Strategic Wildfire Response Decision Support and the Risk Management Assistance Program

Round 1

Reviewer 1 Report

The study presents the history, summary and implementation process of the RMA program developed in the United States to assist decisions during singular fire events, usually large and long termed. The authors provide a general overview of the challenges addressed by the program, the potentiality and capabilities, and they also include three actual cases in which the application of the program has succeeded. They end up the work by highlighting the goodness of such initiatives under a situation where decisions made during fire events will become even more relevant to diminish negative wildfire impacts under the new era of the pyrocene.

The paper is very relevant in the current wildfire context in which more integrated knowledge and tools are needed to successfully approach the wildfire issue globally. The novelty of the paper is the successful application of the long term program and how it can inform and benefit the international community to help the decision making with a science-based approach. Although it does not directly address a forest-related question, it is interesting for the readers of Forests as it can help to manage wildfires, one of the main threats to forest resilience nowadays.

The paper is well structured, presented and extremely well written, with a proper use of English. However, I recommend small changes (minor review):

Main concerns:

1. Until the exposition of the tools and the case studies (L230), the paper has a very theoretical tone, and the reader wonders how such a program may work. The paper would benefit if the authors introduce sentences before line 230 with examples of what the authors are referring to.
2. A diagram with the history of the RMA program would be very well suited in section 3.1., with the addition of the different stakeholders and with a chronological sequence.
3. Figure 3 is not well presented. First of all, the legends should be of similar format (white background in both cases and not in top of the image, etc.), and the scale could only be placed once and avoid repeated information. The color palette makes it difficult to interpret, may be because the similar tone or because they don’t follow an intuitive gradient. The fire progression polygons are not well distinguished on top the map colors. Names on the map are extremely small and not readable. The trail 408 is barely visible. In addition, I have three main concerns regarding this Figure: 1) why the PCL in lower elevations have low values, when these places usually lead to burn all the slope area, and the mountain tops have high values, when wildfire slow down their progression? From my understanding, stopping a fire in the bottom of a slope can lead to a larger reduction in burnt area and stopping it at the top of a ridge. 2) The trail 408 does not seem to have high PCL values. And 3) you should include a new map with the crossing of both information, SDI and PCL, as exposed in Figure 2 and which its colors. This way the reader can also identify areas easily to suppress and with important consequences for fire spread.
4. In Figure 4, you should include the SDI map before and after adding the snag risk information. Thus, we can see the added value of investing in the mapping of the snags.
5. Given the nature of the program, and the section in the manuscript assigned to the TOA, you should include a case study in which different alternatives are considered and where the TOA was performed. This way we can see how the RMA is improving decision making with an actual example using this important tool. In addition, may be an example with the Aviation Use Summary would really improve the coherence of the paper, since examples with the three phases in Figure 1 would be deployed.

Minor comments:

L31-33: finish the sentence, it is not well understood.

L34-35: reference of what the strategy is.

L44-45: explain better which is this historical culture.

L55: Which recommendations? Please give an example. The reader is quite intrigued.

L78: give an idea of what do you refer with large, long duration wildfires, may be in brackets

L102: “growing” too consecutive. Please consider another term

L281: Remember to the reader what PODs are (even done, they are defined quite before)

Line 431: Is the Mullen Fire depicted in Figure 5? If not, shift the reference t Figure 5 to after the Pine Gulch Fire name

 

Author Response

Please see attachment

Author Response File: Author Response.docx

Reviewer 2 Report

General Comments:

 

 I particularly enjoyed – and thought would be interesting/useful to others - the presentation of analytic tools and case studies of how they are used in practice. However, I would like to see some major changes in the manuscript, which I summarized below.

 

• The readers could use some descriptions of how the paper is structured. For instance, it would be helpful for readers to have a clear description of the paper’s structure early-on in the article. Right around lines 70-76 would be good place to detail what each section of the paper will discuss and connect it to the above paragraphs. This is done somewhat with lines 74-76, but I would propose being even more explicit about this (e.g. In section 2, we discuss X. In section 3, we discuss Y). In the next paragraph you begin discussing cognitive biases, which is relevant, but could be disorienting if the reader is not prepared for it. There are other places in the paper where this can be incorporated too (see comments for lines 193, 231-237). I would rank this issue as a highest priority.

 

• The descriptions about SDI (in particular) and PCL were a little light. I need to read further on in the article to understand things like “SDI is a number”, “PCL is like a containment probability”, “SDI should be relativized/standardized” (I believe it should, but an explanation of why this is true is needed). Subsection 3.3 should be like a one stop shop for understanding/introducing all the fire danger analytics that are discussed later in the case studies (section 4) section.

 

• There appear to be at least some inherent limitations to RMA (see comments for line 492) and I would like to see these acknowledged in the discussion section.

 

• Are there other article formats that might be more applicable? This article has definitely value but is not really a research article per se. Double check to see if you want to submit this manuscript as an “article”.

 

Specific Comments:

 

Line 20-22: A large portion of the abstract discusses the history of RMA, followed by a single sentence about what the objectives of the article are. Elaborating on some of the items would balance this and could clarify some of the more vague items in the list (e.g. “we highlight some salient decision support aspects of RMA”, “illustrative use cases” salient/illustrative to whom?

 

Line 31-33: Apparent typo in this sentence.

 

Line 48: Insert “those”?

 

Line 61: Providing IMTs access to these resources?

 

Line 80-82: Unclear if “we focus on” means you the author are going to “focus on” large fires or if you mean firefighting operations focus on large fires.

 

Line 114-123: Most of this section is about the three-stage risk management model, and this information presented early-on seems somewhat unrelated (possibly belongs around lines 280-282). Could this subsection begin with the following paragraph “The structure and individual components of RMA…”?

 

Line 127: Start a new paragraph? Also, a more detailed description of the risk management cycle is required before you begin discussing how each of the RMA tools is used within each step.

 

Line 163-164: Unclear to the reader if…

This is a rule that only existed from 2017-2018 (you had to get a request from an AA in those years whereas it was changed in other years)?

This is a specific request our of many that is to be used as a case study.

This is somewhat resolved in line 311, but it should be clear in this first instance just the same.

 

Line 193: Consider wording “At least two issues became…”. That way the reader knows you are going to raise two points and will implicitly expect to see a “firstly” and a “second”.

 

Line 211: Remove “wildfire” from “Potential wildfire operational delineations”

 

Line 231: Rewrite this sentence without “boundary-spanning nature” which is unnecessarily vague.

 

Line 231-237: Good opportunity to give readers a roadmap to the subsections structure. You are not going to “provide a brief description of a few of the most currently most requested…” you are going to provide a “brief description of three of the most currently requested…” (SDI, PCL, and SnagMap). You could even put a colon after the sentence and list them if you thought appropriate.

 

Line 238-264: The descriptions of these indexes could be clearer. Are the indexes a number? Category? It is implied that PCL exists on a [0,1] scale, but that it isn’t actually a probability. Does SDI have a real-world interpretations or does it simply get arbitrarily larger as certain conditions overlap? On line 341 I see that an SDI> 1.01 is classified as “extreme”, and detailing how these values are computed will help readers understand how that designation was justified.

 

Line 252: “Learns” from what data

 

Line 276-279: Consider breaking up into multiple sentences.

 

Line 277: unclear what is meant by “cross-boundary”

 

Line 283-294: This is an appropriate level of detail about what the index is specifically measuring.

 

Line 295-303: Relate back to the three RMA indexes described earlier. Something along the lines of “The RMA dashboard is the framework in which decisionmakers access these three indexes as well as other important contextual information such as…”

 

Line 304-305: Consider creating subsections for each of the case studies. I like here that you specifically mentioned you have three case studies you are going to investigate.

 

Line 311: On what basis are SDI>1.01 values “extreme”? See comments for line 238-264.

 

Line 378: Would it be possible to describe what these modifications included if SDI was described more deeply?

 

Line 380: Is SDI quantifiably more accurate or does it agree with the perceptions of fire managers? If the former, it would be helpful to provide a comparison of SDI and modified SDI predictions.

 

Line 390-393: Consider moving these ideas into the discussion.

 

Line 413-415: Consider splitting into two sentences.

 

Line 416: Was standardization of SDI done before (in the previous case studies) or was this new (>2018) guidance? This is the kind of information I would like to see in regard to my comments about line 238-264.

 

Line 419-421: “The FBAN reached out to RMA analysts to inquire about the fire weather assumptions of the SDI and PCL products and was cautioned about the “relative” nature of control likelihood under highly variable conditions.” The “relative” nature of (SDI?) should be described in detail somewhere in lines 238-264. What was cautioned to the FBANs are probably related to my questions about the scale of SDI, its interpretation, etc, and should be reported clearly in subsecton 3.3.

 

Line 432: Is this PCL or probability of control? If the latter, that would seem to contradict lines 256-259.

 

Line 441: Summarize these benefits here. On line 501-503 you mention that RMA has improved the quality of decisions, and it would make sense to thoroughly detail those improvements here (line 441ish) so you can just leave the later sentence (501-503) as a kind of a self-evident claim/statement (because the justification for the claim was given on lines 441).  

 

Line 492: I’d like to see more discussion about the limitations of RMA and these indexes in the discussion. In one of the case studies for instance, the SDI values were apparently changed to better reflect perceptions of the fire managers they were intended to inform. On one hand, this is sensible because there are almost certainly going to be unaccounted for variables in any model of fire danger and having flexibility to correct for these factors could be beneficial. On the other hand, making these changes could be a trap for using the same cognitive biases that these objective tools are meant to avoid. This issue doesn’t need to be fully resolved in the discussion (you don’t need to definitively say whether the tools should be modifiable or strictly listened too), but these limitations/problems should be at least acknowledged. 

Author Response

see attached.  Responses to both reviewers are included 

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors satisfactorily addressed my concerns raised in the first review. I have only a few minor comments.

 

Line 79: Explicitly say the number of applications you will go through in this section (Like line 316).

 

Line 210-213: This sentence is a bit unclear to me. It is understandable that “direct” and “indirect” would be “insufficient to truly capture the complexity of the decision space”. However, it is not clearly presented how that objection is related to this idea “when in practice [direct and indirect] are just tactics that could both be employed in different places at different times as part of a broader incident-level strategy“.   

 

Line 284: “different but” is unnecessary

 

Line 324, 357, 411, 453: Use \subsection{} Latex command

Comments for author File: Comments.docx

Author Response

Line 79 Added 'four' to describe the number of examples

Lines 210-213.  Modified the sentence and previous introductory sentence as follows:  Second, the framing of alternatives was at times insufficient and overly simplistic to truly capture the complexity of decision space. For example, some initial TOAs compared strategies that were defined as ‘direct’ or ‘indirect’, when in practice those are just tactics and realistic incident strategies would likely employ both tactics but would vary in terms of the amount of direct relative to indirect line built.

Lines 324, 357, 412, 453 Done