The observation was conducted from June 1 to 30 June 2024, at the Xuzhou Environmental Monitoring Center Station (34.215° N, 117.256° E). Xuzhou is strategically positioned at the convergence of Jiangsu, Shandong, Henan, and Anhui provinces in East China, renowned for their industrial, agricultural, and manufacturing activities (Qin et al., 2017) (Figure 1). The observation site is situated southeast of Xuzhou in a mixed commercial and residential area (Rao et al., 2023). Major roads are located 100 m to the east and 500 m to the south of the site. To the north, extensive farmland and abundant vegetation can be found. Consequently, this site offers an ideal setting for investigating the interplay between human-made and natural sources of pollution (Feng et al., 2023).

Instruments for meteorological data and pollutant concentrations were positioned on the rooftop of the Xuzhou Environmental Monitoring Center Station (34.22°N, 117.27°E). Situated approximately 20 m above ground level, the placement minimized the impact of turbulence near the ground. The specific instruments information are listed in Table 1. Thermo commercial instruments were employed to monitor NOx, O₃, CO, and SO₂ levels, while GC-MS/FID technology was utilized to detect VOCs species, encompassing 107 types of non-methane hydrocarbons (NMHCs) ranging from C₂ to C₁₄, as well as some other OVOCs. To ensure the precision of the measurements, all instruments underwent daily external standard quality control checks at midnight.

The photolysis rates were determined using the Tropospheric Ultraviolet and Visible (TUV) radiation model. To ensure accuracy, the rates were adjusted based on measurements of total solar irradiance (TSI) (Trebs et al., 2009). The validity of this approach has been confirmed through extensive long-term observations (Wang et al., 2020).

A zero-dimensional box model based on the RACM2-LIM1 mechanism was utilized to calculate the explicit local ozone formation and the sensitivity of O₃-NOx-VOCs(Griffith et al., 2013; Stockwell et al., 1997; Tan et al., 2017). To input the necessary data into the model, relevant meteorological and pollutant parameters were compiled into a time-dependent dataset for the model boundaries. In instances where there were gaps in the data due to instrument maintenance or malfunction, averaging or linear interpolation techniques were used to fill in the missing values. The time resolution used for the dataset was 15 min. For species that were not measured, such as H₂ and CH₄, default values were assigned at 550 ppb and 1900 ppb, respectively. To account for the removal of pollutants over a 24 h lifetime, corresponding removal rates were applied to all species in the model. Additionally, a 3 day spin-up period was included to allow the model to stabilize before generating results.

To evaluate the O₃-NOx-VOCs sensitivity, the response of simulated ozone concentration to changes in individual precursors was calculated, which was expressed as the Relative Incremental Reactivity (RIR) in Equation 1 (Cardelino and Chameides, 1995): R I R X = Δ O 3 X O 3 Δ C X C X (1)

∆O₃ represents the change in simulated ozone concentration, O₃ denotes the baseline ozone concentration, ∆C(X) represents the change in precursor concentration, and C(X) represents the baseline precursor concentration. For the sensitivity experiments, a baseline value of 20% was selected for ∆C(X), and then evaluating the impact on the simulated ozone concentration (Yu et al., 2020).

To predict ozone concentrations for the calculation of RIR, a constraint was applied to NO₂ while removing constraints on O₃ and NO in the box model (Tan et al., 2018; Zhang et al., 2024).

Figure 2 illustrates the timeseries of meteorological data and pollutant concentrations during the observation in June 2024 in Xuzhou. The city experienced consecutive extreme heatwaves and intense solar radiation, with daytime temperature peaks surpassing 30 °C, and j (O¹D) reaching a value of 4.0 × 10⁻⁵ s^-1. NO₂ concentrations exhibited a range of 4.73–40.51 ppb, primarily attributed to elevated precursor emissions in the urban environment. PM_2.5 and CO concentrations showed consistent patterns and occasionally reached mild pollution levels on specific days. The average daily peak concentration of VOCs was 16.69 ppb, with a maximum of 36.50 ppb recorded on June 14th. Anthropogenic VOCs (AVOCs) were predominantly composed of alkanes, alkenes, and aromatics. Oxygenated volatile organic compounds (OVOCs) included acrolein, acetone, methyl ethyl ketone, and methyl t-butyl ether, and accounted for over 50% of the total VOC concentration (TVOCs). The intensive emissions of VOCs played a significant role in local ozone formation, leading to consecutive ozone pollution events. During the observation, the daily ozone concentration met the Grade I standard for 21 days and the Grade II standard for 4 days, according to the Ambient Air Quality Standards (GB3095-2012), with daytime peaks reaching 130 ppb.

From June 9 to 16, there was a continuous period of high temperatures, with peaks surpassing 35 °C. According to the Grade of the Heatwave (GB/T 29,457–2012) classification, the scenario can be categorized as an extreme heatwave event (denoted as Extreme period in Figure 3). In contrast, the periods from June 1 to 5 and June 24 to 29 experienced temperature peaks below 30 °C, which was categorized as normal temperature events (denoted as Normal period in Figure 3). Extreme heatwave coincided with a notable increase in ozone pollution, with concentrations rising from 74.14 ppb to 105.51 ppb, representing an increase of over 30% (Figure 3). While the average daytime levels of NO₂ remained consistent across both scenarios, TVOCs significantly increased during the heatwave, climbing from 14.02 ppb to 18.47 ppb. Bio-VOCs emissions, particularly isoprene, exhibited a strong response to elevated temperatures, with peak concentrations nearly doubling compared to the Normal events, reaching as high as 2.86 ppb on certain days.

Given the elevated VOC emissions during heatwaves, the influence of extreme temperatures on the chemistry process of VOC was examined, utilizing temperature as a key variable. The relationship was evaluated through VOC concentrations, hydroxyl reactivity (k _OH), ozone formation potential (OFPs), and secondary organic aerosol formation potential (SOAFPs) to characterize the theoretical generation of secondary pollution generation in diurnal time (8:00–18:00, Figure 4). The total OH reactivity indicates the reciprocal of the OH radical’s lifetime. The OFP for each individual VOC species is determined by multiplying the measured concentration with its maximum incremental reactivity (MIR), and then aggregating the values across all VOC types. Notably, OVOCs were excluded from this analysis due to incomplete measurements.

In terms of relative contribution, alkanes consistently represented the largest share of VOC concentrations, accounting for approximately 62.50% (except for OVOCs). However, the OH reactivity for alkanes decreased significantly, with a reduction of 11.21% during Extreme events and 31.42% during normal scenarios. The increased reactivity observed during the heatwave was primarily influenced by active species, particularly biogenic and anthropogenic alkenes. Despite constituting only 5% of VOCs concentration, isoprene displayed a marked response to elevated temperatures, increasing the contribution to k _OH from 32.00% to 60.04%, which also correlated with a higher ozone formation potential. In contrast, aromatics contributed only around 5% to overall VOC reactivity, differing from findings in cities such as Guangzhou and Shanghai, which suggests a limited local use of solvents and chemical manufacturing sources (Tan et al., 2019; Yu et al., 2020). Nevertheless, aromatics displayed significant ozone formation potential and accounted for over 70% of secondary organic aerosol formation potential (SOAFP).

Given the pronounced pollution characteristics observed during heatwaves, the influnence of extreme temperatures on atmospheric oxidative capacity was examined using temperature as a variable. A box model was employed to quantify key parameters related to atmospheric oxidant sources and sinks, specifically total k _OH and P (ROx) (Figure 5). Due to diurnal variations in meteorological dilution conditions, OH reactivity was found to be lowest in the afternoon and highest during morning traffic peaks, aligning with observed temporal trends (Tan et al., 2019; Zhang et al., 2024). Higher reactivity levels (ranging from 10 to 30 s^-1) was observed in heatwaves, and the proportions of total OH consumption by inorganic species CO and NOx were 21.16% and 19.89%, respectively, indicating a significant contribution of anthropogenic emissions to radical chain termination. Organic species accounted for approximately 60% of the total reactivity, with OVOCs contributing 30.98%. The proportion of k _AVOCs remained consistent with observations during normal temperatures, while a more pronounced contribution from unmeasured species to k _OVOCs, suggesting the potential influence of unknown photochemical processes activated by elevated temperatures.

Considering that photolysis reactions are the primary drivers of radical chemistry, the budget analysis concentrated exclusively on daytime conditions (08:00–18:00). During Normal scenario, the daytime peak of P (ROx) can reach 3.71 ppb/h, which is typical of the summer and autumn seasons in China (Guo et al., 2024; Tan et al., 2024; Tan et al., 2021; Yang et al., 2021). Photolysis reactions accounted for a substantial 86.90% of the total radical production in diurnal time. Among these, the photolysis of O₃ and HONO were particularly significant in generating OH radicals, contributing approximately 29.72% and 18.58% to the total primary sources of radicals, respectively. The photolysis of HCHO and other OVOCs emerged as important primary sources of hydroperoxyl (HO₂) and peroxy radicals (RO₂), contributing 26.58% and 12.09% to P (ROx), respectively. Alkene ozonolysis also played a role in the generation of OH, HO₂, and RO₂ radicals, yielding an average production rate of 0.22 ppb/h during the daytime.

During heatwave events, the peak primary radical source increased significantly, showing a 64.69% rise compared to normal temperatures, reaching 6.11 ppb/h. Ozone photolysis emerged as the dominant contributor to P (ROx), accounting for 35.76% of the total, which is comparable to that from the photolysis of aldehydes and ketones (collectively contributed to 38.63%).

Given the specific impact of extreme heatwaves on ozone pollution, the ozone formation process was reconstructed across the entire temperature range (Figure 6). The species concentration and reactivity are both in the daytime range, when j (O¹D) is greater than 1.0 × 10⁻⁶ s^-1, thereby eliminating the daytime and nighttime effects. Ozone concentrations revealed a strong positive correlation with temperature, with levels increasing from approximately 45 ppb to around 110 ppb, indicating no apparent threshold for this trend. In contrast, PM_2.5 concentrations exhibited a different pattern, peaking between 15°C and 25 °C and subsequently declining at higher temperatures, showing less significant responsiveness.

Compared to concentrations of specific components, changes in the reactivity of VOCs, denoted as k _VOCs, provide valuable insights into the active photochemistry occurring under high-temperature conditions. Previous research highlighted the intricate interplay between anthropogenic activities and vegetation during heatwave events (Li et al., 2024). VOC reactivity was segmented into three main categories: anthropogenic, biogenic, and aldehyde/ketone oxygenated VOCs. The response of k _BVOCs to temperature fluctuations is particularly noteworthy. As temperatures surpass 30 °C, the average k _BVOCs escalates to around 2 s^-1, while the reactivity of anthropogenic VOCs (k _AVOCs) remains relatively stable, indicating a heightened level of activity in biogenic VOC chemistry under elevated temperatures. In the presence of intense photochemical conditions induced by prolonged heatwaves, VOCs interact with OH radical to generate peroxy radicals and other OVOCs, leading to an elevation in k _OVOCs from 2.5 s^-1 to approximately 6.5 s^-1 (Figure 6). The typical tracer for anthropogenic OVOCs (acetone) exhibited a distinct temperature dependency, featuring a pronounced peak during the high-temperature period, which reached a maximum of 16 ppb (Supplementary Figure S4). Consequently, heatwaves trigger an overall increase in the emissions of VOCs originating from both natural and anthropogenic sources, with more prominent contributions from bio-alkenes and OVOCs.

Based on the discussion provided, the pollution event described can be characterized as a “heatwave-ozone” co-occurring extreme event, triggered by high temperatures, low humidity meteorological conditions, and intense radiation. The heatwave accentuated the contributions to VOCs emissions, leading to heightened levels of oxidizing parameters such as radical sources, reactivity, and ozone formation potential. This scenario is likely to exacerbate the co-occurrence of heatwaves and ozone extremes. Despite ongoing reductions in anthropogenic VOC emissions with the current precursor composition (as the constant k _AVOCs in Figure 6), the impact of high temperatures significantly enhanced natural VOC emissions (as the elevated k _BVOCs in Figure 6), counteracting the benefits derived from emission reductions. Therefore, enhanced control of anthropogenic VOCs and reactive OVOCs is crucial during heatwave events.

To effectively control ozone-formation precursors, it is essential to mitigate the increase in biogenic VOC emissions triggered by high temperatures. The study utilized the RIR analysis to quantify ozone responses to individual precursors, focusing on NOx and VOCs (as shown in Figure 7). Apart from CO and NOx, the research specifically examined isoprene and anthropogenic VOCs, such as alkanes, alkenes, and aromatics. It was observed that due to a higher proportion of chain-terminating reactions, the chemistry of NOx did not yield positive benefits, as indicated by a daytime average RIRNOx of −0.39, suggesting that NOx reduction has a detrimental impact on ozone control. Conversely, reducing anthropogenic VOCs demonstrated systemic positive effects, with a corresponding daily average RIR_AVOCs of 0.71, indicating that a 1% reduction in anthropogenic VOCs results in a 0.71% decrease in ozone concentration. Among the different types of anthropogenic VOCs, controlling alkenes exhibited the most significant effect on ozone reduction, with a daytime average RIR above 0.5. Regulating aromatics, alkanes, and CO also had a positive impact on ozone control, with RIR values ranging from 0.067% to 0.18%/%.

The substantial emissions of biogenic VOCs exacerbated by extreme high temperatures underscore the significant control effect on ozone pollution, as elucidated by the quantitative results discussed in Section 4.1. A growing body of research underscores the importance of biogenic isoprene in contributing to ozone pollution in urban areas during the summer months, with global emissions estimated to be around 600 Tg C annually (Guenther et al., 2006). Li et al. determined through modeling that natural emissions enhance ozone pollution by increasing both temperature sensitivity and concentration levels. In future high emission scenarios, such as ssp370 and ssp585, this trend is expected to intensify the concurrent occurrence of “heatwave-ozone” extreme events (Li et al., 2024). From a practical control perspective, choosing plants with low alkene emissions in urban planning could serve as a potential method for managing ozone pollution. Such an approach not only addresses the reduction of biogenic VOC emissions but also enhances urban green spaces, promoting overall environmental health. Furthermore, Song et al. have developed an effective and environmentally friendly photocatalyst designed for the efficient removal of isoprene (Song et al., 2024). When the residence time is increased by 10 s, the photocatalytic rate of isoprene is improved to 79%, and its feasibility for pollution control in various regions has been demonstrated. Simultaneously, new strategies for the synergistic control of primary radical sources (P (ROx)) and reaction chain length (ChL) emerge as additional methods to reduce ozone levels (Liang et al., 2024; Lu et al., 2023). By integrating these approaches, we can develop a more comprehensive framework for ozone pollution management that considers both biogenic emissions and innovative technological solutions.

Long-term observational data on ozone and its precursors in China have revealed a noteworthy trend in the reactivity ratio of VOCs to NOx (VOC^R/NOx^R), which aligns closely with the normalized ozone levels. The relationship suggests that lowering the reactivity ratio serves as an optimal pathway for controlling ozone pollution (Wang et al., 2023). In this context, VOC^R (k _VOCs) is viewed as the primary production term for radical chemistry, while NOx^R (k _NOx) represents the corresponding consumption term.

The impact of temperature on emissions reduction was further investigated utilizing the VOC^R/NOx^R parameter, as depicted in Figure 8A. Within the temperature range of 15°C–30 °C, the reactivity ratio exhibited no significant generation trend, fluctuating between 2.17 and 2.31. The inflection point for the VOC^R/NOx^R trend was observed at 30 °C. Notably, during periods of extreme high temperatures, the increase in ozone production rate (P(O₃)) mirrored the temperature dependence of reactivity ratio (Figures 8A,B). The peak of P(O₃) reached approximately 20 ppb/h, aligning with the synchronous change in ozone concentration.

Global warming and the greenhouse effect further complicate the heatwave issue, as the increase in greenhouse gases such as CO₂ traps more heat in the atmosphere, leading to rising temperatures. The persistent warming trend not only accelerates ozone production but also amplifies the chemical reactivity of VOCs, making pollution control more challenging. By focusing on reducing the reactivity ratio, effective strategies can be developed to manage and control ozone levels in the atmosphere. From a pollution control perspective, regulating temperature by 1 °C reduces VOC^R/NOx^R by 0.12, thereby slowing the efficiency of the OH-HO₂-RO₂ radical cycle and inhibiting local ozone formation. Sensitivity tests under extreme temperature control scenarios demonstrate that the regulation of temperature consistently produces positive outcomes (as shown in Figure 8C). Within the temperature range of −10 to +10 °C, pollution control effects are sustained, though warming exacerbates adverse effects. By adopting a multifaceted approach that integrates temperature regulation and emission control, significant strides can be made towards improving air quality and protecting public health from the harmful effects of ozone pollution. Therefore, implementing measures such as limiting high energy consumption and using artificial rainfall to mitigate extreme heat can simultaneously lower VOC emission intensity and reactivity, facilitating the effective implementation of policies aimed at controlling ozone pollution.