This study presents an NEE dataset based on flux inversion analysis that assimilate 10 years of ACOS GOSAT v9 XCO2 retrievals. To my knowledge, this is the first published flux dataset using the full ACOS GOSAT v9 XCO2 retrievals and believe it to be a signficiant contribution to the community. I find that the dataset is well described and evaluated, and recommend publication after addressing several generally minor comments.

My most significant comment is that the atmospheric growth rate reported here seems to deviate substantially from that reported by GCP2020. From line 227 it is stated that the bias (GCAS2021 minus GCP2020) is 0.25 PgC/yr. Over the 10 year inversion, this amounts to 2.5 PgC or 1.2 ppm (using 1ppm/2.086PgC from https://acp.copernicus.org/preprints/12/C8465/2012/acpd-12-C8465-2012.pdf). I would expect such a difference to be evident in the comparison against independent CO2 data (which does not appear to show this bias). It would be useful to also compare the growth rate with that reported by NOAA (https://gml.noaa.gov/ccgg/trends/gr.html) and discuss this difference some more.

For the comparison against independent CO2 data, it would be useful to show whether these mismatches are decreased relative to the prior (would be fine to have this in the supplement).

L35: remove "could" in "ecosystems could uptake"

L40: remove "," in NBE, = NEE + wildfire carbon emission

L77-78: Please double check wether Liu et al. (2021) actually optimizes fire emissions. I think they just optimize NEE but report NBE because errors in fire would alias into NEE.

L128: What is the data assimilation window length?

L141: I did not see the qauntity "BIO" or "FCC" explicitly defined. Please make sure that these and other abbreviations are explicitly defined in the text.

L162: "pfp" should be "PFP"

L210: "low latitudes (30  S ~ 30  N, TL)" should be "tropical latitudes (30  S ~ 30  N, TL),"

L238: "NEE constraint with" should be "NEE constrained with"

L254-261: "strongest NEE" should be "strongest sink". Similarly, "weakest NEE" should be "weakest sink"

L287-289: In general, OCO-2 XCO2 or GOSAT XCO2 flux inversions find northern sub-saharan Africa to be a strong source of CO2, while in situ CO2 inversion do not (e.g., https://doi.org/10.5194/acp-22-1097-2022). However, due to the lack of validation data for XCO2 and few in situ CO2 measurements, it is hard to know for sure which is more accurate.

L324-325: I would be careful about calling this a "trend". It appears that this is primarily coming from Australia (Fig. 8j) which had very wet seasons at the start of the record (during 2011 La Nina) and has had drought for the last few years. So the "trend" might be strongly impacted by these events at the end points. Interestingly, the variability for Australia in this study looks very similar to the variability for southeast Australia reported in Fig 5 of https://doi.org/10.1029/2021AV000469.

L326: “global land NEE” should be “global land sink”

L326-237: It is unclear which regions are being correlated for each of these numbers. Please state explicitly.

L368: “Unite States” should be “United States”

L368: “Unite States in 2011-2012 (He et al., 2018; Wolf et al., 2016)”. There are several more studies that used atmospheric CO2 to study this event that could be cited:

Liu, J., Bowman, K., Parazoo, N. C., Bloom, A. A., Wunch, D., Jiang, Z., Gurney, K. R., & Schimel, D. (2018). Detecting drought impact on terrestrial biosphere carbon fluxes over contiguous US with satellite observations. Environmental Research Letters, 13(9), 095003.

Byrne, B., Liu, J., Bloom, A. A., Bowman, K. W., Butterfield, Z., Joiner, J., et al. (2020). Contrasting regional carbon cycle responses to seasonal climate anomalies across the east-west divide of temperate North America. Global Biogeochemical Cycles, 34, e2020GB006598. https://doi.org/10.1029/2020GB006598

L375-377: Could cite some previous studies that use atmospheric CO2 to study NEE over Russia in 2010:

Guerlet, S., Basu, S., Butz, A., Krol, M., Hahne, P., Houweling, S., Hasekamp, O. P., & Aben, I. (2013). Reduced carbon uptake during the 2010 Northern Hemisphere summer from GOSAT. Geophysical Research Letters, 40, 2378–2383. https://doi.org/10.1002/grl.50402

Ishizawa, M., Mabuchi, K., Shirai, T., Inoue, M., Morino, I., Uchino, O., Yoshida, Y., Belikov, D., & Maksyutov, S. (2016). Inter-annual variability of summertime CO₂ exchange in Northern Eurasia inferred from GOSAT XCO₂. Environmental Research Letters, 11(10), 105001.

L434: “on extreme climates” should be “to climate extremes”

Figure 7 caption: Give the latitude range of these regions in the caption. Rename “low latitudes” to “tropical latitudes” to be consistent with the abbreviation “TL”

Figure 10: It is hard to see the difference because the trends are so large. Please either add panels showing the data-model differences or add a plot to the supplement showing these differences.

Supplementary figure S2: This caption is incomplete. What are the individual dots? Different Years?

 The authors provide a detailed description of their new GCAS2021 dataset for monthly global NEE data over 10 years from 2010-2019, which were inferred by using their data assimilate system GCASv2 to assimilate the latest GOSAT XCO2 retrievals of version 9.  The manuscript is well written, easy to understand while clearly covering important aspects of the dataset, from their inversion method to the comprehensive evaluation using independent data etc. The resulting dataset is meaningful, and easy to use. Together with other top-down and bottom-up NEE /NBE data, it can be used to improve our understanding of terrestrial biosphere-atmosphere carbon exchange,  although (just like other NEE/NBE datasets) there are still some open questions on its reliablity, particularly over regions poorly covered by GOSAT observations.  Hence I recommend it for publication after minor correction. 

Major comments:

1    As shown in Figure 6, discrepancies from other top-down inversions (such as CMS or CT2019b) are quite significant over tropical regions, and also over South America temperate and South Africa.  As mentioned by the authors, it could be caused by different observation coverages.  

To help the reader understand the impacts of poor (GOSAT) observation coverage over regions (like the Tropical South America etc) on the top-down flux inversion, I suggest the authors include a simple comparison with their own and other groups’ inversions based on the denser OCO-2 XCO2 data for 2015-2019. Such comparisons may also help answer the question whether the high net flux in 2019 (Table 1) is realistic. 

 

2    I’d like to see more details on the assumption of the a priori error covariance,  and like to know how the authors aggregated posterior error across different assimilation windows to calculate the uncertainty for annual flux. 

Table 1 shows that assimilation of GOSAT XCO2 has reduced the uncertainty of the global annual NEE total by about 16% (i.e., from 0.6 PgC/yr to about 0.5 PgC/yr), which seems lower than other literatures.  I am not sure whether the temoral error correlations between assimilation windows have been taken into account in the calculation of those annual uncertainties.  

Minor comments: 

1    Line 31, Page 1: 'We believe that this dataset will contribute to regional or national-scale carbon cycle and carbon neutrality assessment …', 

I think this dataset can be very useful, particularly when combined with other top-down and bottom-up results. But for me,  the above statement is a bit too strong, considering the significant discrepancies with other datasets over several regions critical for global carbon cycle.  I also don’t see direct comparisons/evaluation at national scale.  I think further assessment are needed, and at this moment, it is better to say ‘‘this dataset can contribute to …’  

 

2    Line 73, Page 3:  'data are now available'

    Change to ‘data is now available…’  (to be consistent with line 74 ', which spans…' )

3    Line 135, Page 5: '… the product of CT2017…',

         ‘CT2017’   has not been mentioned before. Should it be ‘CT2019B’ instead ? 

4    Line 150, Page 5: Equation 1. 

Please define i and N.  It is a bit confusing as the gridded product was at a horizontal resolution of 1x1, but the transport model was run at 1.9x2.5. 

5    Line 239, Page 9: '…which also shown'

Should be ‘…which also showed’ 

6    Line 251, Page 9:  '…shown that'

Should be '...showed that ' 

7    Line 262, Page 9 & Figure 6 Caption

    Please use ‘TRANSCOM’ or ‘TRANSCOM 3’ consistently  

8    Line 315, Page 8 & Table 1. 

    It is better to include prior estimates for comparison.  Also,  why were the CMS and CT2019B results  not included in this table ?   

9    Table 1:

    Table 1 shows a high net flux of 6.08 PgC/yr during 2019, which is higher than 2015 (5.95 PgC/yr). It seems inconsistent with NOAA atmospheric CO2 growth rates derived from the in-situ network (i.e., 2.57 ppm/yr (2019) vs 2.96 ppm/yr (2015)). Also, to my knowledge, some inversions based on OCO-2 XCO2 data or based on the surface insitu  network showed  significantly lower net global fluxes  (upto 1 PgC/yr).  Some discussions may be needed. 

10.    Line 425, Page 15: ‘ In the Amazon basin, the simulated, CO2 profiles also agree well with the observations…’

    I think the results (Figure 12) actually suggest there could be systematic bias in the posterior fluxes over Amazon basin. 

Comments for essd-2022-15

This manuscript introduces a new global top-down NEE data from 2010 and 2019 produced by GCASv2 assimilating with a new ACOS GOSAT XCO2 L2 data. The data has been well validated by ground based and aircraft in-situ measurement that provide from OBSPACKv6. Further studies on comparison with GCP data descripts clear in this manuscript. The detail of dataset, e.g. interannual variations, trend, and profile performance is introduced. This manuscript, overall, is well written and organized, and detail is enough for potential readers and data users. I recommend it for publication, but a minor revision is required.

Line 100, the griding method is not clear, would suggest adding some equation description.

Line 155, please state the ‘a global ocean circulation and biogeochemistry model’

Line 225 and Global carbon budgets, the different between top-down estimation NBE and AGR of GCASv2 and GCP2020 comes from LULUC. I would like to indicate the average significant on this different, otherwise the improvement compared a prior and posterior is not clear.

Line 237, ‘In N. America, the distribution of NEE constraint with GOSAT XCO2 agrees well with a recent regional inversion using surface CO2 and 14CO2 measurements, which also shown significant sources over western US and sinks over central and eastern US (Basu et al., 2020).’ Please revise this presentation to avoid over-estimation on the ability of your inversion and GOSAT XCO2 measurement.

Line 265, it seems CMS-Flux using two satellites measurement in their study, incl. GOSAT and OCO-2 comes from different retrieval. The statement of measurement is not clear, e.g. satellite, retrieval algorithm and version, please revise.

Line 387, what is that ‘absolute errors’ mean?

Fig.9, this comparison method is not clear.