Forest Fire Patterns and Lightning-Caused Forest Fire Detection in Heilongjiang Province of China Using Satellite Data
Round 1
Reviewer 1 Report
I have attached a pdf file with my comments for the authors.
SUMMARY
The goal of the manuscript is to combine VIIRS data and forest fire records from a wildfire database to describe some temporal and spatial patterns of wildfires in the in Heilongjing province of China, especially of lightning-caused fires. The study also aims at exploring how many of these fires were detected by the VIIRS sensor, and the reasons why some of the lightning-caused fires were not detected by VIIRS. The manuscript is not long and is relative clear, and the topic is interesting and well suited for the journal. However, there are some important problems, and other small issues, that limit the quality of the manuscript. I am describing below some of the main problems and issues that I found.
MAJOR COMMENTS
1. The word “regular”
What do you exactly mean by “regular” in the manuscript? I do not understand it. Most of the times the word regular is not needed, and other times I have the impression that you mean “patterns”. You should go throughout the whole manuscript and revise each time you use that word. Especially in the title! I can’t understand the current title because of the “regular” word at the beginning.
2. Introduction
The introduction was not very useful and most of it had relatively low interest for the rest of the paper. Thus, I think in general terms the intro may fail to introduce your study properly.
The second (L 55-67) and third (L 68-86) paragraphs of the intro are mainly a list of descriptions of papers on spatial point analyses and fire remote sensing, respectively, mainly in China. However, in my opinion these “lists” are of little value for readers of the paper. Both paragraphs could be easily shorten and combined to simple point out the utility of both types of analyses without actually describing what each paper cited did.
On the contrary, the four paragraph was useful to understand the context of the study. However, in general, in the introduction I miss some information about the human and lightning fire regime in the study area, which could help to establish better the context of your study. There are numerous papers published recently that studied wildfires in the region and lightning fires in the Daxing’an Mountains by different Chinese authors.
Finally, you didn’t write about the specific objective(s) of your manuscript. What did you want to accomplish? And why?
3. Spatial and temporal distribution of forest fires detected by VIIRS
I don’t understand why you choose VIIRS data for this section. Should forest fire record data be better to explain the temporal and spatial patters of wildfires than VIIRS data?
Furthermore, I think the structure of the whole section could be improved. In general, most of the information appears without a clear logic to me, which makes difficult to follow your ideas and find the relevant points. In my opinion, you could organize better the information and improve each paragraph by adding initials sentences that tell us what we are going to find in each paragraph, and final sentences that summarize or give us the main points of each paragraph. You may use another papers for inspiration to decide what spatial and temporal information you want to show and how to organize it, and how to create more valuable figures and tables. For instance, I think you are trying to show interannaul variability in number of fires and burned area, seasonal variation, and spatial differences between districts, but without distinguishing between human and lightning fires.
4. Capacity of satellite for forest fire detection
The main analyses of your work is the processing of the VIIRS data using forest fire points as reference data. However, you don’t explain anywhere how the analyses were done. You don’t say anything about the algorithms or the spatial-temporal rules you applied to go from 1456 VIIRS pixels (Table 1) to 102 fires detected (L 239). In other words, how did you assign a VIIRS pixel to a specific fire records using the temporal and spatial data of both VIIRS pixels and fire points? Did you use a method already published? Or did you come up with your own method? In my opinion, this is ESSENTIAL (i.e. a detailed explanation of this process) and should have included in the section 2.4 Data processing. At least one or two examples of how the VIIRS pixel-fire point matches were done should appear in one new figure.
5. Capacity of satellite for lightning-caused forest fire detection
I think it’s very interesting to assign a cause for a fire not being detected by VIIRS (Table 2), and your 3 classes are good to me (cloud cover, no satellite transit, other). However, you don’t explain how you did this. For example, for the cloud cover, did you MODIS imagery? Or another satellite? What times did you use for the imagery? For the “no satellite transit”, did you use the time of discovery and extinguish from the fire records to know if the VIIRS passes over the fire area within that period of time? One thing you should take into consideration is that lightning fires may be ignited well before they were recorded on the ground (as you point it out previously with the detection times offered by VIIRS). Regarding small fires, did you use a size threshold applied to the burned area recorded in the fire records to decide that certain fire size cannot be detected by VIIRS? I think you should explain well your methodology regarding this aspect on “causes of non detection of lightning fires by VIIRS”. Otherwise we cannot be sure if you approach is good enough. Of course these methodological aspects should have been included in the section 2 of the manuscript.
6. Lightning-caused forest fire cases
This section is the closest one to a discussion in the manuscript. However, I think the authors did not have a specific purpose with this section. The first part doesn’t add anything relevant or discuss the literature on lightning-caused fires in the region. In addition, the two examples described are unrelated to the previous parts of the manuscript and add nothing new. In fact, they are even a bit confusing because the authors mention lightning data (L 329) that was never described earlier in the manuscript. In my opinion, the whole section could be removed or should be rewritten entirely.
I think that one if the biggest weak points of the manuscript is the lack of a proper discussion. The authors don’t discuss their results and put them into perspective regarding what the current scientific literature says on lightning fires in particular and wildfires in general in the region. I think that probably your references from the literature are too few, and this is reflected in the lack of a suitable discussion. I think you should read more on the topic to build a proper discussion for your work.
MINOR COMMENTS
1. Introduction
L 37-42. The first three sentences have little connection with anything else and could be deleted.
2.1 Study Region
Figure 1. I don’t think you need the initial parenthesis in “(In map (a), the …”.
2.2 VIIRS data and forest fire records
L 131. I think it is better to write “from” instead of “in”.
L 135. Why do you cite “Figure 1b” in here?
L 138. What is BST?
L 141-143. It is obvious that a 375-m resolution is more sensitive to small fires than a 1-km resolution? Do you need a sentence and a reference for that? Consider removing the sentence.
2.3 Land cover dataset
L 158. Generally, we don’t write complete names and affiliations in the text of a paper. I suppose the correct form is something like this: “It was created by Yang and Huang [27]” or something similar, please check the submission guidelines from the journal.
L 159-164. I think you don’t need these sentences about the methodology, since it should be described in the original source [27]. However, you could add some more valuable information for the readers about this dataset, such as how many land use classes there are and what accuracy levels these datasets have.
2.4 Data processing
L 176-177. I have a question. You used the Landsat land cover maps to select only VIIRS pixels in forests. How many VIIRS pixels did you exclude from the analyses? Could this be one of the reasons why you only detect a relative small percentage of the recorded fires? The land cover maps are not perfect and you may be excluding too many pixels? I know clouds, passing time and fire size were also reasons for a relatively small detection rate, as you explained this in section 3.3. In addition, how did you decide a VIIRS pixel was in forest land cover if both maps (VIIRS and Landsat) have different spatial resolution? For example, do all the Landsat pixels within a VIIRS pixel have to be forest?
3.1. Spatial and temporal distribution of forest fires decteced by VIIRS
L 183. “Spatial” instead of “spatial”, and “detected” instead of “decteced”.
Table 1. The column “Numbers” means fire records? What about “Affected area”? I guess they are the burned areas that appears in the forest fire records. What is the unit of burned area? Hectares?
L 185-186. If you want to show trends, a graph is better (e.g. Figure 5). Furthermore, the time series is so small that I’m not very sure if these “trends” are significant or not.
L 191-193. The table doesn’t do a good job comparing things because you used absolute terms. Maybe if you add percentages (i.e. 3 more columns) we can see better the comparison between the tree aspects (i.e. # pixels, # fires, burned area).
Figure 2. I don’t understand why you use number of pixels to create this figure? Fire records should provide a better information than VIIRS on the total number of fires (or burned area) per jurisdiction, or not? Furthermore, the number of VIIRS hotspot pixels in forest land cover should not be indicative of the number of fires, since each fire may have a different number of pixels associated to that fire.
Figure 3. I guess the Y axis refers to number of VIIRS pixels in the forest land cover. If you want to show that the fire activity varies per month, should it be better to simply represent the number of pixels per month for the whole study period? This way we could see better the months with more fire activity (e.g. April) than a time series.
L 205. Why do you mean by “M-shape”?
L 210-212. This aspect is better shown in Table 1 since we can see annual data. However, interannual variability in 2-3-4 years does not mean any “improvement”, it may simply indicate strong interannual variability in some of the drivers of fire activity in the region (e.g. fire weather).
3.2 Capacity of satellite for forest fire detection
L 234-237. Regarding this sentence “As shown in Figure 5, out of the 384 forest fires recorded on the ground within the 234 jurisdiction of Heilongjiang Province from 2013 to 2020, 298 forest fires were caused by 235 lightning strikes, with the highest and lowest number of forest fires occurring in 2017 (97) 236 and 2013 (14), respectively.” The first part is not a repetition of what was written in the previous paragraph? The second part appears in Table 1, why don’t you cite Table 1?
L 237-238. Regarding this sentence: “The maximum and minimum number of lightning-caused 237 forest fires occurred in 2015 (73) and 2013 (7), respectively”. Why don’t you include the number of lightning-cuased fires per year somewhere? It could be easily added to Table 1 and we could see the whole time series.
L 238-240. I don’t understand the sentence the way it is written. I guess you mean that < 30% of the fire records were detected by VIIRS (i.e. 102/384).
L 242-243. This sentence is a bit confusing. I think you should add the following: “… for 82.4% of the total number of forest fires detected by VIIRS sensors.”
L 246-248. Unless I misunderstood, I think you made a mistake in this sentence. Figure 5 shows that the number of forest fires detected first by VIIRS is lower every year than the number of forest fires detected first on the ground (i.e. the opposite of what you wrote).
Figure 6. I must recognize that I’m very lost with this figure. First of all, I don’t understand why you using initials (K, J, I) instead of naming each line or axis properly (e.g. proportion of lightning fires, proportion of fires detected by VIIRS). It is very hard to follow the figure with those initials. Second, what is the point of comparing both lines in the same graph? These are two unrelated aspects, aren’t they? Consequently, I don’t see what you want to prove with the linear correlation shown in Figure 6b. For me, it would have much more sense to compare, for example, the proportion of fires detected by VIRRS human vs lightning, for instance to show that it is harder (or not) to detect with VIIRS one of these two types.
3.3 Capacity of satellite for lightning-caused forest fire detection
L 289. Maybe you can add where the 85.2% comes from, for instance saying that 43 of the 292 lightning fires from the Heilongjiang Province were detected by VIIRS. This helps keep track of the number of lightning fires since the manuscript is full of different numbers and percentages and sometimes readers may get lost.
L 295-296. Regarding this sentence, I think you idea here is confusing. You don’t develop the difference here between lightning-caused fires ignited by dry lightning or non-dry lightning. For that reason, I think this sentence should be deleted because you may give a wrong information. In the case of your work, the previous sentence (L 293-295) already points that that lightning-caused fires are associated to thunderstorms, and so it’s likely that the area was covered by clouds during a certain period of time (before and) after the ignition that hampers the detection of these fires.
L 302. I don’t know what errors you mean in the first part of the sentence. In the second part you explicitly say that some fires may be too small to be detected by VIIRS, which is true and a very valid point.
L 307. 298 or it should be 294 (or 292)?
3.4 Lightning-caused forest fire cases
L 319-321. Shouldn’t you add here some references?
L 321-323. I’m not sure what you mean with this sentence. Isn’t it a simple repetition of the previous one? I think you could remove. Otherwise, please be a bit more specific and add references that support your statement.
L 323. I don’t think you meant here “lightning strikes” but “lightning-caused fires with a certain burned area”
L 324-332 and 336-345. Here you talk about two lightning fires. However, I don’t understand what you want to show. In my opinion these examples adds nothing to the paper and they don’t follow what was said before. I think the manuscript would not suffer any lost from removing these two examples in the way they are presented here.
4. Conclusion
L 349-352. I think you could remove these sentences because they have little value.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Please see attachment.
Comments for author File: 
Comments.docx
Author Response
Please write down "Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Comments for author File: Comments.docx
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors made a good effort to improve the manuscript following my recommendations, congratulations! Below you can find further comments to polish the manuscript further. The major comments were all corrected and, in general, I am satisfied with the changes and the answers provided by the authors. Some minor comments are included here because I think the manuscript could be further improved without too much effort if the authors and the editor wish.
PREVIOUS MAJOR COMMENTS
1. The word “pattern”
The title is not clear enough yet. The word “pattern” alone is ambiguous because you don’t specify what kind of patterns you mean. I recommend changing the title, for example: “Forest Fire Patterns and Lightning-caused Forest Fire Detection in Heilongjing Province of China Using Satellite Data”
2. Introduction
The introduction has been improved.
3. Spatial and temporal distribution of forest fires detected by VIIRS
The authors improved the clarity of the section.
4. Capacity of satellite for forest fire detection
The authors included in the text the requested information.
5. Capacity of satellite for lightning-caused forest fire detection
The authors added the information requested. Detailed information about the methods was not really provided, but at least the section has been improved.
6. Lightning-caused forest fire cases
The section was improved.
NEW MINOR COMMENTS
1. Introduction
L 46. Here it should be “patterns” (plural) instead of “pattern”.
L 64. “area” instead of “areas”.
L 68-70. What research? Shouldn’t you add the reference from which you took this value (68%)?
L81. You don’t need “the”. It should be “…the study combines VIIRS 375m …”.
2.3 Land cover dataset
L 144. I don’t understand this sentence: “Please check the submission guidelines from the journal [25]”. I think it should be removed, and the reference [25] added at the end of the previous sentence. Furthermore, “China” should have a capital letter at the end of line 144.
2.4. Data processing.
L 165-181. I recommend that this part (i.e. the new text in red) should be a different paragraph.
3.1. Spatial and temporal distribution of forest fires detected by VIIRS
L 186-187. The current sentence is not ok, it looks like a subtitle. You should build a proper sentence with a verb. For instance: “We first analyzed the inter-annual variation …”.
Figure 4. To avoid misunderstandings, I think you should include the word pixel in the Y axis: “Number of forest fire pixels”.
3.2 Capacity of satellite for forest fire detection
Figure 7. I understand your point: satellite remote sensing is more efficient detecting lightning-caused forest fires than human-caused forest fires. However, the current Figure 7 make this too difficult to understand for the readers. In my opinion, it would be MUCH MORE effective and faster to show the proportion of fires detection every year using two lines, one for lightning and the other for human. Any reader of the paper would understand your point right away with such a figure. In any case, it’s up to you.
3.3 Capacity of satellite for lightning-caused forest fire detection
L 298-299. The sentence “(Worldview, which can be accessed at https://worldview.earthdata.nasa.gov, offers an interactive display of more than 700 global satellite impact layers.)” should be deleted. This information is not relevant for the paper.
L 323-325. This sentence “This section analyzes why forest fires caused by lightning strikes in the study area are not easily detected by satellites and quantifies the number of such fires and the area that remains undetected for various reasons” cannot be at the end of the paragrapgh. It seems an introductory sentence (i.e. at the beginning of a paragraph). I recommend to change it by a proper and concise conclusion, for example something like this: “In summary, most of the lightning-caused forest fires from 2013 to 2020 in Heilongjiang Province were not detected by VIIRS”.
3.4 Lightning-caused forest fire cases
L 337. You clearly need an introductory sentence here for this paragraph. For example: “Two specific examples are presented below.”
L 343-344. This sentence should be changed to improve its clarity. For example: “The darker area within the circle of the true color image of 20 July 2015 in Figure 9 represents the burned area.”
5. Discussion
L 404. First of all, the discussion section goes before the conclusion. I still think that this discussion (i.e. L 404-430) is quite poor and you didn’t really have a clear goal with this section. I propose the following changes:
(i) Rename “3. Results” as “3. Results and Discussion”.
(ii) Perhaps move the paragraph in L 405-478 to the last part of the section “3.4 Lightning-caused forest fire cases”, and delete the paragraph in L 419-430 because it doesn’t really fit the paper.
(iii). Remove section “5. Discussion”. Thus, section “4. Conclusions” becomes the last one of the paper.
L 409. There are hundreds of thunderstorms in spring and summer. I’m not aware of the type of thunderstorms in the region (I’m not a meteorologist but I’m familiar with some thunderstorm rainfall data from other regions), but from your sentence I understand that dry thunderstorms are the norm (i.e. the majority), which is difficult to believe unless some data is provided in the references. I didn’t read your references, but I assume that you sentence can be very misleading. Please, read well the references and considered to rewrite the sentence if dry thunderstorm only represent some of the thunderstorms, and not most of them. For instance: “… resulting in thunderstorms, some of them with minimal rainfall (i.e., dry thunderstorms) …”.
L 419. You wrote “numerous studies” but you didn’t include any reference in this sentence. In any case, I recommended above deleting the whole paragraph.
4. Conclusion
L 362. Since your wrote three conclusions, I would call this section “Conclusions”,
L 375-376. This sentence is not well written in English. Consider deleting it since it is not very relevant for the conclusions.
L 386. “were detected” seems more appropriate than “can be detected”.
L 387-392. These sentences are not well written in my opinion. Maybe something like this would improve the clarity:
“A comparison between the timing of satellite observations of forest fires with ground observations shows that 82.4% of forest fires were initially detected by satellite remote sensing. The proportion of fires detected by VIIRS decreased over time, while the proportion of fires caused by lightning increased in the study period, with a highly significant negative correlation coefficient of 0.89 between the two.
That’s all. Thank you.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The authors addressed all comments adequately.
Author Response
Dear review expert,
Thank you very much for reviewing my English manuscript and providing valuable comments. I am glad to hear that you have no further comments. Your professional guidance has been extremely helpful in improving and refining my paper. I sincerely appreciate your hard work and expert review, which hold great significance for my research and academic development.
Best regards,
Qiangying Jiao
Reviewer 3 Report
Comments for author File: Comments.docx
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
