the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 May 2026
Assessing Air Pollution Drivers in Asia Through Multi-Species Data Assimilation During NASA's ASIA-AQ Campaign
Abstract. Asia accounts for approximately half of global anthropogenic emissions of major pollutants, yet emission inventories remain uncertain and ground-based monitoring is sparse across much of the region. To address these challenges, this study applies a multi-species satellite data assimilation framework to estimate emissions and concentrations of key chemical species during the NASA's ASIA-AQ campaign. The assimilation improves agreement with airborne observations for O3, NOx, CO, and CH2O, with the largest gains for CO (correlation increasing from 0.63–0.64 to 0.77) and CH2O (biases reduced by 41–70 %). Domain-wide, the optimized emissions show increases of 15 % for NOx and 9 % for CO, and a 52 % reduction in isoprene. Comparisons with multiple emission inventories reveal large discrepancies, with normalized standard deviation ranging from 11 % for NOx in mainland China to 68 % for CO in Taiwan. Over Thailand, the assimilation increases fire emissions from 3.6 Tg (GFASv1.2) to 8.3 Tg, while FINNv2.7 produces estimates roughly twice as high, highlighting persistent divergence among fire emission estimates. Source-receptor analysis reveals strong meteorological control on transboundary pollutant: long-range transport contributes up to 62 % of surface O3 in Manila during strong monsoon conditions, whereas Seoul exhibits local NOx-saturated chemistry. During a strong transport episode, Indian emissions account for 72–78 % of the free tropospheric O3 response over Taipei and Chiang Mai, and 24 % over Seoul, highlighting an overlooked transport pathway affecting Asian air quality. These results highlight the value of satellite data assimilation and the need for improved inventories and coordinated action on local and transboundary pollution.
- Preprint
(8240 KB) - Metadata XML
-
Supplement
(12344 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2380', Anonymous Referee #1, 13 Jun 2026
-
RC2: 'Comment on egusphere-2026-2380', Anonymous Referee #2, 10 Aug 2026
Choi et al. perform a comprehensive study aiming to assess air pollution in Asia backed by the NASA ASIA-AQ airborne mission campaign measurements of ozone, NOx, CO and CH2O. To this purpose, they apply a multi-constituent satellite data assimilation approach using MLS profile retrievals of ozone, CO and HNO3, NO2, CH2O and SO2 columns from TROPOMI, and VIIRS AOD products. The baseline framework assimilates CO column data from MOPITT, whereas a sensitivity experiement using TROPOMI CO columns is also conducted and evaluated against ASIA-AQ data. The chemical assimilation method was previously developed and used in a wide range of studies (Miyazaki et al. https://doi.org/10.5194/essd-12-2223-2020). The current Tropospheric Chemical Reanalysis (TCR-3) includes only minor updates relative to the previous versions (TCR-2). After assimilation, the agreement with the airborne data was generally improved, as a consequence of moderate emission increases for anthropogenic NOx and CO, a strong decrease for isoprene, and a pronouced increase in fire emission estimates in Thailand. The analysis reveals that meteorology governs the transboundary pollution in this region. The study is useful for the community. The manuscript is clear, well written and the results look correct. I recommend publication after addressing the comments below.
- The reasons for the large differences (Figure 5) between the two ASIA_AQ ozone measurements acquired with different techniques (in situ cavity-enhanced UV absorption of ozone and chemiluminescence) should be discussed. In Taiwan, for example, if NOx emissions were increased to match the ASIA-AQ NOx observations, this would lead to ozone levels much higher than the measurements. This does not seem consistent. Please expain.
- The use of TROPOMI CO in TCR-3 significantly improves the comparison with ASIA-AQ measurements relative to MOPITT, as seen in Fig. 5. The difference between the two assimilations is quite large but is not adequately explained. Are the TROPOMI clumns higher than MOPITT in this region? Comparisons between TROPOMI and MOPITT (Martinez-Alonso et al. https://amt.copernicus.org/articles/13/4841/2020/amt-13-4841-2020.pdf) indicated an excellent agreement over land, with TROPOMI only slightly lower than MOPITT. Can you provide additional analysis and inependent measurements to explain the large difference? On the other hand, the TROPOMI TCR-3 emissions are much lower that the MOPITT TCR-3 in Figure 8f over Thailand, but this seems to be inconsistent with Fig. 5l. Please explain. Could you provide additional plots as Figure 6(b,e) but for TROPOMI TCR-3?
- The large inferred decrease of isoprene emissions in South East Asia is well supported by ASIA-AQ CH2O measurements, and a even stronger decrease would be required, especially in South Korea and Thailand (Fig. 5o,p). However, recent top-down isoprene emissions based on TROPOMI CH2O data (Sfendla et al. https://doi.org/10.5194/acp-26-733-2026, Li et al. https://doi.org/10.1021/acs.est.5c12927) and CrIS isoprene (Li et al. https://doi.org/10.5194/essd-17-7035-2025) found generally higher annual emission fluxes than the a priori. Could you compare with your results for the specific months? It is also recommended to evaluate against other bottom-up estimates, e.g. with CAMS-GLOB-BIO (Sindelarova et al. https://doi.org/10.5194/essd-14-251-2022).
- The coarse resolution of the assimilation is a limitation when we address short-lived species. As seen in Figure 5, the NOx emission increase in TCR-3 is relatively small and the model remains too low compared to the ASIA-AQ measurements. What is the posterior model bias with respect to the satellite NO2 used in the assimilation? Do you have the possibility to conduct a control run at finer resolution? Could you add a comparison with available Pandora measurements in your target box?Citation: https://doi.org/10.5194/egusphere-2026-2380-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 417 | 213 | 32 | 662 | 64 | 36 | 38 |
- HTML: 417
- PDF: 213
- XML: 32
- Total: 662
- Supplement: 64
- BibTeX: 36
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments:
This manuscript presents a comprehensive study assimilating satellite observations of six key chemical species into the MIROC-Chem model to simultaneously optimize atmospheric concentrations and surface emissions. The assimilated results are extensively validated against the recent NASA ASIA-AQ airborne observations, demonstrating a substantial reduction in model biases. The optimized emissions reveal significant discrepancies in current bottom-up inventories for anthropogenic sources, biogenic sources, and biomass burning. Furthermore, the study provides a detailed source-receptor analysis quantifying the relative contributions of local emissions versus long-range transport to air quality in various Asian cities. Overall, the methodology is solid and robust, and the conclusions are well-supported by the analysis. The manuscript is clearly written and well-structured. I recommend publication after the following specific comments and clarifications are addressed.
Specific Comments:
1. The current inversion framework attributes all simulated concentration biases to errors in surface emissions. How can we be certain that these biases do not stem from inherent deficiencies in the model's physical or chemical mechanisms (e.g., boundary layer mixing, deposition schemes, or chemical reaction rates)? I suggest adding a brief discussion on the potential uncertainties introduced by model structural errors and how they might alias into the optimized top-down emissions.
2. Please elaborate in the methodology section on how the data assimilation system handles this strong non-linear chemical feedback (for example, the strong non-linearity in ozone chemistry). Does this non-linearity pose a risk for optimizing NOx and VOCs emissions during the assimilation process?
3. The study assimilates six gas-phase species but does not mention aerosols (e.g., PM2.5 or AOD). Asia, particularly India and Thailand during the fire season, experiences extremely high aerosol loadings. Heavy aerosol plumes can significantly alter photolysis rates. I recommend evaluating, or at least discussing, the potential errors introduced into the gas-phase photochemistry and the subsequent emission inversions due to the lack of aerosol assimilation (or unconstrained AOD).
4. How is the injection height of Southeast Asian wildfires parameterized in the model? Since the vertical distribution of emissions is critical for simulating long-range transport, did the ASIA-AQ airborne observations provide any constraints or validation for the modeled plume injection heights? Please provide more information on this in the text.
5. Line 376-385: The model scales down biogenic isoprene emissions by a striking 52%. Is this massive reduction primarily due to an overestimation of basal emission factors in the prior model (e.g., MEGAN), or is it caused by inadequate parameterization of the vegetation's response to meteorology (e.g., drought stress) for specific Asian land cover types? I suggest providing a more in-depth mechanistic explanation, possibly by linking the results to meteorological fields (e.g., temperature, PAR) and land cover data.
6. Section 5: The analysis of local and transboundary transport based on regional emission perturbation studies is highly appreciated, as it requires significant computational effort. However, the manuscript does not seem to address the non-linear interactions between emission changes from different regions. For instance, in the pie charts (e.g., Figs. 13 and 14), the sum of individual regional contributions theoretically shouldn't perfectly equal 100% due to chemical non-linearities. Are these interaction terms lumped into the "rest of SE Asia" categories? Please clarify how these non-linear interactions are accounted for in the fractional attributions.
Technical Corrections:
7. Please replace "country" with "country/region" where appropriate (for example, in the colorbar of Fig. 2h or similar figures and texts) to maintain geopolitical neutrality and adhere to standard scientific publication guidelines regarding disputed territories.