The authors have greatly improved multiple aspects of the manuscript. The presentation of the results and discussion are now more focused and easier to follow. Their choice of using a smaller subset of main figures in their manuscript really paid off and improved the overall presentation. As requested, they also provided additional information and figures in a supplementary document, which helped me to understand the more intricate aspects of the paper. I appreciate the detailed comments the authors provided in their rebuttal, and helped to alleviate some of the major concerns I had about the paper. While the readability of the paper has improved, I think the authors need go through the manuscript once more to resolve numerous issues, typos and awkward sentences that are still present. I have listed some of the issues in my specific comments, but certainly not all of them. Furthermore, I still have a few questions and concerns about the methodology which needs to be addressed before publication.
Specific comments:
Line 6: You can leave out “(FIREMo)”. It is already introduced in the previous sentence.
Line 16: “match the fires and observations”. I don’t think you match fires. You only match CO2 observations from OCO-2 and IS. NEE is the buffer, compensating for any missing or abundant fires.
Line 22: “One major conclusion from this work is the strong constrain at global scale of the data assimilated compared to the fire prior used.” This is vague and unclear. Please rewrite.
Line 25: FIREMO => FIREMo
Line 26: “it is not the case for the majority of TCCON sites”
Line 37: “The first uses, since 2017…” You need to write this differently. It implies that all fire models based on burned area use the modification that only GFED4s uses. First explain there are two methods of emission estimation, i.e. burned area approach and FRP approach, and then make a remark that GFED4s works since 2017 somewhat differently because MODIS burned area algorithm has been updated from Collection 5 to Collection 6. That means GFED4s fluxes are not based anymore from the burned area product directly but on the relationships between climatological GFED4s emissions between 2003–2016 and MODIS active fire detections and its FRP.
This explanation should actually be included in the Appendix.
Line 44: “Two emission inventories use this approach…” Are you sure these are the only ones available? I would write “Two examples of emission inventories that use this approach are…”
Line 46: “emissions inventories” => “emission inventories”
Line 51: “trace gases emissions” => “trace gas emissions”
Line 53: “…complicates the inference a great deal ” Not sure what you are trying to say here.
Line 69: were examined
Line 72: “Rather, it is assumed that fire emissions have much lower uncertainty (generally believed to be less than 10%)”
But at line 51 you claim the errors are orders of magnitude for the emissions. Is the 10% uncertainty referring to fossil fuel emissions perhaps? Please check and correct.
Line 76: “This inference is problematic” => “This assumption is problematic”
Line 76: “emissions inventories” => “emission inventories”
Line 92: “was caused not by rising of biomass burning emissions ” Please rewrite
Line 96: “emissions inventories” => “emission inventories”
Line 104: 2016, 2017 and 2018
Line 108: “Finally, these updated fire emissions are imposed in an atmospheric CO2 inversion that constrains CO2 fluxes, using either OCO-2 XCO2 retrievals or in situ data, with different assumed fire emissions and appropriately rebalanced prior biogenic fluxes.”
I suggest to write this differently.
A suggestion:
“Finally, these updated fire emissions and appropriately rebalanced prior biogenic fluxes are imposed in an atmospheric CO2 inversion to constrain the net land and ocean CO2 fluxes using either OCO-2 XCO2 retrievals or in situ data. To evaluate these new emissions, an alternative set of fire emissions and rebalanced prior biogenic fluxes have also been used in this CO2 inversion framework.”
Line 113: You can mention the Appendix as well here.
Line 229: Are MOPITT CO and OCO-2 CO2 observations also aggregated to 3x2 and 6x2 resolution in the inversions (e.g. in the observation operator)? This needs to be explained clearly in the paper how you treated the observations in regard to the simulation resolutions.
Because fire emissions are often very local (say within ~10km) and you use very detailed CO observations at 22km, how can you justify using such a coarse resolution of 3x2 (~300x200km) for transport simulation? You need to elaborate in the methodology or discussion how this can affect your inversion results on CO emissions.
I can imagine that the simulation of pyroconvective plumes and the transport of CO high in the troposphere and stratosphere might be very challenging at very coarse resolutions. Is captured adequately, and if not how would it affect your optimized fire emissions?
I assume IS data is sampled inside TM5 at the correct elevation and location, but please explain on how you obtain a representative sample of CO2 when you simulate at a very coarse resolution of 6x4 (~600x400km). This needs to be discussed in the manuscript as well.
Line 225: “CO inversion :” remove space before :
Lines 270-278: I think is more appropriate here to explain how GFED4s has been modified from 2017 onward.
Line 279-285. Is a length scale of 1000km not too long for local fire events? Please explain. Would a smaller length scale affected your emission results? The Meirink reference is not about fire emissions but about methane emissions. So why is this a valid choice?
Line 284: “The errors are assumed …” This sentence needs to start differently because you switch topic from matrix B to matrix R.
suggestion:
In the observation covariance matrix R we only assume uncorrelated errors, meaning we only have errors along the diagonal.
Also good to explain why or why not this is a realistic assumption.
Line 289: “For each pixel…”
The pixel size is 3x2 right? Good to add that here.
“For each pixel (3x2 degree resolution)”
Figure 1: Because biofuel was added to all CO2 FIRE estimates, should biofuel burning not be included somewhere in the schematic overview?
Line 297: g.mol−1 I don’t think you need to type the dot between g and mol. Same applies for other units elsewhere in the text.
Line 307: Add here that you also included biofuel emissions to obtain FIREMo.
Line 322: The Ott reference links to the dataset of GEOS-Carb CASA-GFED3, but it does not clarify the methodology of respiration modification. Is there another reference available?
Line 331: which is in balance with
Table 3. If I take for 2017 GFED4s emissions (as an example), and add up all the other flux components (biofuel, ocean, NEEre4, and fossil), the sum is not equal the AGR_noaa. Is there something I’m missing in this logic?
4.9 = 0.486 + 10.07 + 1.78 - 4.83 - 2.6
However, AGR_noaa for 2017 is 6.06
Please explain.
Figure 2 and Table 4: Suggestion: It makes more sense I think when you name OCO-2 inversions differently. Instead of CMS-GFED3, GFED3re, GFED4re and MOre, you could name them OCOCMS, OCO3re, OCO4re and OCOMOre. That way they are more akin to the IS inversion names and more easily recognisable as OCO-2 inversions.
In the final column of boxes with inversion names you could give them the same color as used for the bar graphs (e.g. Fig. 5). So, black for CMS, blue for GFED4re, green GFED3re and red for MOre.
Line 367: Typo Coparison
Line 376: “involves” => “produces” or “yields”
Line 387: “south America” => “South America”
Line 390: “are strong” => “are large”
Line 390: “are needed” => “is needed”
Line 397: “seem to observe as mush as the prior GFED4.1s.” I would phrase it differently: “a characteristic that is shared between the MOPITT constrained fire emissions and GFED4s”
Figure 4:
- Include in the legend the colored fire emission categories
- Please provide in the final version of the manuscript a higher resolution of the figure image. The current image is somewhat blurry and hard to read.
Place ‘year’ tick markers and labels in the middle of each grouping of bars
In caption remove space in front of “ :”
Line 401: “Previous studies have shown that peat area and depth, producing large amount of carbon ( 0.60 PgC/yr which represents 26% of the total carbon fire emissions, Nechita-Banda et al. (2018)), were found to have significant uncertainties in Indonesia in the emissions inventories”
Suggestion:
”Previous studies have shown that the parameterization of peat (surface area and layer thickness) resulted in significant uncertainties in emission inventories. This is especially true for Indonesia where combustion of peat can produce significant amount of carbon.”
Line 403: “emissions inventories”
Line 404: “Our posterior have lower emissions than the prior for…” => “Our posterior fire emissions are lower than the prior fire emissions”
Line 404: “southern tropical Asia” => “Southern Tropical Asia”
Line 405: “However, Nechita-banda et al. (2018) assimilated MOPITT and NOAA observations and used GFAS as fire priors, an inversions set-up different to what we used.”
“However, Nechita-banda et al. (2018) assimilated MOPITT and NOAA observations and used GFAS as prior for fire emissions. Also, their inversion set-up was different to what we used.”
Line 406: “Additionally, no evaluation against independent data have been performed in their study to determine if their results are more trustworthy than our results. ”
“Additionally, no evaluation against independent data have been performed in their study, so there is no reason to believe their results are more trustworthy than ours”
Line 410: “that GFED4.1s have a fire peak earlier than MOPITT” => “that GFED4.1s fire emissions have a fire peak earlier than MOPITT constrained emissions”
Line 422: “for the prior4” => “for prior4”
Line 425: “decreasing from 2015 through 2018.” The net sink is actually increasing.
Figure 5
Place tick markers and labels in the middle of each grouping of bars.
“Annual prior CO2 emissions, “ add units => “Annual prior CO2 emissions (PgC/yr)”
“are represented” => “are shown”
“between GFED4.1s (blue), GFED3 (green), MOPITTopt (red) and CMS (black).”
Shouldn’t these not be called either GFED4re, GFED3re, MOre, and CMS-GFED3 according to Table 4, or prior4, prior3, priorMo and priorCMS?
Line 432: “FIRE4 and FIREMo ” labeling is still somewhat confusing. Is it not better to call them prior4 and prior3?
Line 437: “consistent with the fact that GFFED4.1s CO was observing higher CO fire emissions than MOPITT ” I don’t see the added value of this part of the sentence. The first part already makes that point.
Line 439: “observed by ” => “estimated by”
Line 441: “Northern”
Figure 6: “North”
Line 449 and elsewhere: “GFED” write it in full “GFED4.1s”
Line 451: But how sure are we that the MOPITT constrained emission seasonality is “better” (as you write at line 445) if smoke could also impacted the MOPITT observations? Is there evidence in TCCON data? Please elaborate in the paper.
Line 452: “emissions : ” again remove space after :
Line 452: For simplicity sake, I suggest to title section 3.2.2 in a similar fashion as 3.2.1:
e.g. 3.2.2 Posterior NEE and fire CO2 fluxes
Line 462: See previous comment about Figure 2. I think it is much clearer if you name GFED3-CMS as OCOCMS. The distinction between OCO-2 and IS inversions is much easier that way.
Line 464 and elsewhere: “different priors ” better to refer to “different inversions”. Now the use of “prior” labels in this particular section is confusing because we are actually looking at “posterior” fluxes. Please change this throughout the section.
Line 466: “across the priors” => “across the inversions”
Line 470: “across the priors” => “across the inversions”
Line 470: “adjusted downward” meaning larger sink? If so, be clear in the text.
Line 473: “than the other regions ” => “than in the other regions ”
Line 475: “but we know that there are a few in situ data present in 475 the SH Ext and so they have a different data constraint ” => “but we know that there are a few in situ sites present in the SH Ext resulting in a limited constrain on emissions as well”
Line 476: “across the priors” => “across the inversions”
Line 478: “sinks with in situ NEE emissions.” => “sinks with in situ observations.”
Line 479: “across the priors” => “across the inversions”
Line 481: “The net fluxes of ISMOre and IS4re look similar for the Tropical regions, while the net fluxes of IS3re and ISCMS look alike, suggesting the sensitivity in these regions to the fire prior, not only for IS but also for OCO-2 data constraint.” This sentence needs to be rewritten because it is hard to follow.
Line 485: “we can see difference in the carbon balance” => “we can see a number of differences in the inferred carbon balance”
Line 486: I don’t agree with this sentence. If I look at Fig. 7, IS data gives the largest net sink for NH ext.
Line 490: “with few data (north Asia regions).” => “where data is sparse (North Asia regions).”
Line 492: “It is interesting also to see balance ” => “It is also interesting to see the balance ”
Line 493: “sink decreased ” => “sink reduction ”
Line 496: “uptake of around ” => “with an uptake of around ” uptake where? In Europe?
Line 497: Would CarbonTracker Europe not provide a detailed European estimate?
https://carbontracker.eu
Line 501: “in 2017, as opposed to 2016 ” refer to figure 6.
Line 503: Again somewhat confusing sentence: “MOre and GFED4re inversions (respectively ISMOre and IS4re) are similar, while GFED3re and CMS-GFED3 (respectively IS3re and ISCMS) are more similar.”
Why not:
“ISMOre and IS4re inversions providing similar results (both based on either optimized GFED4.1s and default GFED4.1s emissions), while the same is true for IS3re and ISCMS inversions (both based on GFED3 emissions).”
Line 508: “for each priors” you mean “for each inversion”?
Line 512: “observation” => “observations”
Line 513: “southern and northern hemispheres ” => “Southern and Northern Hemispheres”
Line 516: “than the IS are ” => “than the IS posterior emissions are”
Line 519: “then come from different area of observation ” I don’t understand this.
Line 522: “OCO-2 net fluxes ” OCO-2 doesn’t measure net fluxes. “net fluxes derived with OCO-2”
Line 522: “high sources ” => “large sources”
Line 523: “FIREMo and FIRE4 sinks decrease” Again, these labels don’t fit well in the sentence. According to the definition of FIREMo and FIRE4 these are fire emission + biofuel burning. They cannot be sinks.
If you write it like this it is easier to understand:
Posterior net sinks derived with FIREMo and FIRE4 emissions decrease for 2017, however, the posterior net sinks derived with FIRE3 do not.
Similar issues also appear elsewhere in your text. Please correct this.
Line 525: “dependence ” => “known prior dependency of the IS posterior emissions”
Line 526: Again “southern” => “Southern ”
Line 527: Where do we see this? Respiration is not plotted
Line 526-531: This section is in particular hard to follow. Need to be rewritten.
Line 532: “between the priors ” but these are posterior emissions. You mean perhaps “between the inversions”
Line 532-535: Also this part is hard to follow.
Line 535: “Finally, for southern Tropical Asia, the inversions adjusted the NEE sinks for MOre and GFED4re to be larger than the two other inversions in order to accommodate the smaller fires observed with FIREMo and FIRE4. ”
Can be simplified to
“Finally, for Southern Tropical Asia, a larger sink was derived with OCOMOre and OCO4re than with OCO3re and OCOCMS, to balance the smaller fires derived with FIREMo and FIRE4.”
Still I don’t understand what you trying to say here. Would you not expect a larger sink when fire emissions are larger?
Line 485-546: This results section needs some additional editing as some parts are still hard to follow. The final summary paragraph was very informative. I suggest to use this as the main anchor for 3.2.2b. Each sentence of the summary text (L541-L546) can be used as a beginning for a paragraph in 3.2.2.b.
Line 549: “shows a latitudinal gradients ” => “show latitudinal gradients ”
Line 550: “northern hemisphere ” => “Northern Hemisphere ” check this throughout the manuscript
Line 550: “High land values ” => “High concentrations over land”
Line 551: “north tropical south America ” => “North Tropical South America ”
Line 554: “the OCO-2 retrievals (IS data) ” => “the OCO-2 retrievals and IS data ”
Line 559: “This result suggests that the inversion does not change much from the prior, but this result can be explain due to the small number of observations available in these regions. ” Please rewrite this.
Line 561: “For the comparison among the simulations, there is no large difference between the different simulations and the data, particularly for the optimized CO2 measurements. ”
suggestion:
“Among the different simulations, in particular, the posterior concentrations vary little in comparison to OCO-2 and IS data.
Line 586: remove “matches”
Line 587: “priorMO bias are slightly lower or smaller than prior4 ” => “priorMO biases are slightly smaller than prior4”
Line 590: “reduced through the inversion ” => “reduced by the inversion ”
Line 593: “ISMOre can be better at some tropical sites than the other simulation ” => “ISMOre provides a better match at some tropical sites than the other simulations”
Line 595: “all standard deviation ” => “all standard deviations ”
Line 597: “across the priors” => “across the inversions”
Line 609: “with each fire prior” => “with each fire emission estimate”
Line 612: “the recovery period which followed it ” => “the recovery period that followed ”
Line 613-616: Sentence is too long and difficult to understand. Rephrase please
Line 623: “on a short time basis ” => “short-term”
Line 626: “over peat fires ” => “of peat fires ”
Line 636: “have been found higher than with GFED4 ” => “have been found higher in GFED3 than with GFED4.1s”
Line 638: ‘FIREMo emissions are stronger than with FIRE4 emissions but lower than FIRE3 emissions “ => “net emissions derived with FIREMo are stronger than net emissions derived with FIRE4 but lower than net emissions derived with FIRE3 ”
Line 642: “IS4re and ISMOre have higher net sources of carbons ” = > “IS4re and ISMOre provide higher net carbon sources”
Line 649: “and IS observations But” place period at end of sentence.
Line 652: “the full dynamics” = >”the full dynamic”
Line 653: “changing the combustion efficiency and then the gases emitted ” => “changing the combustion efficiency and the ratio CO and CO2 are emitted ”
Line 655: “Further works is ” => “More work is ”
Line 656: “A recent study has shown the underestimation for Africa of MODIS burned area and consequently GFED4s, compared to the new Sentinel-2 burned area product (Ramo et al., 2021). ” => “A recent study has shown that MODIS product most likely underestimate burned area for Africa (Ramo et al., 2021)”
Line 658: “observed with previous studies ” => “The higher fire emissions estimated in previous studies ”
Line 659: “seems to suggest for future work to carefully choose the CO fire prior used in a CO-CO2 study ” Please rephrase
Line 702: “even if with a difference ” Please rephrase
Line 705: “Discrepancies between in situ and OCO-2 inversions occurred over Northern Tropical Africa where OCO-2 inversions have shown net sources while in situ inversions have shown sinks ” => “Discrepancies occurred over Northern Tropical Africa where OCO-2 inversions derived net sources while in situ inversions derived sinks ”
Line 706: “However, over Southern Tropical regions, discrepancies appear between the different set of priors, with higher net sources observed with the inversion using the CO/CO2 emission ratio (MOre inversion) for OCO-2 inversion over Southern Tropical South America and with IS inversion over Southern Tropical Asia, compared to the IS inversions using GFED3 fires. ” = > “However, over Southern Tropical regions, discrepancies appear between the different priors, with larger net sources derived with the OCO-2 inversion using the optimized fire emissions (MOre) over Southern Tropical South America and with IS inversion over Southern Tropical Asia.”
Line 709: “the constrain of priors seems” => “the priors seem”
Line 710: “seems to be better representative of the” => ““seems to be better represent the”
Lines 715-718. This section needs to be rephrased.
Line 719: “We illustrated the potential of using CO/CO2 emission ratio, and the re-balanced respiration and NEE with fire and growth rate, in CO2 inversion for better constraint and accuracy in the CO2 fire prior emissions and biospheric emission estimates. ” This sentence needs to be rephrased.
Line 725: “from the future ” => “from the upcoming”
Line 742: Use PgC/yr, as you used throughout the paper.
Supplement document: Fig. S5 is missing. Please renumber the Figs from S1 to S9 |
I have some concern regarding the emission factors used in section 2.3.1b, page 12f. The authors base their values for the emission factors of CO and CO2 on GFED4.1s, which in turn are based on a blend of Andreae and Merlet (2001) and Akagi et al. (2011). Newer estimates for these emission factors are available in Andreae (2019). These newer estimates, which are based on a much more comprehensive data base than the previous estimates, differ from the ones used here by as much as 30% in some cases. I wonder how much difference it would make if the updated emission factors would be used in the authors’ calculations.
One should also keep in mind, that in particular the EF for CO2 and consequently the emission ratio CO/CO2 are quite difficult to determine accurately in the field for a number of reasons. These include the difficulty of distinguishing the often relatively small fire inputs of CO2 from large biospheric variability, the issue of variable background concentrations, and the problem of accounting for residual smoldering emissions that do not get lofted into the smoke plumes (Guyon et al., 2005; Burling et al., 2011; Yokelson et al., 2013). This introduces systematic errors in the EF(CO2) values that may well exceed 10%. While this problem obviously cannot be mitigated here, it should be at least pointed out to the reader as a significant source of uncertainty and possibly explored by a sensitivity study.
Two minor issues:
In the caption of Table 2, van der Werf et al. (2017) should be cited explicitly (if the authors prefer to keep these emission factors).
I don’t understand what is meant by the sentence: ”Finally, the emission ratio for each vegetation type was divided to the posterior CO fire partitioned as used in Christian et al. (2003) and Basu et al. (2014).” (line 307f).
-----------------------------
Akagi, S. K., Yokelson, R. J., Wiedinmyer, C., Alvarado, M. J., Reid, J. S., Karl, T., Crounse, J. D., and Wennberg, P. O., Emission factors for open and domestic biomass burning for use in atmospheric models: Atmos. Chem. Phys., 11, 4039-4072, doi:10.5194/acpd-10-27523-2010, 2011.
Andreae, M. O., and Merlet, P., Emission of trace gases and aerosols from biomass burning: Global Biogeochemical Cycles, 15, 955-966, 2001.
Andreae, M. O., Emission of trace gases and aerosols from biomass burning – an updated assessment: Atmos. Chem. Phys., 19, 8523-8546, doi:10.5194/acp-19-8523-2019, 2019.
Burling, I. R., Yokelson, R. J., Akagi, S. K., Urbanski, S. P., Wold, C. E., Griffith, D. W. T., Johnson, T. J., Reardon, J., and Weise, D. R., Airborne and ground-based measurements of the trace gases and particles emitted by prescribed fires in the United States: Atmos. Chem. Phys., 11, 12,197–12,216, doi:10.5194/acp-11-12197-2011, 2011.
Guyon, P., Frank, G. P., Welling, M., Chand, D., Artaxo, P., Rizzo, L., Nishioka, G., Kolle, O., Fritsch, H., Silva Dias, M. A. F., Gatti, L. V., Cordova, A. M., and Andreae, M. O., Airborne measurements of trace gases and aerosol particle emissions from biomass burning in Amazonia: Atmos. Chem. Phys., 5, 2989–3002, 2005.
Yokelson, R. J., Andreae, M. O., and Akagi, S. K., Pitfalls with the use of enhancement ratios or normalized excess mixing ratios measured in plumes to characterize pollution sources and aging: Atmos. Meas. Tech., 6, 2155-2158, doi:10.5194/amt-6-2155-2013, 2013.