the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Nov 2025
Surface PM2.5 Air Pollution in 2022 India: Emission Updates, WRF-Chem Model Evaluation, and Source Attribution
Abstract. India experiences some of the highest PM2.5 concentrations globally. Understanding the spatiotemporal variations of PM2.5 and its source attribution requires robust air quality modeling supported by up-to-date emission inventories. Here we present the first WRF-Chem model evaluation and source attribution analysis for India for 2022, supported by updates in sectoral emission inventories and model schemes. We incorporate an updated residential emission inventory reflecting recent transitions to cleaner fuels in Indian households and develop a plant-level inventory for Indian coal-fired power plants. Further major improvements include model updates to the secondary organic aerosol scheme and an improved representation of near-surface pollutant mixing. Collectively our improvements result in a simulation with annual PM2.5 bias of only 0.2±16.9 μg/m3 (0 ± 31 %) across 288 surface monitoring sites in South Asia. We find that, compared to earlier studies, in 2022 India's residential sector remained the dominant source of PM2.5 in the Indo-Gangetic Plain, but nationally ranked second in population-weighted (PW) mean PM2.5 concentrations contributing 15 % (7.3 μg/m3). Instead, industrial emissions emerged as the largest domestic contributor to national PW mean PM2.5 (18 %, 8.6 μg/m3), with urban hotspots including Delhi and Mumbai. The power sector contributions ranked third nationally (13 %, 6.1 μg/m3) and was particularly influential in central India. Transboundary transport contributed more than any individual domestic source nationally (27 %, 12.8 μg/m3). These findings highlight the benefits of India's partial residential sector transition toward cleaner fuels, while underscoring the growing consequence of industrial and power sector emissions that have limited pollution controls.
- Preprint
(8464 KB) - Metadata XML
-
Supplement
(8076 KB) - BibTeX
- EndNote
Status: open (until 05 Jan 2026)
- RC1: 'Comment on egusphere-2025-4947', Anonymous Referee #1, 12 Dec 2025 reply
-
RC2: 'Comment on egusphere-2025-4947', Anonymous Referee #2, 16 Dec 2025
reply
This study presents a comprehensive regional air quality modeling study for India in 2022 using WRF-Chem, incorporating updated sectoral emissions (notably residential fuel transition and a new plant-level coal power inventory), selected model physics/chemistry improvements, and extensive evaluation against surface PM2.5 and satellite AOD observations. The finding that industry has become the largest domestic contributor to population-weighted PM2.5, is potentially impactful and very timely for understanding the changes in India’s PM source structure, epsecially post COVID. I think the paper is suitable for publication after several clarifications.
Specific comments:
- The authors approximate annual PM2.5 using simulations from January, April, July, and October. While this approach is common in resource-limited modeling studies, it may not be suitable for India. The seasons of India is typically characterized as pre-monsoon, monsoon and winter seasons, rather than the four seasons that can be roughtly represented by the four months. Each season also has unique pollution characteristics. I suggest the authors re-organize their simulation and discussion according to the Indian seasons. This may better reveal the contribution of different sectors to annual PM concentration.
- The study used a 100% emission off strategy to estimate the contributions. This is a quite large perturbation, which may result in unrealistic responses of some nonlinear processes. For example, secondary aerosols respond nonlinearly to precursor removal. Some discussion of these potential limitations is needed, and the attribution results should be interpreted as “effective contributions under 100% removal”, and if possible, I recommend the authors add a limited sensitivity test (e.g., 20–30% reduction for one sector) to demonstrate the magnitude of nonlinearity.
- The authors stated that transboundary sources contribute 27% of national PW mean PM2.5. This seems to be a striking finding, so careful clarification is needed. In particular, the authors did not separate natural vs. anthropogenic transboundary sources, and the quantitative attribution may depend on the domain size and boundary conditions chosen. I think at least a separation between dust and anthropgoneic sources should be performed.
- An uncertainty range (or at least rough estimation) of each attribution number should be provided, especially concerning the uncertainty in the emission inventories.
Citation: https://doi.org/10.5194/egusphere-2025-4947-RC2 -
CC1: 'Potential significant under-estimation of residential emissions', R Subramanian, 18 Dec 2025
reply
The two anonymous reviewers have already raised some of my concerns, so I will focus on the new residential emissions inventory from Velamuri et al. (2024). As the authors of the current study state:
"However, recent research highlights challenges in sustaining LPG usage under PMUY, including high refill costs and subsidy delays (Asharaf and Tol, 2024; Gaikwad et al., 2025), which may result in backsliding to a continued reliance on solid fuels which may not be fully captured in the updated inventory."
This can be significant, leading to an overestimate of biofuel replacement by as much as half - based on studies that show LPG refills are consumed at half the rate (or even less) in PMUY households compared to non-PMUY households, e.g. https://india.mongabay.com/2023/09/lpg-subsidy-can-increase-uptake-but-interventions-needed-to-counter-myths-and-improve-access/
It would also be helpful to compare the new residential EI to SMoG-2015 and the newer COALESCE version. EDGAR has much lower PM2.5 and NOx numbers for this sector compared to SMoG, and the new sectoral EI numbers used in the current manuscript are even lower than EDGAR. The COALESCE SMoG (Venkataraman et al. 2024 https://doi.org/10.1029/2024JD040834) captures some of the PMUY transition (2017-2019) but including barriers to uptake, which seems a better approach:
"The SMoG‐IndiaCOALESCE inventory captures the large fuel transition which has taken place in residential cooking and heating from the launch of the Pradhan Mantri Ujwala Yojana (PMUY, 2016) for the substitution of biomass fuels with liquefied petroleum gas (LPG), albeit with barriers which lead to significant levels of continued fuel stacking with biomass."
So at the very least, I would expect an uncertainty analysis that shows the impact of this underestimated biofuel usage (e.g. increase residential emissions by 2x) - the headline finding of this study may look very different.
Citation: https://doi.org/10.5194/egusphere-2025-4947-CC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 349 | 482 | 31 | 862 | 67 | 20 | 22 |
- HTML: 349
- PDF: 482
- XML: 31
- Total: 862
- Supplement: 67
- BibTeX: 20
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments:
This manuscript provides a comprehensive and timely assessment of India’s 2022 PM2.5 air pollution using an updated WRF-Chem modeling framework and substantially improved emission inventories, especially for the residential and power sectors. The integration of a plant-level emissions dataset, updated residential fuel-use information, and modified near-surface mixing represent meaningful advances beyond previous national-scale source attribution studies. The authors also perform rigorous model evaluation using 288 surface sites and satellite AOD, providing a relatively robust validation for India. The study contributes useful insights, particularly the finding that industrial emissions surpassed residential sources as the largest domestic contributor to national population-weighted PM2.5 in 2022, while transboundary pollution remains the largest overall contributor. These findings are policy-relevant and address an existing gap in the literature.
However, several aspects require clarification or strengthening before publication. These include (1) uncertainties associated with the updated inventories, (2) treatment of nonlinear chemical responses in the source-removal experiments, (3) reconciliation between PM2.5 and AOD biases, and (4) clearer articulation of limitations, especially regarding coarse PM and chlorine-containing species, and (5) a more event-focused, daily-resolution evaluation to verify model skill during rapid changes and extreme episodes. Addressing these points will improve interpretability and robustness.
Overall, the manuscript is clearly written, logically structured. Subject to satisfactory revision, it could be considered for publication.
Specific comments:
However, several major issues need to be addressed before the manuscript can be considered for publication.
1. Clarification and Quantification of Emission Inventory Uncertainties
The manuscript incorporates substantial updates to residential and power sector emissions (Lines160-270), but the associated uncertainty ranges are not quantified. The authors should provide uncertainty bounds for the residential fuel-use regression, the coal composition–based emission factor derivation, and the plant-level coal consumption estimation, or alternatively include a table that summarizes the main sources of uncertainty and their likely impacts on simulated PM2.5 concentrations.
2. Source Attribution and Non-linearity (Section 2.4)
The authors use a “zero-out” method combined with a scaling factor (Equation 7) to force the sum of contributions to match the baseline concentration. While this is a common approach to handle non-linearity in chemistry, it can introduce biases. For species in highly non-linear regimes (e.g., nitrate and ammonium, as shown in Figure 10), it is not obvious that linearly scaling the “difference” accurately represents each source’s contribution.
3. The “Simple SOA” Scheme (Section 2.1.2)
The manuscript implements the GEOS-Chem “simple SOA” scheme into WRF-Chem. This scheme uses fixed yields and is computationally efficient, and was originally designed for global, coarser-resolution models. The authors should comment on whether this scheme is sufficiently robust for a regional model at 27 km resolution, particularly in capturing the diurnal variability of SOA in urban hotspots such as Delhi.
4. Interpretation of Biogenic Contributions (Section 3.3)
The results show a net negative contribution of biogenic emissions to PM2.5 due to oxidant depletion (consumption of OH/HO₂) that reduces secondary inorganic aerosol formation. While chemically plausible, this might be confusing for some readers, who could misinterpret a “negative contribution” as implying that biogenic emissions improve air quality. The authors should clarify in the text that biogenic emissions lower secondary inorganic PM2.5 by constraining oxidants, but still contribute to organic aerosol loadings, as indicated in Figure 9 where biogenic sources contribute to the organic component, albeit modestly.
5. AOD-PM2.5 Discrepancy
The model strongly underestimates AOD (−29±14%, Lines 510-524) despite achieving reasonably good agreement for surface PM2.5. The authors should clarify whether this discrepancy is mainly due to missing coarse PM sources (as suggested on Section 2.2), and whether assumptions related to hygroscopic growth or aerosol optical properties (e.g., the internal-mixing treatment in MOSAIC) may also contribute to the bias.
6. Coarse PMcoarse Underestimation
The manuscript notes that PMcoarse is strongly underestimated compared with CPCB observations (Line 522), but this is not quantified. The authors should explicitly report the magnitude of the PMcoarse bias and to discuss potential missing or underestimated sources, such as road dust beyond the adjustments already made for transportation, construction dust, and industrial fugitive emissions.
7. Extreme-Event Underestimation and Time-Varying Emissions
The model underestimates both surface PM2.5 and, to an even greater extent, AOD during extreme events. Fires from Myanmar, internal crop burning from within the domain, and severe air pollution events from small industries located outside of city centers are not captured, especially during their most intense phase. The authors should clarify whether this is related to a time-changing emissions dataset that is not currently considered.
8. Daily-Resolution Evaluation
Since the model has daily-resolution data, comparisons should be made with daily data. It is essential to do this, especially during times when aerosol emissions and removal change rapidly (e.g., biomass burning and monsoon arrival).
9. Coarse-Mode/Mixing Discussion: Alternative Possibility
The authors should add to the coarse-mode argument and the mixing argument that another possibility is that the particle sizes are still fine, but have multiple peaks (i.e., not reasonably represented by a single-peaked lognormal, as assumed). The authors should also note that absorption and extinction enhancement due to mixing may contribute to the discrepancy.