The paper: "Methane gas emissions from savanna fires: What analysis of local burning regimes in a working West African landscape tell us" presents some useful data on fire behaviour and methane emissions for a range of fires in West Africa. The authors aim to replicate the types of fires set by people in "working savanna landscapes" so as to provide data to test ideas around manipulating fire regimes in these landscapes to reduce GHG emissions. 

They demonstrate that because amount of biomass consumed, the moisture of the fuels, and the weather conditions all change through the course of a fire season, it is not simple to predict the emergent effects on MCE or EF. They also argue persuasively that any attempts to change the way that fires burn currently in these systems are unlikely to have the desired climate mitigation effects. 

I find huge value in the data they present, but it would have been even more useful if they could have moved beyond the crude discussions currently playing out in the literature about "early" "mid" and "late" season burning, to derive empirical relationships between the various factors affecting EF and methane emissions. If they could have presented regressions of methane emissions factors and combustion completeness against fuel moisture, rate of spread, and weather conditions then it would be a step forward in terms of understanding the trade offs between burning when fuels are moist but CC low, or when fuels are dry but CC high, and what the emergent methane emissions are under different conditions. 

I found the methods quite hard to follow and indicate where I was confused in my detailed comments below. I careful re-write aiming to help the reader through the experimental design would greatly improve the manuscript. 

Detailed comments: 

Abstract line 27-29: I thought you recommended using seasonally-varying EF for W. African fires? Not sure that your conclusions in the discussion and your conclusions here totally align. Perhaps have another look at what you want the key messages of this paper to be. 

Line 107 - "emit less"

Line 109 - Reference Russell-Smith et al JEM 2021 - this paper works hard to suggest that a shift to EDS fires is an appropriate approch to reducing GHG emissions from fire in Africa, but ends up having to conclude that there is uncertainty and that fuel moisture (in particular vegetaiton greenness/curing) is key and has not properly been resolved. 

Line 112-114: I agree completely. However, I am still a bit vague about what the "key factors noted above" are. Can you make a table or a text box or something where you list the key factors affecitng emissions factors and give an indication of the level of knowledge/data around them? that would be very helpful. 

Line 136 over 750mm as you say before,  or over 900mm?

Line 162. So you only account for curing in your moisture content measurements? What about variation due to weather conditions or position on the ground? ... Data to show that curing over rides these others?

164: more details on the Viney method? How did you dry it and did you determine fuel moisture content?

198 It is usually better to give fires a chance to get some energy before measuring rate of spread. Did you always light in the direction of the wind?

170 what 10m plot you haven't explained this yet?

200 repetition?

205. Oh, I see, here is wher you explain your calculation of fuel moisture. Can I suggest that you rework this section to make it a bit easier to follow? 

also, it seems a shame to have to use a model then you are out in the field with a scale and some real fuels....I am not sure leaf litter and grass litter respond to environmental conditions that same way as some are lying on the ground and others are standing up. Somewhere it will be nice if you present the range of fuel moisture from weather and the range of fuel moistures from curing that you calculated - I am in a discussion with global fire modellers about which of these sources of moisture is most important to get accurately and your data will be useful for this discusison

209 repetition again? 

213 - it would be nice to have more detail on how the smoke was collected? Did you wait for the fire to be fully lit? How did you get the canisters into the smoke? Did you manage to do multiple samples per fire? 

213 - also mention here that you couldn't collect LDS EF 

252 I think you have your units confused in this equation or in the text below. Is it BA percent or Ha?

Overall however I support your proposal to alter the equation. I would argue that the weather variables are accounted for by seasonal variation in CC however so it is only the variation in fuel curing/moisture that needs to be additionally included as an adjustment for EF. 

249 - I think you will have to be more clear about how you distinguish between BA (burned area) and BE (burn efficiency). As someone who has watched a lot of savanna fires I am assuming that your BE term refers to the "patchiness" of the fire - i.e. how much of the area that the fire spread through actually encountered a flame. Often in cool early-season burns there are patches of unburned fuel inside the burn scar. If this is what you mean, then I think most people account for this by taking more measures of biomass before and after the burn, and assuming that some of these biomass measures will not be burned - i.e. some of their measures of combustion completeness will be 0. 

258. I don't understand....see comments above about clarity and distinguishing between BE and BA. But now you say this can't be determined? Can you explain your logic more clearly? (PS Some people use burn severity to indicate how patchy a fire is - either way, it might be good to suggest a way for people to measure this or calculate it). 

272.I don't see how moisture content can go up when plants are curing and weather driven moisture is going down?

280. Where is table 1?

291 - aha! so the moisture content goes up in the Mid dry season (see question above) because it is perennial grasses that burn (presumably annual burn in the early dry season)? This is very important information, so needs to be clarified. I would like to develop some more processed-based way of assessing curing and moisture content that could easily be integrated by remote sensing and modelling products to account for the variability you describe here, so being clear about these processes is important if your data are to be useful

Line 306 - this is the first time you mention MCE - can you define and give units please?  (you could do this in line 64-66 where you define emissions factor. 

Table 3: indicate that MCE  is a ratio so no units

Results: I don't see anywhere where you report on the curing of the grasses in the different seasons. There are two sources of fuel moisture, and your study is able to distinguish between these. It would be good to know which of these is driving the changes in CC and BE through the season so is it possible to present these data? 

Results: as a way to move this debate forwards would it be possible to try to develop some predictive models of emission factors in relation to some of the key drivers you mention here? i.e. plot regressions of EF vs biomass, fuel moisture, and rate of spread for example. Then we can start to get at the processes driving the patterns that you find. 

Line 393 - well, it is not that different from your results - you show that EDS produce slightly lower methane emissions?  But I think this entire debate is confused by different defintions of "early" "mid" and "late" dry seasons. I would advocate for using a more empirically-based approach -see general comments above. 

Line 365 - agreed: your data show a very small change in methane emissions over the year and you can explain this by contrasting patterns of fuel amounts vs fuel moisture. How about testing whether you can explain this?  

Line 374 - agreed - as someone who has tried for years to get managers to burn savannas during conditions when fire is unlikely to spread I know that people on the ground know when fires will spread, and are reluctant to try to burn when conditions are not suitable. Theoretically, however, it would be good to have a model which predicts how CC vs fuel moisture interact to drive total emissions over the entire range of CC values. 

Line 382 - ah, finally you define BE and link it to burn patchiness. I would like to see this further up (or alternativley I suggest replacing the term BE with patchiness to avoid confusion with other terms which include the word "efficiency" in this paper. 

Line 385 - it is still not clear to me why BE and CC cant be subsumed into one measurement. Combined they represent the proportion of biomass consumed in the area of the fire scar. Even if you measure them separetly in the field, there is not reason to have two parameters in your equation (for example, when it comes to fuel moisture you just have one value, not the two that you measured in the field). So I am interested to see how they change independently over time, but no need to have two terms in your equation (I am also interested to see how curing and dead fuel moisture change independently over time)

Line 392 - very important point: this is probably why you had less large increases in methane emissions over time. 

Line 414 - 416: agreed. I think there is no one size fits all with these policy recommendations: they need to be made with reference to the particular ecological and social conditions and you demonstrate this very clearly here. 

Line 420 - good point: increasing the area burned in the EDS requires that people burn fuels that are not yet fully cured, so this will immediately increase EFs. So in the W-African situation you describe there are no social, ecological, or biogeochemical reasons to do so. 

438-440: lovely - it would be nice to bring this mechanistic explanation into the abstract more clearly 

Sally Archibald

General comments

This paper is potentially an important field-based study focusing on methane gas emissions that can provide useful savanna fire characteristics values suitable for West Africa. The authors quantified, compared, and analyzed these values by fire type and by seasons. Finally, the authors linked their findings to some practical fire management implications. I have a few specific comments that can help to improve the paper. I also suggest the authors carefully check if the reported values in the text are consistent with those in the tables.

Specific comments

Line 28: the value of 0.862 should be better clarified in the text, how is it calculated?

Line 80: From a fire modeling perspective, emission factor was usually simplified defined in most global fire models(1, 2), I am thinking if you can add some “discussion” about the possible implications of your study to the fire modeling field to broaden the interests of this paper.

Line 138: what is the relationship between cool, hot dry season and early, middle, later dry season?

Line 157: clarify what is “time of day”, ignition time?

Line 175: more description of “burn efficiency” is needed

Line 226: delete a replicated “in”

Lines 243, 252: The same symbols and/or units (for example, “Area” in equation 5 but “BA” in equation 6) should be used in equations 5 and 6 to better see their difference.

Line 271: Where did these values come from? I cannot see 4.62%, 4.09% in Table 1.

Lines 306, 308: Similar to the previous comment, 3.47 or 3.3? 2.5 or 2.9? Please check all of the values carefully. If I misunderstood, please clarify this in the text.

Line 308: The authors should also be consistent with the number of digits after the decimal point in the text and in the tables.

Line 311: 313.4 kW/m to 109.0 or 366.9 to 124.8 in Table 3? Did I misunderstand something?

Line 425, 427: 2 or 3 digits after the decimal point? should be consistent.

Tables 1-3: I suggest using “early dry season (EDS)”, “mid-dry season (MDS)”, and “late dry season (LDS)” to replace “early”, “middle”, and “late”, respectively

1. S. Hantson et al., The status and challenge of global fire modelling. Biogeosciences 13, 3359-3375 (2016).
2. F. Li et al., Historical (1700–2012) global multi-model estimates of the fire emissions from the Fire Modeling Intercomparison Project (FireMIP). Atmos. Chem. Phys. 19, 12545-12567 (2019).

This study conducts several site experiments using an approach grounded in the burning practices of people who set fires to working landscapes and collect fire-related data in West African Savanna. They find that in the dry season, methane emission factors ranged from 2.86 g/kg to 3.71 g/kg and methane emission densities ranged from 0.981 g/m² to 1.102 g/m². Overall, the results improve estimates for savanna fire emission and have important implications on earth system model development and policy making. However, I have some concerns about the presentations and hope the authors can further improve this study.

My major concern is that the connections between fire emissions and environmental factors are not explored in depth. For example, the study measured simultaneous meteorological conditions including temperature, air humidity and wind speed. How these parameters affect the fire emissions of CH₄? The authors need to plot some figures to show the relationships between emissions and weather conditions, and to identify the possible driving factors determining the differences of emissions at different stages.

To make the results more robust, the author should add more discussion about (1) the possible causes of the differences in fire-related variables at early, mid-, and late dry season and (2) reasons for different change trends among variables in dry season. In addition, the representativeness of the selected two sites and uncertainties of experiment methods need to be discussed explicitly. This manuscript is full of tables, one or more of them can be converted into figure (figures) to make the information more intuitive.

Pay attention to the consistency of tenses in the manuscript. For example, “finds” in Line 22 and “found” in Line 23.

Acronyms should be marked in the main body at the first time, not just in the abstract. For example, the author does not specify what does “MCE” stands for in the main body.

Line 191-195: The unit of I in equation (1) is wrong. The unit of product of H (kJ/kg), w (kg/m²), and r (m/sec¹) might be kJ/m/sec¹.

Line 272: “12.04% and 3.65% in the LDS” might be “12.04% in the MDS and 3.65% in the LDS”.

Line 285: “85.3% to 92.3 to 99.2” might be “85.3% to 92.3% to 99.2%”.

Equation 6: What’s the differences between CC and BE? How the inclusion of BE will improve the estimate of fire emissions without introducing more uncertainties?

Reviewer: Comments on ms bg-2020-476-manuscript-version1

Methane gas emissions from savanna fires: What analysis of local 2 burning regimes in a working West African landscape tell us

General Comments:

 The ms tackles an interesting issue and a globally important one at the same time. Improve estimation of methane (one of the most important GHG) in the fire continent (Africa) is a very interesting study. Moreover, this study will help to improve estimation of methane emitted during different fire seasons in West Africa (with high fire activity). This study is also very important for management purposes, since it is helpful to mitigate GHG emission through the best fire season applying.

The manuscript will benefit from a revision of the plan, data analyses, results and discussion sections. Data analysis section have to be rewrite. I suggest that Fire intensity and fuel moisture formula to be include in methods section. Surprisingly, all the result tables do not include statistical analysis. Statistical analysis must be done and explain in data analysis section. Arguments in this manuscript are based on the comparisons in the table 1, 2 and 3, whereas no statistical analysis have be done. In discussion section, I suggest author to integrate the many important other works undertaken in West Africa savannas on fire ecology in Burkina Faso, Côte d’Ivoire, Togo, ….

The conclusion is too long, you have to keep the main results only.

Specific Comments:

- Line 19: Among the 97 experimental fires, how many for EDS, MDS and LDS? I suggest the precision of the number of experimental fires considered for methane emissions.

- Line 28: I suggest author to propose this value for the study area as they suggest themselves and it is true that emissions are strongly influenced by the vegetation/fuel type, fire season, weather conditions,…that defer strongly among West Africa savannas.

- Line 119 to 133: this part have to be move in methods section. A resume part could be kept there.

Line 137 and 119: standardize according to the precipitation in your study area; above 750 or 900 mm?

Line 150 to 153: The clear description of vegetation (main tree and grass species, density, savanna types …) is necessary as emissions depends on it, and as precise by the authors themselves at line 61-62. Moreover, in abstract (line 20) authors suggest that they considered these aspects in their study for better estimation of emissions.

- Move from line 178 to 187 (Plot design) at line 155 (before Data collection)

- Plot design section (line 178-187) have to be clearly describe. How many plots? How many for EDS, MDS and LDS. How many repetitions for statistical purpose? The dimension of each plot for back and head fire? The distance between plots and sites? Is the seasonality define for each site based on long term data, as one site could burn during the EDS one year and during the MDS the next.

- Line 198: Why do the amount of ash is take into account for the calculation of amount of fuel consumed since usually the pre-fire and unburnt fuel are consider. Ash being a part of fuel consumed.

- Line 214: for the 36 experimental fires used for emissions estimation, how many were in EDS, MDS and LDS, back and head fires?

- In all the result section, authors have to based commentary on clear statistical analysis. Statistical analysis conclusion (for example Tuckey HDS test) could be include in table 1, 2 and 3 (comparison of different parameters between fire seasons). Sections 3.2, and 3.3 are concerned. For example (line 286): the characteristics of the fires vary by season…; the BE increased as the dry season progressed, and elsewhere…this sentences have to be based on statistical analysis showing statistical difference between seasons for BE.

- Line 286: decline BE as you use it for the first time in result section. I suggest to do it in all the manuscript…..(line 305: for MCE…).

- Line 286 and table 1 and 2: The lower fire intensity in MDS plots could be explain by higher total fuel moisture (table 1). But I don’t understand that while total fuel moisture content is higher in MDS, fire rate of spread slower in comparison to EDS and LDS, the burn efficiency could be higher in MDS than EDS. Could you explain that result please? May be that is not statistically different? I read your explanation from line 405 to 411 and I’m more confused. At line 409 you argue that during EDS grasses are often too moist to carry a fire, whereas your data showed highest moisture content during MDS.

- Line 334: may be delete the parenthesis.

- Please add dry (1) between early and season at line 374 and 392, (2) between the and season (line 379)

- Line 410: you could add the other important and recent studies on fire characteristics in West African savannas.

Your conclusion have to be reduce please.