the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
The Fire Inventory from NCAR version 2.5: an updated global fire emissions model for climate and chemistry applications
Christine Wiedinmyer
Yosuke Kimura
Elena C. McDonald-Buller
Louisa K. Emmons
Rebecca R. Buchholz
Wenfu Tang
Keenan Seto
Maxwell B. Joseph
Kelley C. Barsanti
Annmarie G. Carlton
Robert Yokelson
Abstract. We present the Fire Inventory from NCAR version 2.5 (FINNv2.5), a fire emissions inventory that provides publicly available emissions of trace gases and aerosols for various applications, including use in global and regional atmospheric chemistry modeling. FINNv2.5 includes numerous updates to the FINN version 1 framework to better represent burned area, vegetation burned, and chemicals emitted. Major changes include the use of active fire detections from the Visible Infrared Imaging Radiometer Suite (VIIRS) at 375 m spatial resolution, which allows smaller fires to be included in the emissions processing. The calculation of burned area has been updated such that a more rigorous approach is used to aggregate fire detections, which better accounts for larger fires and enables using multiple satellite products simultaneously for emission estimates. Fuel characterization and emission factors have also been updated in FINNv2.5. Daily fire emissions for many trace gases and aerosols are determined for 2002–2019 (the Moderate Resolution Imaging Spectroradiometer (MODIS)-only fire detections) and 2012–2019 (MODIS+VIIRS fire detections). The non-methane organic gas emissions are allocated to the species of several commonly used chemical mechanisms. We compare FINNv2.5 emissions against other widely used fire emission inventories. The performance of FINNv2.5 emissions as inputs to a chemical transport model is assessed with satellite observations. Uncertainties in the emission estimates remain, and recommendations for future evaluation and use are given.
- Preprint
(2634 KB) -
Supplement
(1178 KB) - BibTeX
- EndNote
Christine Wiedinmyer et al.
Status: open (until 24 Apr 2023)
-
RC1: 'Comment on egusphere-2023-124', Johannes Kaiser, 02 Mar 2023
reply
The paper presents an update of the FINN inventory of biomass burning emissions based on satellite-based observations of thermal anomalies. Since FINN provides essential boundary conditions for CTMs and GCMs, the paper addresses a relevant modelling question within the scope of GMD. Furthermore, the calculation of emissions of various chemical smoke constituents from burnt area (observed or modelled) is itself a relevant model parameterisation.
The paper presents a novel concept for calculating burnt area from hot spot observations by the satellite-based MODIS and VIIRS instruments, new compilations of fuel loads and emission factors, the resulting emission data covering 2002-2019 and a comparative assessment thereof. This certainly constitutes sufficient advance to justify a publication.
The methods and assumptions are generally clearly presented. I would just suggest to explain the merging of the data with 2-day resolution into the product with 1-day resolution more clearly, e.g. do the values repeat for pairs of consequtive days? Furthermore, it should be justified that different pixel sizes of the two instruments both can be interpreted as 100% burned; this make sense for clusters of fire pixels but not necessarily for isolated fire pixels.
The interpretations of the data are somewhat limited due to the lack of reference observations for smoke emissions but sufficiently supported.
The description is complete and precise, in particular since the entire code is published in github. (Personally, I failed to in install the software in macOS due to the error "dpkg-deb: error: 'tini.deb' is not a Debian format archive", though. Help would be appreciated.)
Related work is properly credited and the title reflects the contents of the paper. The abstract is concise and almoste complete: I would suggest additionally summarizing the results of the data assessment and the identified regions of largest uncertainties in the abstract and the conclusions.
The overall presentation is well structured and clear. Its language is fluent and precise. Symbols and abbreviations are adequately explained except for VCF and LCT, which are, admittedly, names of MODIS products but also are abbreviations.
All parts of the paper should be published and the references and supplementary material are appropriate. Just Fig. 2 doesn't really carry relevant information and may be moved to the online supplement: Fig.2 (c) appears not to be used at all, while (a) and (b) appear to play only minor roles in the following calculations.
Finally, here are a few minor suggestions:
L.137: I would state explicitly that this results in a temporal resolution of 1 hour.
Fig.1(d): The detection rectangles near the bottom are much larger than the 110% mentioned in Line 154. Please clarify the discrepancy in the text.
Fig.1(e): Please replace "VCF" in the legend with the plotted quantity, i.e. tree cover in percent.
Fig.1(g): The IGBP land cover classes shown here are only introduced in the text in the next section. Is there a smooth way to introduce them early enough for showing in the figure?
L.221: Since annual MODIS products ared used, please state how you are planning to replace them after the lifetime of the MODIS instruments.
L.235: Please summarize the process decribed in the earlier publication for the convenience of the reader of this paper.
L.295: Please specify for each of the four instruments which time period it was used for.
Fig.4: I am somewhat surprised that BC emissions are compared but not those of organic carbon/matter. I thought that AOD is usually dominated by the latter and that emission factors for BC are comparatively uncertain. Is there a reason for this choice?
Figs.8,9: In the difference plots, it would be nice to distinguish small positive and negative values more clearly, e.g. with small transparent range around zero.
Fig.9: I suggest to also plot absolute value fields, either from the model or the observations.
I have the impression that "amount that is burned" in L.437 and "fuel consumption" in L.439 refer to the same uncertainty. If so, maybe mention it only once. If not, please clarify what the difference is.
L.474: Please justify that FINNv2 "better simulate"s emissions in the western US. This can surely not have been shown in the cited publication from 2011 when FINN2 didn't exist.
L.480: In github, the version of FINN is specified to be 2.2, with a subdirectory "v2.5_emissions_code". Which parts of the repository/project in github is described by this paper? Please clarify.
Citation: https://doi.org/10.5194/egusphere-2023-124-RC1 -
RC2: 'Comment on egusphere-2023-124', Anonymous Referee #2, 21 Mar 2023
reply
The manuscript is generally well written and informative. It presents essential information regarding an important modeling tool for fire emissions and global emissions. I suggest some minor edits to the structure of the presentation as follows:
The Results secion (section 3) begins by reviewing what is esentially part of the methods for this manuscript. The comparison to other studies should be presented in the methods section. If desired, the authors could present "model framework" methods section (what is now the Methods) and model comparison methods (move materials to a mehtods section out of Results. Or they could make one Methods section.
Similarly Sec 3.2. presents parts that are actually methods, not resutls (line 384-386: "To further evaluate ..."). I suggest this should also be moved to methods and then the results can be presented as a set of outputs, rather than in the piescemeal way currently presentedd.
Otehrwise, the material presened is excellent and worthwile.
Citation: https://doi.org/10.5194/egusphere-2023-124-RC2 -
RC3: 'Comment on egusphere-2023-124', Johannes Kaiser, 22 Mar 2023
reply
I have one more practical concern: The label "FINNv2.5" appears to denote the methodology and implementation on the one hand and the resulting emission dataset on the other hand. That is fine. However, two different datasets have been produced from different satellite input. The manuscript refers to datasets as "FINNv2.5" (e.g. Fig. 4 and Line 411), "FINNv2.5(MODIS)" (e.g. Fig. 4) and "FINNv2.5(MODIS+VIIRS)" (e.g. Fig. 7). I recommend to explicitly define which dataset is labelled "FINNv2.5", define another label for the other dataset and use these labels consistently throughout the manuscript.
Citation: https://doi.org/10.5194/egusphere-2023-124-RC3 -
RC4: 'Comment on egusphere-2023-124', Anonymous Referee #3, 22 Mar 2023
reply
This paper provides a technical description of the methods used to produce daily global fire emissions estimates from the FINNv2.5 model, which is an update to the widely used FINN model. The most significant updates to the existing model are the inclusion of fire detections from VIIRS, the estimate of event-specific area burned based on geospatial analysis, and updates to fuel characterization and emission factors.
While nothing in the paper is particularly novel, it serves as an important reference to others who will use FINNv2.5 results and should be published, pending some minor clarifications. The paper is well written and clear in most sections. The validation section is reasonable, though it is difficult to assess the uncertainties. The method of modeling CO or AOD to compare with satellite retrievals itself contains many sources of uncertainty, so it cannot be used directly to quantify the accuracy of the fire emissions. This is a shortcoming of the field, and not this specific work. I believe the manuscript could be improved by performing an additional sensitivity analysis using FINNv2.5 with VIIRS detects only.
Minor comments:
- Line 116. How was the value of 20% determined?
- Line 124. I would not describe the burning in the southeast US as agricultural burning. It is primarily small-scale prescribed burning on forested land. You could call it "managed burning."
- Line 145. Can the authors clarify how time is incorporated into this procedure? What time range is used to generate the information in Fig. 1b, for example? Is it a single day? The full year? If it is a single day, is any effort made to link events that overlap in space and are contiguous in time? Are emissions estimated for multiple days if the same location is detected multiple times? If it is a longer time period, how are the emissions allocated back to daily?
- Line 180. This series, especially Fig 1c, suggests that MODIS and VIIRS are seeing different burned areas, or perhaps MODIS is seeing a subset of what VIIRS is seeing, but with less spatial accuracy. A model run using only VIIRS and an analysis of the differences would be very instructive.
- Line 191. How are the Voronoi polygons determined? What are the points? It appears that they are some subset of the original MODIS/VIIRS detections. What is the purpose of this step? Is it to provide finer spatial resolution?
- Line 305. The differences between FINNv2.5 and FINNv1.5 are significant. The draft mentions that this is "primarily due to new processing of area burned," but can that be further quantified? Clearly emission factors play a large role for some species (e.g., CH2O).
- Line 399. Looks like a typo (...an a priori profile...)
Citation: https://doi.org/10.5194/egusphere-2023-124-RC4
Christine Wiedinmyer et al.
Christine Wiedinmyer et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 433 | 188 | 18 | 639 | 52 | 12 | 11 |
- HTML: 433
- PDF: 188
- XML: 18
- Total: 639
- Supplement: 52
- BibTeX: 12
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1