the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Aug 2026
Regime-dependent expression of regional aerosol signals in urban PM10 and PM2.5 across arid northern China
Abstract. Air pollution in arid regions emerges from the interaction between regional aerosol transport and local atmospheric processes, yet it remains unclear whether inter-city synchronization reflects direct transport or shared meteorological forcing. This ambiguity limits the interpretation of urban air quality and the design of effective control strategies in dust-influenced environments. Here, we analyze PM10 and PM2.5 observations across 47 cities in northern China during 2015–2025. We quantify regional coupling using explained variance (R2) and introduce a differential regional control metric (ΔR2) to resolve the relative sensitivity of coarse and fine particles to regional forcing. We find that strong inter-city synchronization coexists with substantial heterogeneity in underlying processes. Although the regional signal explains comparable variability in PM10 and PM2.5, ΔR2 varies widely across cities (−0.57 to +0.34), revealing distinct pollution regimes. These regimes are systematically linked to aerosol composition and proximity to desert sources, indicating that similar levels of synchronization can arise from fundamentally different mechanisms. Event-scale analysis further shows a nonlinear dependence of aerosol concentrations on wind speed, with extreme PM10 occurring under both weak and strong winds. This reflects two contrasting regimes: accumulation under limited ventilation and transport-driven dust events. Together, these results demonstrate that regional synchronization is not a unique indicator of transport but an emergent property of multiple interacting processes. The proposed framework provides a process-oriented basis for distinguishing these regimes and improving the interpretation and management of air quality in arid regions.
- Preprint
(1352 KB) - Metadata XML
-
Supplement
(823 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3468', Anonymous Referee #3, 15 Oct 2026
-
RC2: 'Comment on egusphere-2026-3468', Anonymous Referee #4, 24 Oct 2026
This manuscript investigates regional synchronization of PM10 and PM2.5 concentrations across 47 cities in arid northern China during 2015–2025. While the long-term analysis is valuable, I have substantial concerns about the methodology and the physical interpretation of the results. In particular, the analysis relies primarily on statistical correlations, but the authors use these relationships to infer the contributions of regional and local processes and identify distinct pollution regimes without sufficient supporting evidence. The use of monthly mean data also raises concerns about the ability to capture short-term dust events and atmospheric transport processes. In addition, the manuscript requires substantial improvements in clarity. I believe that major revisions and additional analyses are needed before the main conclusions can be adequately supported.
Major Comments:
- The Introduction mixes scientific questions, gaps in previous research, and new concepts without a clear logical flow, making it difficult to follow. For example, it raises two related but distinct research questions (Line 59–61 and 73–78), followed by a discussion of comparing PM10 and PM2.5 (starting at Line 79), without clearly explaining how this comparison addresses the questions raised earlier. I suggest reorganizing the Introduction to better connect the research questions, knowledge gaps, and proposed approach.
- Why did the authors choose to use monthly mean data rather than daily or hourly observations? Please clarify the rationale. Dust storms, synoptic-scale weather systems, and aerosol transport typically occur on timescales of hours to days, which may not be adequately captured by monthly averages. In addition, Section 2.10 and 3.4 defines then discusses “extreme events” using monthly mean, respectively, how can monthly mean well represent “extreme events”? Please clarify.
- I have concerns about interpreting R2 as the contribution of regional processes to urban PM variability (Line 179-181). The proposed regional signal is simply the average PM concentration across other cities, and R2 quantifies the squared Pearson correlation between this signal and the concentration at a given city. Such statistical relationship does not establish causality or distinguish the contributions of regional transport, and shared meteorological conditions. In addition, what does the unexplained fraction of variance (1-R2)? Does that indicate local contributions? It is therefore unclear how to intepret this R2 in terms of the underlying physical/chemical processes.
- Following the previous comment, in Line 306–314, the authors seem to interpret stronger correlations with the regional PM10 signal as evidence of greater mineral dust influence, and stronger correlations with the regional PM2.5 signal as evidence of greater contributions from secondary aerosol formation and anthropogenic sources. These explanations are not easy to follow for readers. Please clarify the physical basis for these interpretations.
- Section 2.7 and table 1: the criteria used to classify cities into four categories are not mutually exclusive. For example, a city could simultaneously meet the criteria for “strong regional control” and “PM10-dominated regional control”. Please clarify how such overlapping cases were handled, or revise the criteria to ensure that the four categories are mutually exclusive.
- Current analysis considers only wind speed but not wind direction. Wind direction is important for identifying potential dust transport pathways and distinguishing regional dust transport from local dust emissions. Without this information, it is unclear how the authors can attribute extreme PM10 concentrations under high-wind conditions to transport-driven dust events. Please clarify. In addition, meteorological conditions favorable for local pollutant accumulation cannot be characterized by wind speed alone. Indicators of vertical ventilation, such as planetary boundary layer height (Wang et al., 2018), temperature inversions (Zhou et al., 2024), or more complex boundary layer features (Huang et al., 2018), should also be considered.
- Figs 2–5 lack detailed captions, containing only brief titles without explanations of the individual panels. In addition, the figures appear blurry when zoomed in. Please provide more informative captions and higher resolution figures. In addition, in Fig 2c is difficult to interpret. Some bars appear to be stacked, while other show only one color. Also, what is the bin width, and what range of correlation coefficients does each bar represent?
- Line 332, “… additional process-based analysis, which is addressed in the following sections.” However, Sections 3.2–3.4 are largely based on statistical correlations and relationships with geographic and meteorological variables. It is unclear which analyses directly investigate the underlying physical processes. Please clarify.
- Line 421-425, the authors interpret higher R2 for PM10 as indicating transport-dominated dust regimes. However, this is inconsistent with the interpretation for Haixi (Line 365-370), where a higher R2 for PM10 is attributed to “the dominance of local dust processes and limited coupling with the broader regional dynamics.” Please clarify. In addition, why does a higher R2 for PM10 necessarily indicate dust transport? If so, could higher R2 for PM2.5 also result from regional transport of fine particles? These claims appear to have limited physical basis and need more clarification.
- Line 559–564: The authors highlight the nonlinear relationship between PM10 concentrations and wind speed as a key finding. However, similar relationships, with elevated PM concentrations under both low- and high-wind conditions, have been reported in previous study (Yin et al., 2016). Please clarify the novelty of this finding.
Minor Comments:
- Line 48-51 “As a result, particulate matter concentrations in urban areas reflect not only local emissions but also the influence of large-scale processes, making both the interpretation of observations and the development of effective air quality management strategies more challenging.” Based on my understanding, any urban areas face the challenges of local emissions and transboundary transport of pollutants. What makes the focus on arid and semi-arid regions, along with its dust background special here, please clarify.
- Line 56-58, what about the impact of wind on dust emissions?
- Line 159, please clarify how the seasonal components were removed and whether this procedure was applied only to the standardization and anomaly analysis or to all other analyses.
- Line 251-258, I don’t see how amplified IQR during March 2021 can support the claim that “extreme concentration levels are primarily driven by regional-scale processes, while local conditions modulate the magnitude of the response in individual cities”. The increased IQR indicates greater spatial variability in standardized concentration anomalies but does not directly establish the contributions of regional and local processes. Please clarify.
- Line 272-278, can you clarify how the lower inter-city variability in correlations with the regional signal for PM2.5 reflects differences in aerosol sources and formation processes between PM2.5 and PM10?
- Line 287-290, this sentence seems to imply that, given Haixi’s arid environment and proximity to potential dust sources, a high correlation with the regional signal for PM2.5 and a low correlation for PM10 would be expected. Please clarify the reasoning behind this expectation. In addition, the following sentence states that “This indicates the dominance of local dust processes and limited coupling with the broader regional dynamics.” If so, why does Haixi exhibit a relatively stronger correlation with the regional PM10 signal?
- Line 293, Zhu et al., 2026 is missing in the reference list.
- Line 316-319, this is very general and it is unclear how this conclusion is supported by the preceding analysis. Please clarify.
- Line 326–328: “superposition of multiple partially independent modes of variability” is vague. What specific processes or modes of variability are the authors referring to, and how do they differ from the common regional forcing mentioned earlier? Please clarify.
- Line 454-455, this is inconsistent with Fig 5c, where low-wind conditions shows overall lower PM10 concentrations compared with two other wind regimes.
- Line 498–506: This paragraph largely repeats the interpretation of ΔR2 already presented in Line 419–427, with little additional discussion. Please avoid unnecessary repetition between the Results and Discussion sections.
- Since the authors find that extreme PM10 concentrations occur under both low- and high-wind conditions, it would be interesting to examine whether similar relationships exist for PM2.5.
Reference:
Huang, Q., Cai, X., Wang, J., Song, Y., and Zhu, T.: Climatological study of the Boundary-layer air Stagnation Index for China and its relationship with air pollution, Atmospheric Chemistry and Physics, 18, 7573-7593, 10.5194/acp-18-7573-2018, 2018.
Wang, X., Dickinson, R. E., Su, L., Zhou, C., and Wang, K.: PM2.5 Pollution in China and How It Has Been Exacerbated by Terrain and Meteorological Conditions, B Am Meteorol Soc, 99, 105-119, 10.1175/bams-d-16-0301.1, 2018.
Zhou, M., Xie, Y., Wang, C., Shen, L., and Mauzerall, D. L.: Impacts of current and climate induced changes in atmospheric stagnation on Indian surface PM(2.5) pollution, Nat Commun, 15, 7448, 10.1038/s41467-024-51462-y, 2024.
Yin, Q., Wang, J., Hu, M., and Wong, H.: Estimation of daily PM(2.5) concentration and its relationship with meteorological conditions in Beijing, J Environ Sci (China), 48, 161-168, 10.1016/j.jes.2016.03.024, 2016.
Citation: https://doi.org/10.5194/egusphere-2026-3468-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 144 | 76 | 29 | 249 | 76 | 31 | 30 |
- HTML: 144
- PDF: 76
- XML: 29
- Total: 249
- Supplement: 76
- BibTeX: 31
- EndNote: 30
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Filonchyk et al. investigates regional synchronization of PM10 and PM2.5 across 47 cities in arid northern China using a leave-one-out regional signal and R² and ΔR² metrics. The long-term, multi-city dataset and the comparison between PM10 and PM2.5 provide a useful basis for examining spatial differences in aerosol behavior. I found the distinction between regional synchronization and direct aerosol transport particularly important, as similar temporal patterns across cities may arise from different underlying processes.
The manuscript addresses an important question and presents an interesting approach. However, I have several concerns regarding how some of the statistical results are interpreted. In particular, the interpretation of R² in terms of regional and local contributions needs further justification, the construction of the regional signal should be better supported, and some of the proposed physical mechanisms appear stronger than what can be directly established from the analyses presented. I therefore recommend minor revision before the manuscript is considered for publication.
Major comments:
1.The interpretation of R² as a quantitative measure of regional and local contributions requires further justification. In the current framework, R² = r² measures how strongly the temporal variation in each city's PM concentrations is associated with the constructed regional signal. While this provides a useful measure of regional coherence, it does not necessarily mean that the explained variance represents a regional contribution or that the remaining variance represents a local contribution. I therefore suggest that the authors clarify this distinction throughout the manuscript and reconsider statements that the framework "separates regional and local contributions". If this interpretation is retained, additional justification or supporting analysis would be needed.
2. The construction of the regional signal requires further justification. The leave-one-out approach appropriately avoids including the target city in its own regional reference. However, for each target city, the regional signal is calculated as an equally weighted average of all other cities, despite the considerable geographic and environmental differences across the study region. It is therefore unclear whether all cities are equally representative of the regional conditions affecting a given target city. This is important because R², ΔR², and the resulting regime classifications all depend on how the regional signal is defined. I suggest that the authors provide a clearer justification for this choice and evaluate whether the main results are robust to alternative definitions of the regional signal, for example using geographically closer cities or subregional groupings.
3. The interpretation of the statistical relationships in terms of specific physical processes could be presented more cautiously. The observed relationships between ΔR², the PM2.5/PM10 ratio, distance to desert sources, and wind speed are consistent with the proposed roles of dust, atmospheric accumulation, and aerosol formation. However, these relationships alone do not directly demonstrate that these processes are responsible for the observed patterns. I suggest that the authors make this distinction clearer and use more cautious wording when discussing specific mechanisms, unless additional evidence is available to support these interpretations.
Minor comments:
Table S1 reports the correlation coefficients for PM10 and PM2.5 and the difference between them, while the regime classification is based on R² and ΔR². To make the table more consistent with the methodology and easier to interpret, I suggest reporting R² and ΔR² directly in Table S1.