more cases like this.

3.     L127-128: Sentences are not complete. Please check the English grammar thoroughly.

4.     Is the Harmonized World Soil Database v1.2 product used for the elevation information used well in the field? The authors need to add what is evidence for choosing this data.

5.     In Figure 5, (a) says false color image synthesized with channel 21/22/31, but (b) is also false-color pictures synthesized by using channel 21/22/31. Why do (a) and (b) look different?

6.     In 2.4.1 Otsu algorithm, the concept of the Otsu method was well explained. However, a detailed description should be added that how does the Otsu method do the binarization segmentation?

7.     Weather data has high spatiotemporal variability, and using weather data from the nearest station increases uncertainty in areas where stations are sparsely distributed. As a result, it is necessary to add a description of the time and distance ranges to the assigned station, as well as the resulting uncertainty.

8.     In Table 1, the range of the brightness temperature threshold 1 is 7.01 to 182.48. Is the range of BT reasonable? What is the unit?

9.     In 2.7.3 Data set partitioning, the leave-one method is used for validation. Is the single sample used here a pixel or a fire cluster? It is a reasonable validation method if the fire cluster is used. If pixel units are used, each fire cluster should be separated from the others for accurate validation, so that other pixels in the same fire cluster are not used for both train and validation at the same time.

10.  There is an error about Figure 10. Figure 10 (c) is not shown in the figure, and the explanations for (a) and (b) in the text seem to have changed. Make sure to check that part out.

Author Response

We have carefully studied your revisions and have made careful changes. The manuscript of the paper has also been touched up by a professional organization (MDPI). Thank you for your dedication in this review process.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I think that the article can be accepted as presented.

Author Response

Thank you for your great contribution in reviewing the manuscript.

Thank you for your support!

Reviewer 2 Report

It is recommended that the contents of the weather data distribution, as well as the time and distance range, be included in the manuscript as well as response. Since this part is closely related to the novelty of this paper, a more extensive examination can greatly improve the quality. It is also strongly advised to include a discussion on this.

Author Response

Please see the attachment.

Author Response File: Author Response.docx