Dead Fuel Moisture Content Reanalysis Dataset for California (2000–2020)
, Jack Ryan Drucker 1, Jeffrey Mirocha 2, Philip Cameron-Smith 2 and Adam Krzysztof Kochanski 1,*Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors present a historical DFMC reanalysis dataset by leveraging a data assimilation system that integrates simplified fuel moisture model and the observations from RAWS. This system outperforms a much more complex model of Nelson model. However, this work is generally about the extension of previous FMDA system, and in the Materials and Methods section, a clearer statement or list of the modifications proposed in this paper would be preferred to understand the main contributions. Overall, this paper provides an interesting result with improvement of wildfire model predicting the critical variable of fuel moisture. I am happy to recommend it for acceptance.
Author Response
The authors present a historical DFMC reanalysis dataset by leveraging a data assimilation system that integrates simplified fuel moisture model and the observations from RAWS. This system outperforms a much more complex model of Nelson model. However, this work is generally about the extension of previous FMDA system, and in the Materials and Methods section, a clearer statement or list of the modifications proposed in this paper would be preferred to understand the main contributions. Overall, this paper provides an interesting result with improvement of wildfire model predicting the critical variable of fuel moisture. I am happy to recommend it for acceptance.
The authors really appreciate the feedback from the reviewer. To improve clarity in the paper, the authors added the modifications conducted as part of this study as a new paragraph at the end of the Materials and Methods section.
Reviewer 2 Report
Comments and Suggestions for Authors-
1. What is the spatial resolution of the reanalysis dataset of dead fuel moisture content (DFMC) across California?
2. Over how many years was the DFMC dataset created?
3. What time span does the DFMC dataset cover?
4. What was the reduction in the mean absolute error achieved by the presented dataset compared to the widely used Nelson model?
5. How many fuel moisture observations (in hours) were integrated from Remote Automated Weather Stations (RAWS)?
6. How many different types of fuel moisture (in hours) does the presented product provide gridded hourly moisture estimates for?
7. How many types of time-lag fuels are analyzed in the dataset?
8. What is the improvement percentage in the accuracy of the presented dataset compared to the Nelson model?
9. What is the spatial resolution (in kilometers) at which DFMC data is available for the entire state of California?
10. Over how many years was the DFMC dataset developed (from 2000 to 2020)?
11. How many types of fuel moisture categories are used in the dataset (e.g., 1-hour, 10-hour, etc.)?
12. How many hours are used to categorize the fuel moisture for the most prolonged time-lag fuel (1000-hour fuels)?
13. What is the number of different fuel moisture categories for which the presented dataset provides hourly estimates?
14. By what percentage was the mean absolute error reduced in the presented dataset compared to the Nelson model?
15. For how many types of time-lag fuels does the presented methodology demonstrate significant advancements in accuracy and robustness?
Author Response
What is the spatial resolution of the reanalysis dataset of dead fuel moisture content (DFMC) across California?
The spatial resolution of the dataset, as stated in the Abstract, is 2 km resolution.
Over how many years was the DFMC dataset created?
As part of the title, the dataset expands from the start of 2000 to the end of 2020 (21 years).
What time span does the DFMC dataset cover?
As part of the title, the dataset expands from the start of 2000 to the end of 2020 (21 years).
What was the reduction in the mean absolute error achieved by the presented dataset compared to the widely used Nelson model?
As stated in the Abstract, the mean absolute error was reduced by over 2%.
How many fuel moisture observations (in hours) were integrated from Remote Automated Weather Stations (RAWS)?
The FMDA assimilates data in an hourly cycle and contains all observations from RAWS measured at the hourly interval. For more information about the cycle mechanism, the reviewer should look into Vejmelka et al. 2016.
How many different types of fuel moisture (in hours) does the presented product provide gridded hourly moisture estimates for?
As stated in the Abstract, the dead fuel classes provided as part of the dataset are 1h, 10h, 100h, and 1000h. Although only 10h fuel moisture observations are assimilated as part of the system, the other classes are adjusted based on the assimilation by the equilibrium moisture content.
How many types of time-lag fuels are analyzed in the dataset?
All metrics and analysis focus on the 10h fuel class since observations from RAWS measure 10h fuel moisture.
What is the improvement percentage in the accuracy of the presented dataset compared to the Nelson model?
It depends on the metric of interest, for the SMAPE, the improvement is more than 2%.
What is the spatial resolution (in kilometers) at which DFMC data is available for the entire state of California?
Figure 4 shows all RAWS data in California providing 10h fuel moisture measurements. The other DFMC datasets are at 2km resolution as stated in the Abstract.
Over how many years was the DFMC dataset developed (from 2000 to 2020)?
Yes, this is correct.
How many types of fuel moisture categories are used in the dataset (e.g., 1-hour, 10-hour, etc.)?Same question as 6.
How many hours are used to categorize the fuel moisture for the most prolonged time-lag fuel (1000-hour fuels)?
All the dead fuel moisture classes are provided hourly. The only difference is that the most prolonged time-lag fuel will adjust slower to the atmospheric conditions than the other classes.
What is the number of different fuel moisture categories for which the presented dataset provides hourly estimates?
Same question as 6. So, 4 classes (1h, 10h, 100h, and 1000h).
By what percentage was the mean absolute error reduced in the presented dataset compared to the Nelson model?
Same question as 4.
For how many types of time-lag fuels does the presented methodology demonstrate significant advancements in accuracy and robustness?
The presented methodology demonstrates a significant advancement in accuracy and robustness in terms of 10h fuel moisture estimates. The other fuel classes are more difficult to validate given the sparse availability of data and the accuracy of the sampling techniques. However, if the 10h fuel moisture is improved by the assimilation, it means that in general the model will use more accurate equilibrium moisture contents that will improve the estimate of the other fuels.
Reviewer 3 Report
Comments and Suggestions for AuthorsThis article presents a novel high-resolution reanalysis dataset of dead fuel moisture content for California from 2000 to 2020. The paper is written well, however, there are some points that should be explained more carefully.
Please consider the following suggestions.
1. Line 83-97. Please, formulate the goal more precisely. What is a main core here? To outline modifications implemented to the Vejmelka et al. (2016) system and to assess the accuracy of the final product? Or to carry out the comparative analysis with the Nelson model? I guess that from this point of view, the article’s title should be corrected.
2. Why the Authors did not provide the comparative analysis/test (or did not discuss) with weather reanalysis datasets like the Climate Forecast System Reanalysis (CFSR) and North American Regional Reanalysis (NARR)? They also present gridded data of fuel moisture content. Discussion in this context will widen the scientific significance and the novelty of the article.
3. Is the reduction of the mean absolute error by over 2% significant estimate?
Author Response
Comments from the reviewer and the author's answers can be found in the attached document.
Author Response File: 
Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsREANALYSIS DEAD FUEL MOISTURE CONTENT DATASET OF CALIFORNIA [2000 – 2020]
This manuscript presents the dead fuel moisture content dataset reanalysis for California between 2000 and 2020 with a modified fuel moisture data assimilation system based a model previously developed by the authors.
This fuel moisture data assimilation system uses simulated high-resolution regional weather data and 10-hour dead fuel moisture data from remote automatic weather stations to create hourly 2-km spatial resolution dead fuel reanalysis moisture data.
The main claim made is that this new model presents a mean absolute error less than more than 2% compared to the commonly used Nelson model, when estimating fuel moisture content to the observed values.
The paper is well written, and the reader will be able to follow the arguments made by the authors. The main concern with the manuscript is that the reader would have to refer to and study the authors’ previous publications, or at least be conversant with them, to fully grasp the present work.
A few notes are recommended below for the authors’ consideration:
· Line 21: Any more recent studies? These range from the years 2010 to 2019!
· Lines 24 to 26: What are the sources of this assertion? What other factors affect wildfires? [for the uniformed reader].
· Line 46: Please add [10] next to McCandless et al. [reference number].
· Lines 92 and 118: Please add [18] to in place of (2016) after Vejmelka et al. [reference number].
Author Response
Comments from the reviewer and the author's answers can be found in the attached document.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsI have no any significant comments to this version of the manuscript.
However, I reccommend to the authors add some sentences (probably, in the Discussion) about the advantages of their dataset in comparison with reanalysis data of fuel moisture components from the Canadian Forest Fire Hazard Assessment System (CFFWIS) from ERA5. This dataset contains daily values of three components (indices) of the fuel moisture at different layers with a spatial resolution of 0.25° × 0.25° in latitude and longitude: Fine Fuel Moisture Code (FFMC), Duff Moisture Code (DMC) and Drought Code (DC). Or probably, the Authors will compare with other datasets of fuel moisture that are available in the ERA5 reanalysis data.
Author Response
Comments from the reviewer and the author's answers can be found in the attached document.
Author Response File: 
Author Response.docx
