The study, led by Qian Li and colleagues at the Beijing Municipal Ecological and Environmental Monitoring Center together with Chunsheng Zhao of Peking University, examined surface ozone measurements from 23 national monitoring stations spread across urban and suburban Beijing between 2013 and 2023. The team introduced a concept borrowed from North American air quality practice, the Ozone Season, defined not by a fixed calendar but by the continuous period from the first to the last day in a year when the maximum daily 8-hour moving average ozone concentration, known as MDA8, exceeds 160 micrograms per cubic meter, China’s official threshold for an ozone pollution day. By tying the season to the photochemical lifecycle of ozone itself rather than to April through September, the definition naturally captures the regional and seasonal heterogeneity of ozone pollution that a fixed warm-season framework misses.

What the record shows is a season under pressure from both ends. The first ozone pollution day of the year advanced from 5 May in 2013 to 30 March in 2023, an average shift of roughly 3.5 days per year, while the last pollution day slipped from 11 September to 27 September, a delay of 16 days over the same span. Together, those shifts extended the Ozone Season by an average of 5.0 days per year, about 3.2 percent of the average season length of roughly 155 days. The season ranged from 116 to 176 days across the decade, with 2021 the shortest and 2022 the longest. April hosted the first pollution day in seven of the eleven years, and the Mann-Kendall trend statistics for the first pollution day and season length, after two meteorologically anomalous years were excluded, climbed to 1.80 and 1.89, brushing against the 1.96 critical value for statistical significance at the 95 percent confidence level.

That near-significance matters, because ozone photochemistry is exquisitely sensitive to weather. Temperature, solar radiation, and synoptic-scale circulation inject substantial year-to-year noise into an eleven-year record, and the authors identified 2017 and 2021 as anomalous years with markedly shorter seasons, likely attributable to such meteorological quirks. Once that noise was filtered out, the underlying tendency toward an earlier onset and a longer season emerged more clearly, even though the timing of the last pollution day, which fell in September in 88 percent of years, retained a weaker trend signal. The researchers attribute the lengthening of the Ozone Season to a combination of climate change and shifting pollutant emission patterns, and they argue that a longer high-risk period demands adaptive pollution control strategies rather than the static seasonal frameworks now in use.

The concentration statistics tell an equally counterintuitive story. Low-percentile ozone levels rose sharply and significantly: the minimum MDA8 climbed from 4 micrograms per cubic meter in 2013 to 14 in 2023, a 225 percent increase, while the 25th and 50th percentile values rose by 24 and 31 micrograms per cubic meter, at rates of 2.4 and 3.1 micrograms per cubic meter per year. The minimum and 25th percentile trends passed the Mann-Kendall test with confidence. Yet the mean and high-percentile values barely moved. The maximum MDA8 recorded was 311 micrograms per cubic meter in 2014, the lowest was 256 in 2021, and the 90th percentile actually declined at about 2.5 micrograms per cubic meter per year. In other words, Beijing’s ozone floor is rising while its ceiling holds steady, a signature consistent with global findings that low-percentile ozone has increased everywhere while high-percentile behavior diverges by region.

The daily timing of exposure has shifted too, and this is where the findings turn directly toward public health practice. The team defined the ozone exposure window as the 8-hour period containing the MDA8, and tracked when that window begins. In the cold season, the starting hour moved from predominantly 10:00 to 11:00 in 2013 to 12:00 to 13:00 by 2023, a more pronounced delay than in the warm season, where the window shifted from 11:00 to 12:00 toward 12:00 to 13:00. During April through September, the exposure window now spans roughly 14:00 to 21:00, meaning residents are breathing elevated ozone well into the evening and nighttime hours, a period when traditional daytime monitoring and advisories offer little protection. On pollution days specifically, the exposure window is tightly concentrated between 13:00 and 14:00 at its start, reflecting the dominance of intense photochemical production under favorable meteorology, whereas non-pollution days show a much broader and more delayed distribution shaped by weaker photochemical forcing and non-local transport.

The flattening of the diurnal cycle is perhaps the study’s most striking technical result. Each year’s typical diurnal ozone profile, after screening out days with rain, strong winds, or overcast skies, showed the classic unimodal shape: concentrations fall from midnight to a minimum between 06:00 and 07:00, rise to a peak between 16:00 and 18:00, then decline. But over the decade, the peaks dropped at 3.6 micrograms per cubic meter per year while the troughs rose at 1.7, and the hourly trend analysis found significant declines during the 14:00 to 19:00 afternoon window and significant increases during the 01:00 to 10:00 overnight and morning window, both at p below 0.05. The diurnal amplitude, the swing from trough to peak, shrank from about 150 micrograms per cubic meter in 2013 to 104 in 2023. The chemistry behind this is a two-pollutant story: falling nitrogen oxides have reduced nighttime nitric oxide concentrations, weakening the titration reaction that normally destroys ozone after dark, while stricter volatile organic compound controls since 2017 have suppressed daytime ozone production, pulling the peaks down.

The team confirmed that interpretation by computing the hourly ozone change rate, the difference in concentration between consecutive hours, which integrates all local production and loss processes after long-term averaging suppresses advective noise. The maximum production rate rose slightly to 29.1 micrograms per cubic meter per hour in 2015, then fell steadily to 18.2 by 2023, a decline of about 1.0 per year, with its timing advancing from 11:00 to 13:00 toward 10:00 to 12:00. The maximum loss rate, typically occurring between 20:00 and 22:00, weakened at 1.2 micrograms per cubic meter per hour per year, and the average production speed across the 9-to-11-hour production period declined at roughly 0.69 per year. Both the engine and the brake of Beijing’s ozone photochemistry are slowing, and the brake is slowing faster, which is precisely why the diurnal profile flattens and the overnight baseline climbs.

Notably, pollution days retain more production capacity than the season as a whole. Peak concentrations on pollution days declined at only 2.5 micrograms per cubic meter per year, 30.9 percent slower than the seasonal rate, while troughs rose 37.3 percent faster, and the hourly gap between pollution days and the Ozone Season widened over the decade, with nocturnal differences growing from 19.2 to 34.2 micrograms per cubic meter. Analysis of the MEIC emission inventory showed that nitrogen oxide and VOC emissions fell in both seasons, ambient nitrogen dioxide tracked emissions with correlations of 0.95 to 0.97, yet warm-season MDA8 ozone showed no significant correlation with either VOC emissions or the VOC-to-NOx ratio, a decoupling that underscores how nonlinear ozone photochemistry and meteorological variability dominate over precursor reductions. The authors are candid that their observational approach cannot quantitatively separate emission-driven from climate-driven changes, and that chemical transport modeling will be needed to disentangle the contributions.

What emerges is a baseline for how urban ozone evolves under sustained emission controls, and a warning about what comes next. Beijing’s mean MDA8 rose at 2.4 micrograms per cubic meter per year over the decade, its Ozone Season lengthens by five days annually, its overnight ozone climbs as titration weakens, and its pollution days stubbornly retain production capacity even as typical days lose it. For a city of 22 million people, the implications reach beyond policy into daily life: health protection for sensitive populations may need to extend into evenings and shoulder seasons that current frameworks ignore, and the two-pollutant strategy of coordinating VOC and NOx reductions, which the data suggest is already bending the peaks downward, may be the most promising lever for a pollutant that has proved far harder to tame than the soot and sulfur Beijing conquered in its miracle decade.

Subject of Research: Decadal evolution of surface ozone and the Ozone Season in Beijing under emission controls, 2013-2023

Article Title: Shifting patterns of Ozone Season due to policy and human activity in Beijing between 2013 and 2023

Article References: Shifting patterns of Ozone Season due to policy and human activity in Beijing between 2013 and 2023. (n.d.). https://doi.org/10.5194/acp-26-14559-2026

Image Credits: AI Generated

DOI: 10.5194/acp-26-14559-2026

Keywords: Shifting, patterns, Ozone, Season, policy, human, activity, Beijing, between, scientific research