Multiscale Interactions between Local Short- and Long-Term Spatio-Temporal Mechanisms and Their Impact on California Wildfire Dynamics

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General reaction:

 

This study attempts the ambitious goal of connecting local meteorological variations in California driven by large scale climate features (teleconnections) to fire occurrence and fire risk using meteorological fields based on observations, and observed fire activity over a 37 year record.   It is generally very well written. The figures are clear, and generally do a good job in illustrating the messages delivered by the paper. I think the authors have a lot of insight into the ways that the meteorological features and teleconnections may influence fire risk and occurrence in the three regions and those messages are delivered quite clearly. The questions appearing on lines 165-171 are interesting and important.

 

I feel the study is not yet ready for publication, but that it does have the potential to be a nice paper.

 

I am not a fire expert (I am a climate scientist who has written some papers on the fire and its role in the climate system), so this review may reveal a lack of familiarity with relevant approaches, or previous literature.

 

I do have a couple general concerns: 1) I would appreciate some additional clarification on the methodology (more discussion in specific comments below) used to create the dataset used for the analysis (this is a relatively minor issue); 2) I am also concerned because there isn’t any discussion on the dangers of using correlation to infer causality, and there are no attempts to use modern ‘causal inference’ or ‘causal analysis’ methodologies e.g., Barnes et al (2019) to do a more rigorous job of connecting cause and effect. I know how difficult it is to use those techniques rigorously. Given the relatively short data record, and the fact that some of the features (for example the PDO) have long time scales it may not be possible to be rigorous about it, but I feel you should a) attempt such an analysis if you feel it is possible, or b) discuss the issues, and why you think your conclusions are robust if you do not attempt such an analysis.  You might also discuss how the study could be extended or improved (for example with a longer observational record and/or using a coupled GCM with a fire model (e.g., Zou et al., 2021) to make your work more robust.

 

On a related note, could the datasets analyzed and presented in the paper be used to both “organize” and “quantify” the relative importance of the various climate oscillation to each region in a predictive, and quantifiable manner? For example, would it be possible to use the data to create an equation of the form 

 

FWIP(i) = function(i) (ENSO, PDO, BSISOphase, BSISOamp)

 

Where FWIP is an estimate of the observational estimate of FWI for region i, and ENSO, PDO and BSISO are the indices for the climate oscillations and BSISO is a measure of the BSISO phase and amplitude? And “function(i)” is the mapping function from climate indices to FWI for region i? I could imagine that a multivariate regression or AI technique could be used to create such a mapping. Then one could might be able to make statements like these: ENSO explains X% of the variance in the FWI in region R1. PDO contributes Y%, and BSISO information contributes Z%. Combining information from the 3 regions explains XX% of the signal. Since XX is larger than X+Y+Z  it shows the synergy resulting from simultaneous information about ENSO, PDO and BSISO. Such a relationship would be useful because if any of these features are predictable (and they are) then this would make fire risk more predictable.

 

Specific comments

 

1. Line 94: You say “upward”, but I think you mean “northward”. Your phrasing might be interpreted as referring to altitude.

2. Line 244-245: It would be useful to explicitly define “Wildfire Onset Day” which I assume is the day that a particular location triggers a class F-L wildfire classification  on 1000 acres or more as defined on lines 183-184. I am assuming that for every location on your grid at 0.25 degree resolution you have one or more wildfire onset days defined. You might also state explicitly that this categorization will subsequently be used for figure 3 where (I believe) you simply count the number of wildfires each year at all locations in california and partition the count into a category (by phase of ENSO or PDO), and for figure 4 where you partition by region. Can you be explicit about whether you record only the first day that a wildfire exceeding class F is triggered, or do you also record it again if it expands and then transitions into another category? I believe this sort of info will be important if your audience intends to use this methodology.

3. Line 260-272: You should be explicit that the precipitable water (PW) is a measure of the mass of water vapor in a column (kg/m2, or mm). This concept is familiar to atmospheric scientists, but maybe not to fire scientists. Similarly, the SPW is a measure of the water in the column if each layer were saturated.

4. Line 279: You are missing a set of parentheses in equation (4).

5. Lines 387-750: This is where I feel it might be useful to make use of Causal inference analysis similar to that found in Barnes et al (2019).

6. Line 438-750: I believe that all the figures in sections 3.2-3.4 make use of all 38 years of data. Because you started by characterizing wildfire occurrence days it took a little time for me to realize that the analysis had changed from analyzing fire occurrence to connecting meteorological features to a fire risk index. I think the transition is a little abrupt and that a sentence or two near the beginning of the section might make it easier for your audience to see the transition

7. Lines 751-845: This is a nice summary of the paper results. After reading it, I felt that there were a few topics they might discuss that were missed. These are discussed in my general comments.

 

Barnes, E. A., Samarasinghe, S. M., Ebert‐Uphoff, I., and Furtado, J. C.: Tropospheric and Stratospheric Causal Pathways Between the MJO and NAO, J. Geophys. Res. Atmos., 124, 9356–9371, https://doi.org/10.1029/2019JD031024, 2019.

 

Zou, Y., Rasch, P. J., Wang, H., Xie, Z., and Zhang, R.: Increasing large wildfires over the western United States linked to diminishing sea ice in the Arctic, Nat Commun, 12, 6048, https://doi.org/10.1038/s41467-021-26232-9, 2021.

 

 

Author Response

Responses to the Reviewer's comments are attached below.

Thank you.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have presented a study on relation between wildfire occurences in diffferent regions of  California with climate drivers PDO (decdal scale), ENSO (several years scale) and BSISO (monthly scale).

The authors should note that PDO modulates ENSO and therefore  the interaction between PDO and ENSO should be considered together rather than separately (e.g in Figure 3). The authors conisder the interaction between PDO and BSISO but do not analyse the ENSO and BSISO interaction. The authors state that "this study suggest that the PDO can enhance or suppress the characteristics of specific BSISO phases, thereby influ-651 encing FWI patterns". (line 650-653). This can also be seen in PDO modulation of ENSO as well. So it can be that BSISO and ENSO are more related to each other.

The interaction PDO-ENSO-BSISO can be analysed on the effect on FWI, moisture deficits, VPD,.. in different regions of California. 

Minor comments:

(1) Line 256-258: Equation (1) and (2). The authors need to define what  es, eso, L and Rv are

es is defined as saturation pressure later but it should be defined earlier in here

Overall the results are ineesting but complex interaction of PDO-ENSO and BSISO whould be considered.

 

I recommend the manuscript to be accepted for publication with minor revision

 

Author Response

Responses to the Reviewer's comments are attached below.

Thank you.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

As the impact of global climate change becomes increasingly evident, the frequent occurrence of wildfires has become a major environmental and social issue faced by many regions around the world. This study analyzes the relationship between climate patterns such as BSISO, ENSO, and PDO and wildfire activity in California, providing a scientific basis for understanding the impact of climate change on the frequency and intensity of wildfires. This helps to improve wildfire prediction and management strategies and enhance the ability to cope with extreme climate events. Here are the suggestions for revision:

1. The spatial resolution of the climate data is 0.25°, and the area of each grid is approximately 625 km², which is much larger than the area of most wildfires. The uncertainty brought by the overly large spatial resolution of climate data to the results needs to be discussed;
2. The climate models and Fire Weather Index (FWI) used in the study have certain uncertainties, especially under the interaction of different climate patterns. These uncertainties may affect the accuracy and reliability of the research results and need to be seriously discussed;
3. Although the study discusses the impact of climate patterns on wildfire risk, the suggestions on how to specifically implement climate change adaptation measures to reduce wildfire risk are somewhat general. It is recommended to add specific and operable management strategies and policy recommendations.

Author Response

Responses to the Reviewer's comments are attached below.

Thank you.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The paper has been appropriately revised according to the reviewers' comments. I agree to publish the paper.