the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Aug 2026
Interprovincial supply chains reshape the pattern of nitrogen deposition in China
Abstract. China faces severe atmospheric nitrogen deposition, with domestic consumption acting as a hidden key driver. Domestic interprovincial supply chains can reshape deposition patterns by reallocating emission burdens across provinces, yet its role in China’s nitrogen deposition remains poorly understood. Here, we quantify how domestic consumption reshapes China’s nitrogen deposition pattern by combining multi-regional input–output analysis with GEOS-Chem simulations. We find that domestic consumption was responsible for 10.5 kg N ha⁻¹ yr⁻¹ of China’s nitrogen deposition in 2017, accounting for 67 % of the national total. Reduced nitrogen (NHₓ) and wet deposition dominate this footprint, reflecting the strong influence of agricultural NH₃ emissions and precipitation-driven removal. Among all consumption demand categories, investment accounts for the largest share, followed by household and government consumption. We find a clear spatial mismatch that the provinces with high consumption affordabilities tend to net outsource emissions, while the provinces with intensive agricultural and industrial productions tend to be with net emission inflows. Sectorally, NH₃ emissions are emitted from livestock and crop production, mainly driven by domestic demand for food and agricultural products. In contrast, NOₓ emissions mainly originate from transport, power generation, and heavy industry, driven by consumption demand for infrastructure investment, manufactured goods, transport services, and energy-intensive products. Our findings show that interprovincial supply chains reshape China’s nitrogen deposition pattern by transferring emission burdens from developed coastal provinces to inland agricultural and industrial regions. Effective mitigation therefore requires coordinated strategies that combine production-side emission controls with consumption-based responsibility sharing.
- Preprint
(3629 KB) - Metadata XML
-
Supplement
(1188 KB) - BibTeX
- EndNote
Status: open (until 31 Oct 2026)
- RC1: 'Comment on egusphere-2026-3564', Anonymous Referee #1, 03 Oct 2026 reply
-
RC2: 'Comment on egusphere-2026-3564', Anonymous Referee #2, 10 Oct 2026
reply
General comments
This manuscript combines a multi-regional input–output (MRIO) framework with GEOS-Chem to investigate the role of interprovincial supply chains in shaping nitrogen deposition in China. The study provides a useful perspective on the relationship between economic activities and atmospheric nitrogen deposition.
However, several methodological issues require further clarification and validation before the conclusions can be considered robust. I recommend that the authors address the following major comments.
Major comments
1 Insufficient description and validation of the GEOS-Chem sensitivity experiments
The description of the GEOS-Chem sensitivity experiments is insufficient for readers to fully understand and reproduce the attribution of nitrogen deposition to different final-demand categories. Table S4 provides only a brief summary of the emission settings for the control and sensitivity simulations. Since these experiments form the basis of the main quantitative conclusions, I recommend moving Table S4 to the main manuscript and providing a more detailed explanation of the experimental design.
In Figure 2, the contributions of the three final-demand categories sum to exactly 100% for both NHₓ and NOᵧ deposition. However, GEOS-Chem includes nonlinear chemical processes, meaning that the sum of individual emission perturbation responses may not equal the response to their combined perturbation. It is therefore unclear whether these percentages reflect approximately additive model responses or were normalized to sum to 100%.
The authors should clarify the calculation method and quantitatively assess the additivity of the sensitivity experiments by comparing the sum of the individual deposition responses (Control-X_INV, Control-X_GOV, and Control-X_HH) with the combined response (Control-X_DOM). The absolute and relative differences should be reported to demonstrate whether nonlinear effects can be neglected in the attribution of nitrogen deposition.
2 Justification of the GEOS-Chem spin-up period and clarification of initial conditions
The authors state that a 10-day spin-up period was used for each nested GEOS-Chem simulation. However, this duration appears relatively short for regional full-chemistry simulations. GEOS-Chem developers have generally recommended a spin-up period of at least one month for nested-grid simulations, although the required duration depends on the model domain and the chemical species of interest. Given that this study focuses on nitrogen deposition involving coupled gas-phase chemistry, aerosol thermodynamics, atmospheric transport, and deposition processes, the adequacy of a 10-day spin-up should be demonstrated. I recommend that the authors consider extending the spin-up period. Alternatively, a sensitivity test comparing results obtained with the current 10-day spin-up and a longer spin-up period would help establish whether the current configuration is sufficient.
In addition, the manuscript should clarify the initialization procedures for both the global and nested simulations. Specifically, the authors should identify the sources of the restart files used in the global simulations, describe whether and how the global simulations were spun up, and explain how the initial conditions for the nested simulations were generated. It should also be clarified whether the different emission scenarios used their own corresponding global simulations and restart files. These details are necessary to assess whether the simulated nitrogen deposition responses could be influenced by the initial chemical conditions.
3 Temporal representativeness of the four simulated months and uncertainty in annual estimates
The authors simulated January, April, July, and October as representative months of 2017 and used these simulations to estimate annual nitrogen deposition. I understand that this approach can substantially reduce the computational cost, particularly when multiple GEOS-Chem sensitivity experiments are required. However, the temporal representativeness of these four months needs to be quantitatively evaluated.
Nitrogen deposition is strongly influenced by seasonal variations in both emissions and meteorological conditions. For example, agricultural NH₃ emissions can exhibit pronounced seasonal peaks associated with fertilizer application and temperature-dependent volatilization. Consequently, selecting one month per season does not necessarily provide an unbiased estimate of annual nitrogen deposition, particularly for wet deposition and agricultural NHₓ.
The authors should first explain the specific method used to extrapolate the four monthly simulations to annual deposition totals, including any temporal weighting applied. I also recommend a quantitative assessment of the resulting uncertainty. For example, the authors could conduct a full-year simulation for the control scenario and compare the resulting annual nitrogen deposition with the estimate derived from the four representative months. The comparison should ideally include total nitrogen deposition, NHₓ and NOᵧ components, and wet and dry deposition, together with their spatial distributions. This comparison would assess the annualization bias in total deposition, although it would not by itself establish the temporal representativeness of the category-specific attribution.
4 Lack of quantitative uncertainty analysis, particularly for the MRIO-based attribution
The authors discuss several potential sources of uncertainty near the end of the manuscript. However, these uncertainties are discussed primarily in qualitative terms, without a quantitative assessment of their impacts on the main results. Although the comparison with observed nitrogen deposition in Figure S4 provides useful information about the overall performance of GEOS-Chem, model–observation agreement alone does not quantify the uncertainty in the MRIO-based attribution of emissions and deposition to individual provinces or final-demand categories.
This limitation is particularly important because the MRIO framework is central to the novelty of the study. The estimated emission responsibilities depend on sector-specific emission intensities, economic input–output relationships, and the mapping between different emission inventory sectors and MRIO economic sectors. Uncertainties in these components could propagate into the calculated contributions of investment, government consumption, and household consumption.
I recommend that the authors provide a quantitative uncertainty or sensitivity analysis for the main attribution results, with particular attention to the MRIO framework and emission inventory integration. Where supported by the available input data, uncertainty ranges should be reported for the principal quantitative findings. At a minimum, the authors should quantify the sensitivity of the major conclusions to the most influential MRIO and emission-inventory assumptions.
5 Limited information on province-level emission origins
Figure 3 shows the difference between consumption-based and production-based emissions for each province, but net values alone can mask substantial bilateral flows and remove the contribution of emissions generated within a province to satisfy its own final demand. As a result, the figure does not fully reveal the geographical composition of emissions associated with each province’s consumption.
Figure S3 provides additional information, but the aggregation into five broad regions and the omission of diagonal entries still prevent readers from distinguishing within-province, within-region, and cross-region emission sources. I therefore suggest that the authors provide a province-level decomposition of consumption-based emission origins for NH₃ and NOₓ, showing how much of each province’s final-demand-driven emissions occurs locally and how much occurs in other individual provinces. Both absolute amounts and relative shares would be useful, and a complete 31 × 31 bilateral matrix could be provided in the Supplement.
In addition, the manuscript should clearly state that Figures 3 and S3 describe MRIO-based embodied emission transfers, rather than atmospheric transport or deposition source–receptor relationships. This clarification, together with a more detailed province-level analysis, would strengthen the interpretation of the results and the manuscript’s central argument.
Citation: https://doi.org/10.5194/egusphere-2026-3564-RC2 -
RC3: 'Comment on egusphere-2026-3564', Anonymous Referee #3, 16 Oct 2026
reply
This manuscript quantify how domestic consumption reshapes China’s nitrogen deposition pattern by combining multi-regional input–output (MRIO) analysis with GEOS-Chem simulations. Overall, it is an interesting paper. It is well written and structured, and therefore easy to follow and understand. I would like to recommend its publication after an adequate major revision.Major Comments:1. The authors should provide a more comprehensive discussion of the limitations associated with the control and sensitivity simulations, particularly in relation to their integration with the MRIO framework. Within an interconnected input–output economic system, removing a particular category of final demand may lead to adjustments in production activities and intersectoral economic relationships, potentially requiring the remaining economic system to be rebalanced. The current approach appears to assume that other components of the economic system remain unchanged. The authors should clarify this assumption and discuss how the absence of economic rebalancing may affect the interpretation of the sensitivity simulation results and the estimated contributions of different consumption categories.2. In Section 3.3, the authors identify whether individual provinces are net importers or exporters of embodied NH3 and NOx emissions. However, the current analysis does not adequately illustrate the directions and magnitudes of interprovincial emission transfers. Figure 3 presents the net emission balances of individual provinces but does not reveal the underlying bilateral transfer relationships. I suggest that the authors further visualize these interprovincial emission flows, either by adding directional flow lines to Figure 3 or by introducing a chord diagram to illustrate the relationships between emission-exporting and emission-importing provinces.3. The sectoral contribution analysis presented in Figure 4 would benefit from a more detailed explanation of the differences between NH3 and NOx emissions. It would be interesting to examine how emissions originating from different production sectors are redistributed across consumption sectors through upstream and downstream economic linkages.4. The authors mainly acknowledge that the static MRIO framework cannot capture dynamic adjustments in supply chains in response to consumption interventions or price signals. However, the uncertainties inherent in the MRIO database itself are not adequately addressed. MRIO tables are subject to uncertainties arising from data collection, sectoral aggregation, interregional trade estimates, and the estimation of technical coefficients, all of which may propagate through consumption-based emission accounting and affect the nitrogen deposition estimates. The authors could expand the uncertainty discussion to explicitly acknowledge these data-related uncertainties and their potential implications for the results. Previous studies on uncertainties in environmental footprint accounting based on MRIO models could provide useful references for strengthening this discussion.Minor comments:1. The authors should ensure consistency in the measurement units used throughout the manuscript. For example, Figure 2 uses ‘Tg N’, whereas Figure 3 is labelled simply as ‘Mt’. It is unclear whether ‘M’ refers to ‘Mt N’ or the mass of the corresponding pollutants (e.g., NH3 and NOx). The authors should clarify the units in Figure 3 and ensure consistent and accurate unit notation across all figures and tables to improve the clarity and comparability of the results.2. The authors should ensure consistent font styles and subscript formatting for pollutant names throughout the manuscript. Currently, inconsistencies are observed in the notation of pollutants such as NHx and NOx, particularly in the font size and formatting of the subscript ‘x’.Citation: https://doi.org/
10.5194/egusphere-2026-3564-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 193 | 75 | 34 | 302 | 75 | 39 | 34 |
- HTML: 193
- PDF: 75
- XML: 34
- Total: 302
- Supplement: 75
- BibTeX: 39
- EndNote: 34
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments:
Overall, the manuscript fits well Atmospheric Chemistry and Physics (ACP) in principle: it uses GEOS-Chem to simulate NH₃/NOₓ → NHₓ/NOᵧ and wet/dry deposition, links emissions to atmospheric transport and deposition, and draws general implications for China’s nitrogen (N) cycle. ACP favors studies with atmospheric-modelling and deposition-process content rather than pure economic accounting, so the paper should sharpen the “chemistry–deposition” contribution and not read like an input–output economics paper with a model appendix.
The MRIO+CTM attribution is novel; N deposition by NHₓ vs NOᵧ and wet vs dry speaks directly to atmospheric reactive-N research; interprovincial source–receptor mismatch is of broad interest. To pass ACP standards, the paper must (a) demonstrate rigorous GEOS-Chem evaluation by species/form/pathway, (b) discuss atmospheric processes (NH₃ volatility, NOₓ–ozone–aerosol chemistry, transport, deposition velocities, meteorology/topography), and (c) avoid presenting investment/household/government categories without standard economic definitions.
Main innovations of the manuscript include:
Specific comments (Recommended revisions):
a) Define the three final-demand categories rigorously and rename the umbrella term.
In national accounts, “consumption” normally means household + government final consumption; “investment” is gross capital formation. The text currently uses “domestic consumption” to include investment, government and household, which will confuse both economists and atmospheric reviewers. In Methods/Intro, explicitly state that Y is split into (i) household final consumption (resident households, food, energy, goods, services, residential), (ii) government final consumption (public administration, education, health, defence, etc.), and (iii) investment/gross fixed capital formation (construction, infrastructure, equipment, inventories if included). Map these to MRIO Y columns and, if possible, to CEADs/SAIC categories. This directly prevents “investment = consumption” ambiguity.
b) Update the literature and position against 2024–2026 N-deposition evidence.
The reference list is heavily pre-2024. Add recent China-focused work: Liu & Liu et al. 2026 Nature Reviews Earth & Environment on 1980–2020 drivers and the post-2010 shift back to NHₓ/agricultural-ammonia dominance; Wen et al. 2024 review on geographical distribution and policy response (oxidized N declining, reduced N stable); Ge et al. MICS-Asia III ACP multimodel N deposition evaluation; 2025–2026 regional studies on inorganic/wet–dry deposition; and 2026 work showing GEOS-Chem/WRF-Chem underestimation of particulate dry N in China, which is directly relevant to your dry-deposition evaluation. Also cite newer MEIC/GAINS/NNDMN updates if available. This addresses updated references and helps the discussion say whether 2017 NHₓ-dominance is consistent with post-2020 trends.
c) Justify the 2017 baseline and add a post-COVID/outlook section for the three categories.
2017 is reasonable for MRIO completeness, but nearly a decade old. Add a discussion paragraph (or supplementary scenario) on how the attribution would likely change in 2020–2025: post-COVID construction/restimulus may sustain investment-driven NOₓ/NH₃ initially, but real-estate downturn, green-infrastructure shift, electric vehicles (EV) penetration and power-sector decarbonization would reduce NOₓ per unit investment; household rebound in meat/dairy/e-commerce/services affects agricultural NH₃; government spending on transport, healthcare, education and “trade-in”/subsidy programmes changes sector mix. Macro context is useful here: 2021–2024 domestic demand contributed substantially to growth, with final consumption averaging ~56% and investment ~30% in some NDRC summaries, while 2025 analyses note consumption-stimulus dependence and slowing infrastructure/manufacturing/real-estate investment. If CEADs 2020/2022 MRIO and 2020s MEIC are available, a short rerun or sensitivity would be ideal; if not, frame 2017 as a structural baseline and clearly caveat time-transferability.
d) Strengthen the atmospheric-modelling rigour for ACP.
Four representative months (Jan/Apr/Jul/Oct) with annualization can bias seasonality, especially for agricultural NH₃, monsoon wet deposition and winter heating NOₓ. Expand evaluation beyond total deposition: show NMB/rms/seasonal metrics for NHₓ vs NOᵧ, wet vs dry, and compare against NNDMN/regional networks; discuss sensitivity to NH₃ spatial allocation, boundary conditions, deposition velocities and particulate-N dry flux in light of recent GEOS-Chem dry-N bias studies. Also clarify the X_DOM/X_INV/X_GOV/X_HH perturbation design: whether removed emissions are gridded with the same MEIC spatial proxies, how double-counting across categories is avoided, and how MRIO sector—grid mapping uncertainty propagates to deposition. This makes the paper “atmospheric enough” for ACP rather than an IO-accounting add-on.
If these four are addressed—especially clearer final-demand definitions, newer references, explicit post-2017 outlook, and species/resolution-resolved model evaluation—I will strongly recommend to accept the manuscript to be published in the ACP journal.