| 19 Jan 2026
Top-down estimate of regional carbon sinks over East Asia for 2010–2019 using satellite observations
Abstract. Given East Asia's highest CO2 emissions, quantifying natural carbon sinks in this region is essential for improving climate projections and informing mitigation strategies. We estimated the Net Ecosystem Exchange (NEE) and ocean carbon fluxes over East Asia (18.5° N–54° N, 73° E–146° E) during 2010–2019 using a Bayesian inversion framework. The GEOS-Chem chemical transport model was combined with GOSAT ACOS v9.0 XCO2 retrievals, and region-specific prior uncertainties were assigned using standard deviations from land and ocean models. Posterior estimates show enhanced carbon uptake relative to the prior, with NEE increasing from −0.17 ± 0.08 to −0.31 ± 0.06 PgC yr⁻¹ and ocean uptake changing slightly from −0.20 ± 0.03 to −0.21 ± 0.03 PgC yr⁻¹. Simulated CO2 concentrations based on posterior fluxes agreed better with independent observations than those from prior fluxes. Most subregions in East Asia acted as net carbon sinks over the past decade. Enhanced Vegetation Index (EVI) trends also support strengthened carbon uptake. However, several regions showed temporary net carbon releases in 2015–2016, likely linked to the strong 2015/16 El Niño. East Asia released a net flux of +3.45 PgC yr⁻¹ to the atmosphere during 2010–2019. Natural sinks offset only ~13.6 % of fossil fuel emissions, leaving a substantial residual source. Despite strengthened posterior sinks, they remain insufficient to counter regional emissions, sustaining elevated CO2 levels and continued outflow from East Asia.
The authors estimate regional carbon sinks over East Asia using a GOSAT-based atmospheric inversion system. Overall, the manuscript is generally well written, and the study is clearly structured, particularly in terms of how the authors evaluate the inversion results and interpret the resulting flux estimates. I also appreciate that the analysis focuses on East Asia, where carbon budget estimates remain highly uncertain, and that the results are further broken down by country/region to discuss both regional carbon budgets and their responses to ENSO-related climate variability.
However, there are several aspects of the inversion configuration that are not sufficiently described or justified (e.g., boundary conditions and the treatment of the initial state). In addition, the manuscript would benefit from a more thorough discussion of previous efforts to quantify carbon budgets over East Asia. The evaluation of the inferred flux estimates relies on a limited set of comparison datasets, and more appropriate and comprehensive datasets should be considered to strengthen the credibility of the results.
I encourage the authors to carefully address the comments below, as I believe the study has strong potential to make a valuable contribution once these issues are resolved.
Especially, the RECCAP (REgional Carbon Cycle Assessment and Processes) efforts are not discussed. Given that RECCAP provides coordinated regional carbon budget estimates in East Asia, these studies should be properly acknowledged. Where possible, the authors should also compare their inferred flux estimates against RECCAP-based budgets and related regional assessments to provide an independent benchmark for evaluating the reliability of their results.
Below I provide a key RECCAP-related reference that I recommend the authors consider. I also encourage the authors to review the references cited within this paper and incorporate additional relevant studies into the manuscript as appropriate.
Wang, X., Gao, Y., Jeong, S., Ito, A., Bastos, A., Poulter, B., ... & Piao, S. (2024). The greenhouse gas budget of terrestrial ecosystems in East Asia since 2000. Global Biogeochemical Cycles, 38(2), e2023GB007865.
For example, since this study applies a regional inversion system, one potential advantage could be the use of higher-resolution meteorological forcing compared to global inversions.
In this context, the authors’ statement that using the retrieval uncertainties provided in the GOSAT product represents a “new strategy” is not well supported. Many existing inversion systems already use the reported retrieval uncertainties from satellite products; the key issue has been whether to assimilate ocean observations given the potential for systematic biases. Thus, this aspect should not be framed as a clear methodological novelty relative to prior work.
Even if the authors are not able to explicitly account for fossil fuel emission uncertainty within their inversion framework, this issue should be discussed in more detail, particularly in the context of East Asia.
If the boundary conditions were not constrained by observations, the estimated enhancement of terrestrial carbon uptake could potentially reflect biases in the boundary conditions rather than a true observational signal within the domain. Therefore, the reliability of the inversion results may depend strongly on how the boundary conditions were prescribed, and this aspect should be described more clearly.
Based on my understanding, a common approach is to conduct an inversion for several months using an independent observational dataset and then use the optimized CO₂ concentration fields as the initial state. Alternatively, some studies prescribe published optimized CO₂ fields as initial conditions. Which approach was adopted in this study? The authors should clarify how the initial atmospheric state was generated.
In addition, the authors state that uncertainties are large over ocean regions. However, land XCO₂ retrievals generally have larger reported uncertainties than ocean retrievals. The main challenge in optimizing ocean fluxes using satellite-based inversions is not necessarily the retrieval uncertainty itself, but rather the fact that ocean fluxes are about an order of magnitude smaller than land fluxes at the grid scale. As a result, the contribution of ocean fluxes to XCO₂ variability is relatively small over the Northern Hemisphere, making it difficult for satellite observations to effectively constrain ocean fluxes. Therefore, the interpretation in this part should be reconsidered and clarified.
However, the current discussion mainly provides an indirect interpretation and suggests possible mechanisms without clearly demonstrating them. To strengthen this part, I encourage the authors to include a more direct analysis, for example by examining precipitation and/or soil moisture anomalies and quantitatively evaluating their correlations with the inferred carbon flux anomalies. Such an analysis would provide stronger support for the proposed ENSO-related interpretation.
Furthermore, CMS-Flux provides estimates of NBE (NEE + fire) as well as ocean fluxes, which could provide a more consistent benchmark. The Global Carbon Project (GCP) also provides multiple pCO₂-based ocean flux estimates beyond those from GOBMs. In addition, the OCO-2 v10 MIP provides satellite-based NBE estimates for the 2015–2020 period, which would be highly relevant for comparison. Finally, RECCAP-2 provides country-level carbon budget estimates for East Asia based on both top-down and bottom-up approaches, and incorporating these results would substantially strengthen the evaluation.
Overall, I recommend that this section be substantially revised based on more widely used inversion products and published carbon budget estimates, in order to provide a more rigorous and credible comparison framework.
Minor comments
The author's inversion system does not directly provide information on vegetation carbon uptake alone, but rather constrains net carbon fluxes of the terrestrial systems.