Two Decades of Satellite Observations of Carbon Monoxide Confirm the Increase in Northern Hemispheric Wildfires

Round 1

Reviewer 1 Report

This communication work estimates of CO emissions from biomass burning using a simple box model in the case study of the Northern Hemispheric Wildfires Increase.

 

1.     For easier understanding, please explain more briefly information about methodology before subsection 2.1

2.     Based on the correlation in table 1, why do R in USA and Japan lower than in other sites? Please included the discussion.

3.     Please added some briefly information about GFED4 data in methodology section.

4.     Because GFED4 data has a resolution of 0.25 degree, how do author compare those GFED4 data with AIRS data (Figure3) ? Please included some explanation in the texts.

5.     Also, please some briefly information about FTIR and MOPITT data in methodology section.

6.     Table 9, how do authors used the MOPITT data? Are there some algorithms to use those MOPITT data? If so, please explain in some texts.

 

 

Author Response

The authors would like to thank both reviewers for reading our MS and for their helpful suggestions to improve it.

 

A general improvement that facilitates a better understanding of techniques and results is a new Figure (Figure 2 in the new version). A leading role of boreal forest fires and a striking difference between 2020 and 2021 fire seasons became more pictorial.

 

Another important point is a fundamental drawback of TIR satellite data: namely, a reduced sensitivity to the lower part of atmosphere. To our mind, this point needs a further clarification, that is beyond the scope of this brief communication. Generally speaking, there is a consensus in our community that midtropospheric data for gases (e.g., CO, CH4) are more or less reliable. Also the TIR instruments are practically insensitive to the first km or km and a half of the altitude. But a usefulness of the range 2-4 km is a matter of discussions. We use here Xco (TC); the total column is influenced by this drawback, even a suggested ~30% correction factor does not eliminate this problem completely. This correction, however, temporarily allows to assign ±19% error caused by the sensitivity problem. A brief discussion is added to the 2.2 Validation Results section.

 

Specific comments

1.     For easier understanding, please explain more briefly information about methodology before subsection 2.1

Fig. 2 and Lines 115-125.

2.     Based on the correlation in table 1, why do R in USA and Japan lower than in other sites? Please included the discussion.

See lines 131-132

3.     Please added some briefly information about GFED4 data in methodology section.

See lines 174-186

4.     Because GFED4 data has a resolution of 0.25 degree, how do author compare those GFED4 data with AIRS data (Figure3) ? Please included some explanation in the texts.

See lines 184-185

5.     Also, please some briefly information about FTIR and MOPITT data in methodology section.

See lines 116-126. MOPITT data are not used in this study, only as citation and comparison in Table 2.

6.     Table 9, how do authors used the MOPITT data? Are there some algorithms to use those MOPITT data? If so, please explain in some texts.

Table 9? The reviewer has Table 2 in mind, doesn't he? The last two columns are citation of [9]. Accuracy of MOPITT for long-term measurements is a special issue. Its design (gas-correlation spectrometer) is principally different from that of AIRS. We do not need to discuss these issues further in this communication. Moreover, our analysis is based on AIRS only.

Reviewer 2 Report

The manuscript provides an additional method to investigate the variation in the CO concentrations emitted from forest fires using the remote sensing data. The study results are interesting, and the writing is sufficiently good. A minor revision is recommended.

1. Several corrections are necessary in lines 56, 72, and Table 1 (first row). Lines 65-66 also need correcting with the position of “2000-2019”.

2. More information on the methods and fitting parameters to calculate the average seasonal cycle (trend + cycle) and the polynomial trend should be provided in lines 114-118 and 164-170.

3. Please also check Figure 2 for polynomial trend and trend + cycle. It is difficult to distinguish between graph lines.

Author Response

The authors would like to thank both reviewers for reading our MS and for their helpful suggestions to improve it.

 

A general improvement that facilitates a better understanding of techniques and results is a new Figure (Figure 2 in the new version). A leading role of boreal forest fires and a striking difference between 2020 and 2021 fire seasons became more pictorial.

 

Another important point is a fundamental drawback of TIR satellite data: namely, a reduced sensitivity to the lower part of atmosphere. To our mind, this point needs a further clarification, that is beyond the scope of this brief communication. Generally speaking, there is a consensus in our community that midtropospheric data for gases (e.g., CO, CH4) are more or less reliable. Also the TIR instruments are practically insensitive to the first km or km and a half of the altitude. But a usefulness of the range 2-4 km is a matter of discussions. We use here Xco (TC); the total column is influenced by this drawback, even a suggested ~30% correction factor does not eliminate this problem completely. This correction, however, temporarily allows to assign ±19% error caused by the sensitivity problem. A brief discussion is added to the 2.2 Validation Results section.

Specific comments

1. Several corrections are necessary in lines 56, 72, and Table 1 (first row). Lines 65-66 also need correcting with the position of “2000-2019”.

It is done. See lines 64 and 80, also in Table 2

2. More information on the methods and fitting parameters to calculate the average seasonal cycle (trend + cycle) and the polynomial trend should be provided in lines 114-118 and 164-170.

See lines 152-159 and tables A2 and A3

3. Please also check Figure 2 for polynomial trend and trend + cycle. It is difficult to distinguish between graph lines.

Thick dash line is used now for trend in Figure 3 (former Figure 2)

Round 2

Reviewer 1 Report

I have accepted a manuscript in the current version.