The Dolan Fire of Central Coastal California: Burn Severity Estimates from Remote Sensing and Associations with Environmental Factors

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study titled “The Dolan Fire of Central Coastal California: Burn Severity Estimates from Remote Sensing and Associations with Environmental Factors” uses Landsat derived NBR to explore the impacts of various environmental factors on the Dolan Fire’s level of burn severity. The study is well written and is an important examination of the relationship between burn severity and potential environmental controls on the degree of burning. The authors have done a good job in providing context for the importance of the study, outlining the various datatypes used in their research, and providing an exploration of the results. However, there is room for improvement with this manuscript.

The PDF document uploaded with his review highlights and provides comments for parts of the manuscript which can be improved. The issues which need to be addressed in order of importance are:

        I.            The statistical analysis is limited. Most of the statistical analyses are only descriptive in nature, with very little in the way of inferential statistics. The manuscript would be far more impactful if further statistical analysis was performed to better explore the impact of various environmental factors on levels of burn severity. This can take the form of expanded correlation analyses, simple regression analyses within each environmental factor, and a multivariate regression analysis between multiple factors. In its current form, the study lacks sufficient statistical analysis to draw strong conclusions.

      II.            There are aspects of the materials and methods, results, and discussion which can be improved. For example, there are parts of the results which would be more appropriate in the materials and methods. There are also parts of the discussion which refer to tests performed on the data which are not mentioned in the methods and results. Please see the PDF for more details.

    III.            There is a figure used in the materials and methods (Figure 2) which appears to be borrowed from another study. Was proper permission granted for use of this figure?

    IV.            There are some typos and improper use of words/units in the paper which should be addressed.

This study has potential and I encourage the authors to address these concerns as I believe they will make the paper more impactful.

Comments for author File: Comments.pdf

Author Response

Kindly find the responses to Reviewer #1 in the attached document.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Some fire studies examine a large number of fires; others look at one or a few fires in depth. This study falls in the latter category, but the depth isn't very deep. Specifically two elements are missing.

 

Firstly, the study would benefit from an exploration of the wind-slope-vegetation interactions that could explain both fire spread and burn severity. Given the DEM, a wind model could be used to derive wind speed and direction at high resolution, either using the prevailing wind direction, or (even better) using weather observations during the course of the fire.

 

Secondly, there is hotspot and burn perimeter information available for this fire, which would allow for a reconstruction of daily fire spread. There is a lot of information about the Dolan fire online, including perimeter maps from NIROPS made at multiple times during the fire's progression. CalFire must have additional information and maps, considering the scale of the suppression efforts. Hotspots are readily available from FIRMS and can be used to map and characterize fire spread at better than daily resolution. There may also be other Landsat-resolution imagery available (e.g. Sentinel-2).

 

The goal of both of these elements would be to obtain the best possible approximation of wind speed and direction for each pixel at the time of that pixel's burning. Knowing the day and approximate time of burning allows for the use of daily or hourly meteorological observations. If the temporal resolution of observed fire spread is still inadequate, use a fire growth model to fill in the gaps.

 

It seems that the Anderson fuel models aren't specific/detailed enough for a study like this. Landfire has a lot of other maps describing and quantifying the characteristics of the vegetation/fuels, which are not considered in this study.

 

Lastly, there is a significant number of typos, questionable grammar and punctuation, and potentially confusing terms and sentences that reduce the effectiveness of the paper. I know many of these are trivial (others are not) and don't matter individually, but taken in aggregate the effect is to slow the reader down and ultimately prevent the authors' point from getting across.

 

Line 43. "Belligerence" is defined as "a warlike or aggressively hostile nature, condition, or attitude; an act of carrying on war; warfare". I have not previously seen this word to describe arson. There are various factors motivating arsonists, many of which, such as mental illness, have nothing to do with belligerence. Also in line 419.

43. "Population growth has increased the number of humans and infrastructure in the wildland-urban interface." True, but also significant is the enormous expansion of the WUI as a result of people wanting to, and being able to, live in or near forests.

51. Flame height is not a characteristic of vegetation/fuel.

53. Live vegetation retains its moisture, even during drought, while dead vegetation dries out completely. In other words, the drying comes after the dying. However, we know that the trees don't die every summer. Rather, needle-leaf trees are very flammable even while alive, as are some broadleaf species (e.g. eucalyptus).

57. "the Sierra Nevada daily meteorological variable" - which variable is that? It sounds as though the Sierra Nevada has only one meterological variable, which is unspecified, unlike the rest of the world where weather includes multiple variables like temperature, humidity, wind speed, precipitation, etc.

67. "which usually flows uphill" - not sure how true this is, or how helpful it is to mention it here. It's true that fires spread more rapidly uphill, even in the absence of wind. When there is wind, fire will happily spread downhill. In fact California is famous for downslope winds. Foehn winds can produce rapid downhill spread, and contributed to the disaster in Lahaina in 2023.

68. It is not primarily wind gusts that cause crown fire initiation, but flame length and ladder fuels.

70. "plume" - Consider clarifying here that you're referring to the turbulent/buoyant plume, not the smoke plume. Better yet, cite a paper about using wind modelling for fire behavior modelling. A bunch of them are listed at the bottom of this page https://www.fs.usda.gov/research/rmrs/products/dataandtools/tools/windninja (For the record, I have no involvement in this site or any of these papers.)

77. This sentence essentially restates what has just been said, with the confusing phrase "from the fire line to the direction of steeper slopes". Note that whether the "radiant energy transfer" is "controlled" in this way also depends on aspect, wind speed, and wind direction.

78. There is an incongruous mixture of small scale (microclimate) and large scale (fire regime) here.

80. "choose": Vegetation doesn't choose. Of course it's common to attribute intelligent agency to animals, plants, and all kinds of objects, but primarily in popular rather than scientific literature.

81. Wordy, circular sentence

91. "Previous studies were limited by... empirical relationships with ecological metrics." I haven't read reference [22], but could it be that studies were limited because they failed to take into account relationships with ecological metrics?

92. "Some studies attempted to compare... methods using a limited number of vegetation classes." Perhaps use "inadequate" instead of "limited". All studies use a limited number of vegetation classes. There is no alternative.

93. Other studies used machine learning. Is there anything wrong with that? This appears to be mentioned as part of a list of reasons why previous studies were "limited".

102. Slope and aspect are not topographic features. Mountains, valleys, plains, outcroppings, gullies, plateaus are topographic features. (Also in line 312.) Suggest: How do slope and aspect affect burn severity? Better: How does topography (slope and aspect) interact with wind and fuels to influence burn severity?

104. "formulism" means "adherence to or reliance on formulas". This word could be removed.

105. The phrase "the fuel model" (also in line 99) implies there is only one fuel model, whereas in fact there are several.

105. The sentence beginning with "An overview..." is incomplete.

106. "on" should say "in".

Table 1. NIR and SWIR are often used interchangeably, as can also be seen from the complete overlap of the frequencies given. This overlap however is obscured here with with the use of microns for one range and nanometers for the other.

113. Why was this fire selected for study? Is it unusual? Typical? Mysterious? Impactful?

118. "is": Present tense here is jarring. Activities that happened in the past should be in the past tense, as in the preceding lines in this paragraph. (Also in multiple places elsewhere in the manuscript, e.g. lines 174, 314; cf. past tense in lines 318, 320.)

122. Move "average" to the beginning of the sentence.

123. "potential vegetation fire fuels": How about "The fuel types include..." or "the vegetation types include..."

124. Comma splice - also in lines 170, 176, 189, 374

126. Cone Peak is not pictured in Figure 1.

172 "by which water vapor condenses onto" should be "onto which water vapor condenses"

190. "geospatial and field data": Field data is also geospatial data, so it's unclear what these two categories refer to.

190. "vegetation fire fuels" - This could say simply "vegetation" or "fuels".

191. Why is "polygons" plural here?

192. This sentence should be in the section below on weather data.

198. I'm not familiar with the Anderson models, but aren't flame height and spread rate outputs of the model, rather than characteristics of the fuel? The spread rate for example depends on the wind speed and fuel moisture, so that any given fuel (vegetation) type can produce a wide range of spread rates. Consider adding an explanatory sentence for oblivious people like me.

199. The phrase "before fire has occurred" makes this sentence confusing. Is the moisture of extinction one value before the fire, and another afterwards?

233. Repeats what is said in line 71ff.

234. "them" - unreferenced pronoun

237. "is determined by" - should be "is"

242. "NBR burn index"... How about "Burn severity, as represented by dNBR, was calculated using Equations 1 and 2"

246. The MTBS perimeter contains unburned pixels, which presumably/hopefully are not included in the average dNBR. There may be burned pixels outside the MTBS perimeter as well; it would be worthwhile to look for burned pixels within a reasonable buffer distance outside the MTBS perimeter.

266. Hotspot data could be used to derive fire progression at a higher temporal resolution.

283. "FMFM13"

299. The abundance of young brush doesn't "change" per se, as if young brush appears and disappears on the landscape.

Figure 8. It would be more intuitive to have a chart for each burn severity class. Though it's carefully explained otherwise in the text, the impression conveyed by these charts is for example that 60% of the young brush was unburned, while 40% burned at low severity, etc. The use of a line chart rather than a bar chart gives the impression that there is temporal progression from unburnt, to low, to moderate, etc., which isn't the case. Note that table 6 could be expanded to include additional columns, one for each severity class. But the goal here (I think) is to convey that some fuel types are more prone to high-severity fire than others. Could this be done better in a different way? Maybe a stacked bar chart?

317. These averages are pretty similar. Are the differences statistically significant?

317. Would it be reasonable to expect that aspects more exposed to the sun (i.e. south) would be dryer and have higher burn severity? If so, why aren't we seeing that? It could be for example that the south slopes have lower fuel density because lack of moisture limits vegetation growth. Slopes facing the wind may dry out more quickly, or alternatively may receive more rain due to orographic uplift. Lee slopes may be in a rain shadow and therefore be dryer. These and other possibilities should be considered.

332. Yes, the results suggest there is a relationship; and what is that relationship? Maybe the burn severity decreases at very steep angles because the fuel density is lower, i.e. fewer trees able to grow on steep slopes. Maybe the steep slopes have rocky soil or nutrient-poor soil.

345. "WNM"

356. "There is a lack of studies regarding the burn severity of the Dolan Fire". This implies that there ought to be several studies of the burn severity of the Dolan Fire, and by extension, every other fire as well. This would constitute thousands of studies.

360. "were developed were adept"

372. This is the first time we hear that the burn severity of this fire has been mapped already by someone else. This deserves more than just a passing mention. Reading between the lines suggests that MTBS used a different method for mapping burn severity, and different criteria for the classes. What was that method and what were those criteria? How do we know that your method is better?

381. "Slopes facing in the direction of oncoming fire spread were more susceptible to increased burn severity." This contradicts line 408, which says that north and east-facing slopes had higher burn severity.

388. Here's a partial explanation of the Anderson fuel models. This should be in the data/methodology section, not in the discussion section.

395. Just to clarify what's going on here: The dead fuel load data you're correlating comes from Anderson, not from the Dolan fire. So the fuel loads associated with the Anderson fuel models selected in this case were right, or happened to line up with the observed severity. However, Anderson wasn't right about the flame height or spread rate. In other words, this is not a test of which fuel characteristics contribute to burn severity, but a test of (a) whether the correct fuel models were selected by Landfire in the Dolan fire area, (b) whether the various fuel characteristics are always associated in the way Anderson expects, (c) whether 13 fuel models can be expected to accurately represent the variety of fuels in this area and indeed in the whole country, and (d) whether Anderson was a good choice for this study. In any case, the variables that showed no correlation (flame height and spread rate) depend in part on the wind speed - as is acknowledged in line 400.

397. "severity. ty"

400. Temperature, topography, etc. don't only affect severity, but also flame height and spread rate. It seems the Anderson system specifies flame height and spread rate without knowing these variables, which is a significant weakness.

412. Preheating fuel uphill is described now for the third time (after lines 71, 233).

413. "east facing slopes are adjacent and trend in the same direction of [as?] the wind" - yes, but the slope is downhill, and not only on south-facing slopes as in 415. What's true for south must be true for east, but it's not. Why not?

421. It's not temperature changes per se that strongly influence fire, but high temperatures. A temperature change from -30 to -10 is also "massive" but doesn't produce extreme fire behavior.

425. "Without better temporal resolution": yes, but better temporal resolution is available.

432. Yes, but the fire wouldn't have spread if the fuel was wet. While there was no "drought" as defined by EDDI, fuel moisture content is a well-established and significant factor in fire behavior. Maybe the conclusion to be drawn here is that EDDI drought categories are not helpful in predicting fire spread.

452. The Black Dragon paper is not the only one out there about burn severity and Carbon emissions, and unlikely to be the best because the fire occurred at a time when far less satellite data was being captured. See for example Hall et al. 2008. Remote sensing of burn severity: experience from western Canada boreal fires. International Journal of Wildland Fire 17(4): 476-489.

 

Comments on the Quality of English Language

Included in general comments.

Author Response

Kindly find the responses to Reviewer #2 in the attached document.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have done a good job addressing my primary concerns.

Author Response

Comments in attached PDF

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for considering some of the suggestions in the first review, and for all the responses and changes made to the paper. In particular I appreciate the new figure 8. I have only a few nitpicky comments (you know me by now)...

67. "display" should be "displayed"

Table 1. The column headers (titles) are missing, and the second row of the table is incomplete (missing "Landsat-8 Collection 1")

143. "designed to dis." The rest of the paragraph is missing.

193. "includes" should be "include"

347. The relationship is weak because the response is not linear. Fitting a straight line to a curved response results in an underestimate of the strength of the relationship. A quadratic function would fit better.

Figure B1. In the PDF I downloaded, only the bottom portion of the image is visible, and is overwritten by the title of Figure A4. 

 

Comments on the Quality of English Language

Included in general comments.

Author Response

Response in attached PDF

Author Response File: Author Response.pdf