Evaluating the Persistence of Post-Wildfire Ash: A Multi-Platform Spatiotemporal Analysis

Round 1

Reviewer 1 Report

The paper provides an in-depth case study of mapping ash cover/depth with different sensors. It is very well written and has a very high overall merit. I have only a few minor remarks:

• l 309 ff: please be more explicit about the ANOVA, e.g. indicate in the text how the repeat measurement was treated in the ANOVA, e.g. are the samples on different days after fire considered to be stochastically independent and can this be the case if the samples are only 1 m apart? In the linear mixture modelling repeat sampling is considered (l. 317).
• Figure 7: The meaning of the x-axis is not very clear. I assume each tick represents a transect. Please clarify. Would a bar chart possibly be more appropriate than a line chart?
• l. 450: Please be more specific here. I assume the soil fraction of the soil-char mixture increases?
• Figure 10: indicate left (ash depth from field) and right panel (BNDVI) from Sentinel 2) in the figure caption to make life easier for the reader
• Figure 13: pls. refer to left and right panel of figure in caption
• Table 5: pls. briefly indicate in text how UAV costs are estimated
• Table 5, last row: should be Landsat instead of Lansdat
• l. 764: better present these pro and cons in a shorter table format.
• please format tables according to journal style.

Author Response

Reviewer 1

The paper provides an in-depth case study of mapping ash cover/depth with different sensors. It is very well written and has a very high overall merit. I have only a few minor remarks:

l 309 ff: please be more explicit about the ANOVA, e.g. indicate in the text how the repeat measurement was treated in the ANOVA, e.g. are the samples on different days after fire considered to be stochastically independent and can this be the case if the samples are only 1 m apart? In the linear mixture modelling repeat sampling is considered (l. 317).

We initially used an ANOVA for a simple analysis of the difference between groups, but did not take into consideration the possible autocorrelation between samples. In the revised manuscript we used a mixed model with repeated measures for all statistical analyses, replacing the ANOVA analysis that was used for Figures 7 and 12. We appreciate you pointing this out and we feel this has made our analysis more robust. The Methods Section 2.6 now reads: “A linear mixed-effects model [58] was run in SAS (ver. 9.4, SAS Institute, Cary, NC) to evaluate post-fire ash cover as the dependent variable using BNDVI index values, post-fire day, and burn severity fixed effects, the plot as a random effect, and sample day as the repeated measure unit. Least significant differences were used to compare differences in Tukey-adjusted least squared means of ash cover by post-fire day and by burn severity. Similarly, a mixed-effects model with a repeated measure of sample day was run for two simpler analyses: 1) to assess the difference between spectral indices over time, where the index value was the dependent variable and time post-fire was the independent variable and 2) to assess the difference in blue reflectance over time with two imagery types, with reflectance as the dependent variable and time as the independent variables. All models were considered different at p<0.05.”

Figure 7: The meaning of the x-axis is not very clear. I assume each tick represents a transect. Please clarify. Would a bar chart possibly be more appropriate than a line chart?

We agree that the x-axis was somewhat confusing and have changed both the labels on the figure as well as the figure caption to help clarify the tickmarks and data points. The figure caption now reads “Reflectance indices calculated from Sentinel-2 data over 4 image dates from the Mesa Fire. Each transect has two data points, representing each endpoint. Transects T1–T2 are high burn severity and T3–T6 are low/moderate severity. Different lowercase letters on each plot indicate significant differences between image days at p<0.05 as evaluated with the mixed model.”

1. 450: Please be more specific here. I assume the soil fraction of the soil-char mixture increases?

Thank you for suggesting this clarification. Line 458 now reads: “As expected, the uncharred soil has the highest reflectance, followed by the light gray ash, and then the reflectance generally decreases as the soil fraction of the soil-ash and soil-char mix increases.”

Figure 10: indicate left (ash depth from field) and right panel (BNDVI) from Sentinel 2) in the figure caption to make life easier for the reader

We changed the figure caption to read: “Comparing Redford Canyon Fire field and BNDVI from Sentinel-2 spectral data. Ash depth (left) and BNDVI (right) are plotted over time by transect.”

Figure 13: pls. refer to left and right panel of figure in caption

We agree that the figure caption was somewhat confusing. We edited the figure caption to now read: “Mesa Fire ash cover from 2 discrete field sampling periods compared to single-date WorldView-2 BNDVI values. The BNDVI image on the left is from 26-September (post-fire day 62); darker areas indicate high BNDVI values (i.e. green vegetation) while light gray pixels are areas of ash cover. The graph on the right compares BNDVI and ash cover values with the data subset into pre- and post-WV2 image collection.”

Table 5: pls. briefly indicate in text how UAV costs are estimated

We have contracted with private vendors to acquire UAV imagery several times and this is an approximate fair-market value. We have revised the text to acknowledge this. We added a footnote to Table 5: “This is an approximate fair-market value from the past 2 years (~$1000/day). This cost can vary widely, especially as UAS ownership and instrument availability increase.”

Table 5, last row: should be Landsat instead of Lansdat

We corrected the spelling of Landsat, thank you.

1. 764: better present these pro and cons in a shorter table format.

Thank you for this suggestion. We incorporated the bullet points into Table 5. We agree that this is a better way to compare the different sensors.

Please format tables according to journal style.

We made changes to all tables for consistency: bold headings, centered columns, and removed extra horizontal lines. Thank you for pointing these inconsistences out.

Author Response File: Author Response.pdf

Reviewer 2 Report

This is an interesting study that investigates the persistence of post-wildfire ash by using both field measurements and remote sensing images. The paper is generally well written. Minor revision is recommended.

The authors suggested the better correlations between BNDVI and ash cover. However, it should be noted that the blue band reflectance used by BNDVI is highly sensitive to atmosphere correction. In my experience, in some cases the blue band reflectance may be even negative values due to poor atmosphere correction. Such limitation should be discussed.

I acknowledge that mapping ash cover from remote sensing is challenging because of the similarity of reflectance spectra between ash and soil. As you can see that the cited literature about remote sensing of ash cover were published decades ago. The challenging and future direction for mapping ash should be carefully discussed.

Table 5 - please use the standard table format.

Author Response

Reviewer 2

This is an interesting study that investigates the persistence of post-wildfire ash by using both field measurements and remote sensing images. The paper is generally well written. Minor revision is recommended.

The authors suggested the better correlations between BNDVI and ash cover. However, it should be noted that the blue band reflectance used by BNDVI is highly sensitive to atmosphere correction. In my experience, in some cases the blue band reflectance may be even negative values due to poor atmosphere correction. Such limitation should be discussed.

Thank you for pointing this out. We had some discussion on the blue band in the original submitted manuscript, and have since revised both the methods and discussion to read: Methods on lines 295-298: “The Blue Normalized Differenced Vegetation Index (BNDVI) (Table 2) has been used for crop health evaluation and burned area mapping, and is sometimes favored with aerial imagery because the blue band has a low signal-to-noise ratio and may help account for atmospheric interference” and in the and in the discussion on lines 685-692: ” While not especially common in vegetative or burned area indices, the blue spectral region has been associated with accounting for atmospheric effects when used as a burned area index [56]. The Landsat and Sentinel satellite imagery evaluated in this study was corrected to surface reflectance, or bottom of atmosphere (BOA) reflectance, while the WorldView imagery was corrected with a dark object subtraction. Both corrections were implemented to minimize atmospheric effects during analyses. The BNDVI rather than the NDVI is sometimes used when crop mapping with aerial imagery that is difficult to atmospherically correct.”

I acknowledge that mapping ash cover from remote sensing is challenging because of the similarity of reflectance spectra between ash and soil. As you can see that the cited literature about remote sensing of ash cover were published decades ago. The challenging and future direction for mapping ash should be carefully discussed.

We agree with the reviewer that most or all of the research on remote sensing of wildfire ash was done 10 or so years ago. We have cited the most recent paper we found (Chafer et al., 2016) and we did a thorough literature search and did not find more recent papers – we do not think they exist but would welcome additional resources if the reviewer or editor knows of any. This is part of what we feel makes our research timely and relevant, as ash transport after wildfires is a high-priority concern for watershed managers. The ability to create a map of ash cover or load in a timely manner after a wildfire is needed for hydrologic modelling. Our goal was to evaluate multiple methods of creating a map of ash cover and present pros and cons of these different options. We also do not feel that we overstated our findings. We had positive relationships between BNDVI and ash cover on two wildfires, and we presented our findings as recommendations rather than absolute results. Please see lines 807-813: “Managers and data scientists need to prioritize their image needs, and time and financial constraints, when selecting an imaging platform. We also acknowledge that oftentimes it is solely an availabity issue, and for that reason, we have presented a case for mapping ash with several different platforms with reasonable success. As a final recommendation however, for the goals of this study Sentinel-2 seems to be the best-suited for mapping ash over time when the eventual next step is to incorporate the results into other models.”

Table 5 - please use the standard table format.

Thank you for this suggestion, we made changes to all of the tables for consistency: bold headings, centered columns, and removed extra horizontal lines. Responding to a suggestion from another reviewer we also incorporated the bulleted list of pros and cons into Table 5. 

Author Response File: Author Response.pdf