DA(data assimilation)is a crucial technical method for improving the accuracy of atmospheric chemical forecasts by integrating the results of atmospheric chemistry models with multi-source observational data,reducing uncertainties in model input data.Centering on DA techniques for atmospheric chemistry models,the transformation process of initial field assimilation for pollutant gases and aerosols from single state variables to multi-state variables was systematically reviewed.Meanwhile,the important progress of pollutant emission source assimilation inversion using ensemble methods and four-dimensional variational methods was focused on the improvement of emission source accuracy,optimization of spatiotemporal resolution,and enhancement of pollutant concentration prediction performance.With the explosive growth of observational data,a core challenge in the current field lies in fully leveraging high-resolution geospatial and remote sensing data for atmospheric chemical DA.The deep integration of DA with artificial intelligence algorithms represents a key research direction to break through this bottleneck and significantly enhance the accuracy of atmospheric composition analysis and forecasting.

Abstract Mineral dust drives Arctic climate variability, but accurate source attribution remains difficult due to plume mixing from the different regions. Using observations and HYSPLIT modeling, we characterize a record‐breaking trans‐continental dust intrusion (March 13–19, 2013) traversing from North Africa to the Arctic. Distinct from single‐source transport, a novel “cumulative‐snowballing” phenomenon‐Saharan dust uplifted into the free troposphere mixed with Central and East Asian emissions‐was identified. This blended plume was dominated by East (45.22%) and Central (23.62%) Asian sources, while retaining a non‐negligible Saharan contributions (17.34%). Driven by extreme multi‐regional emissions, the mid‐latitude westerly jet exported this dust across the North Pacific in the middle to lower troposphere, further a blocking‐like anticyclone opened a meridional pathway for continuous poleward intrusion. These findings demonstrate that Arctic aerosol loading can result from consecutive multi‐continental injections, highlighting the importance of considering complex mixing states in climate assessments.

This study critically examines the prevailing hypothesis that water vapor from fossil fuel combustion inherently exacerbates winter PM2.5 pollution, specifically focusing on the impacts of coal-to-gas (CTG) policy in the Beijing-Tianjin-Hebei (BTH) region, China. While existing literature often links combustion-derived humidity to enhanced aerosol formation and worsening haze, the effect of water vapor specifically from increased natural gas use remains contested. Employing a combined empirical and modeling approach-Propensity Score Matching with Difference-in-Differences (PSM-DID) and the WRF-Chem model-this research quantifies the influence of CTG policy on atmospheric moisture and its subsequent influence on PM2.5 concentrations. Our analysis reveals that the CTG policy significantly increased wintertime relative humidity by approximately 4.205 units. Contrary to expectations, model simulations indicate that the CTG-induced humidity rise does not lead to net enhancement of PM2.5 concentration. On average, an 8.7% increase in relative humidity correlates with a slight 0.44% decrease in PM2.5 levels during December 2017. Importantly, the elevated humidity contributes to substantial PM2.5 reductions under polluted conditions (PM2.5 > 75 μg m-3), with declines of up to 10% during extreme pollution episodes (PM2.5 > 400 μg m-3). Mechanistically, we find that while increased humidity promotes aerosol liquid water and secondary aerosol formation, it concurrently improves key meteorological diffusion conditions-like temperature, wind speed, and planetary boundary layer height-that enhance pollutant dispersion. The latter effect dominates under heavy pollution scenarios, leading to an overall amelioration. These findings challenge the simplified view of combustion-derived water vapor as a uniform aggravating factor for haze and underscore the context-dependent complexity of its role. The study provides crucial evidence for refining emission control strategies, suggesting that the benefits of fuel-switching policies may extend beyond direct emission reductions to include indirect and potentially favorable modifications of local meteorological conditions.

Global climate change and air pollution are closely linked environmental challenges. As the most abundant greenhouse gas, CO2 requires further investigation in relation to co-occurring air pollutants. Based on hourly observations from 16 monitoring-site groups across Europe, the United States, China, and Canada, this study examines the temporal and seasonal variability of CO2 and six air pollutants, including CO, NO2, SO2, PM2.5, PM10, and O3. The study periods differ among site groups and are defined by the common valid observation periods of CO2 and the six pollutants, with an overall temporal coverage from 2014 to 2025. Pearson correlation analysis quantified the relationships between CO2 and air pollutants, hierarchical clustering classified monitoring sites according to correlation patterns, and stepwise multiple regression assessed the effects of time scale and pollutant concentrations on CO2 variability. Results show that CO2 exhibits a clear seasonal cycle, with higher concentrations in winter and lower in summer. CO and NO2 show seasonal variations consistent with CO2, whereas O3 displays an opposite pattern (r: -0.83∼-0.09). The coupling between particulate matter (PM) and CO2 strengthens in winter but weakens in summer. CO2 is positively correlated with most pollutants except O3 in winter, while correlations decline substantially in summer. Cluster analysis separates the sites into anthropogenic emission-dominated and natural source-regulated groups. Stepwise regression (the maximum R2 is 0.92) indicates positive within-period linear tendencies and seasonal variability in CO2 at most sites, although model performance is limited where geographical and vegetation-related factors are excluded.

Accurate emission inventories are essential for effective air quality prediction and management. Traditional bottom‐up methods and top‐down data assimilation approaches struggle to balance accuracy, timeliness, and usability. This study presents a novel inversion method combining deep learning (DL) and 3D‐Variational (3DVar) assimilation, termed DL‐3DVar. Based on the prior emissions and SO 2 concentration observations, this method was applied to invert China's grided hourly SO 2 emission inventory during the 2023 Chinese New Year (CNY). The results show that DL‐3DVar estimated an hourly SO 2 emission increment of 1.95 × 10 6 kg at midnight on CNY's Eve relative to the prior emissions, primarily due to extensive fireworks. Forecast experiments showed that using DL‐3DVar inverted emissions reduced the root‐mean‐square error (RMSE) by 52.9% and increased the correlation coefficient (CORR) by 331% compared to the prior emissions. This high‐temporal‐resolution emission inventory can improve pollutant forecasting and provide robust support for air pollution control policymaking.

Understanding the composition of mercury (Hg) in the atmosphere is important for confirming its sources and to preventing and reduce the production. To explore the morphological distribution characteristics of wet Hg concentrations in Xi'an Shaanxi Province, China, total Hg (THg), dissolved Hg (DTHg), reactive Hg (RTHg) and particulate-bound Hg (PTHg) (Hg insoluble in water) were measured at 72 precipitation in Xi'an from September 2020 to July 2022, and their average concentrations were 3.035 ± 3.035, 1.352 ± 1.943, 0.414 ± 0.556, and 1.797 ± 1.681 ng L-1, respectively. Hg in wet deposition was mainly affected by particulate matter, and the proportion of PTHg in THg ranged from 16% to 92%, with an average of 55%. The observed seasonal concentrations variation order of THg was: winter > spring > summer > autumn. Positive matrix factorization analysis showed that Hg in precipitations mainly originated from four sources, including coal burning, traffic emission, mineral dust, and industrial emissions, accounting for 14.9%, 27.6%, 13.1%, and 14.2% of THg, respectively. Backward trajectory analysis showed that the air mass from Northwest China was the main air pollution source in Xi'an. Converting the amount of PTHg in the atmosphere (PBM) absorbed by the human body inhalation into the amount of smoking, the amount of PBM absorbed in one day corresponds to a reduction from 0.61 cigarettes to 0.28 cigarettes, after each precipitation event, which means precipitations have a significant dilution effect on PBM in the atmosphere.

This study investigated the spatial distribution and temporal variations of winter PM2.5 concentrations and light precipitation in North China from 2013 to 2022, and explored the relationship between them. The results show that winter PM2.5 concentrations in North China exhibit a south-high-north-low spatial pattern, with higher concentrations in the east than in the west, and a maximum average PM2.5 concentration exceeding 100 μg/m3. The overall interannual PM2.5 concentration shows a decreasing trend, which is most significant in the northern part of Hebei Province. Winter precipitation in North China demonstrates a similar south-high-north-low spatial pattern, with an upward trend in mean annual precipitation. PM2.5 concentration changes predominantly influence the frequency of light rainfall events. The relationship between PM2.5 concentrations and light rainfall frequency shows a clear spatial pattern: a significant positive correlation is observed west of the Taihang-Yanshan mountain range, and a negative correlation east of the mountain range. This occurs because rainfall frequency initially increases then decreases with increasing PM2.5 concentration. The differing background PM2.5 concentrations on either side of the mountain range result in opposing trends in light rainfall frequency relative to PM2.5 concentration changes.

In recent years, the random forest model has been widely applied to analyze the relationships among air pollution, meteorological factors, and human health. To investigate the patterns and influencing factors of respiratory disease-related medical visits, this study utilized data on medical visits from urban areas of Tianjin, meteorological observations, and pollution data. First, the temporal variation characteristics of medical visits from 2013 to 2019 were analyzed. Subsequently, the random forest model was employed to identify the dominant influencing factors of respiratory disease-related medical visits and to construct a statistical forecasting model that relates these factors to the number of visits. Additionally, a predictive analysis of medical visits in Tianjin for the year 2019 was conducted. The results indicate the following: (1) From 2013 to 2019, the number of medical visits exhibited seasonal fluctuations, with a significant decline observed in 2017, which may be directly related to adjustments in hospital policies. (2) Among the meteorological factors, average temperature, relative humidity, precipitation, and ozone concentration significantly influenced the variation in medical visits, while wind speed, precipitation amount, and boundary layer height were of lesser importance. Furthermore, different linear relationships exist among the meteorological factors; specifically, meteorological factors show a negative correlation with pollutant elements, and there is a strong correlation among the pollutant factors. (3) When the number of medical visits ranged from 50 to 200, the predictions made by the random forest model closely matched the actual values, demonstrating strong predictive performance and the ability to effectively forecast daily variations in medical visits over extended periods, thus exhibiting good stability and generalization capability. (4) However, since the random forest model relies on a large amount of data for model validation, it has limitations in capturing extreme variations in medical visit numbers. Future research could address this issue by integrating different models to enhance predictive capabilities.

Nitrogen dioxide (NO2) exposure has rarely been explored with respiratory outpatient visits in China. The Generalized Additive Model (GAM) is first employed to assess health related risk (RR) with NO2 and PM2.5 exposure in Tianjin during the winter of 2019/2020. The RR of NO2 exposure is 1.018 (95 % CI: 1.012-1.024) on lag 3 days, while that of fine particulate matter (PM2.5) exposure is 1.005 (95 % CI: 1.002-1.008) on lag 4 days per 10 μg m-3. The Weather Research and Forecasting model coupled to Chemistry (WRF-Chem) model is used to simulate the pollutant concentrations to calculate the total number of excess visits (EN) in the North China Plain (NCP) in the 2019/2020 winter. The EN of NO2 exposure is 0.97 million (95 % CI: 0.67, 1.27), while for PM2.5 exposure is 1.19 million (95 % CI: 0.50, 1.84). With the anthropogenic emission mitigation, the PM2.5 concentration has dropped by 40.3 μg m-3 on average in the NCP, but the NO2 concentration has rebounded by 1.8 μg m-3. The emissions mitigation reduces NO2-related EN by 0.76 million (95 % CI: 0.53-0.98) and PM2.5-related EN by 1.11 million (95 % CI: 0.48-1.69). The findings underscore that reducing NO2 emissions could yield more substantial health benefits.

Emissions from open crop straw burning (OCSB) during crop harvest have exerted considerable impacts on the air quality in China, but the quantitative evaluation of OCSB to air pollutants remains insufficient. In the study, the WRF-Chem model is used to quantitatively evaluate the impacts of OCSB on the air quality in Hunan, China from October 15 to 26, 2024. During the study episode, OCSB contributes 45.7% and 16.8% of the anthropogenic emission of organic carbon and non-methane volatile organic compounds, respectively. Model simulations show that OCSB emissions account for 2.8% of the maximum daily 8-hour average ozone (MDA8 O-3) concentration, but play a more substantial role in particulate pollution, contributing 25.6% to near-surface fine particulate matter levels. OCSB is an important organic aerosol (OA) source, increasing primary OA by 46.2% and secondary OA by 62.7%. The enhanced OA accounts for 54.8% of the total PM2.5 increase induced by OCSB emissions. Sensitivity experiments show that the concentrations of aerosol species generally decrease linearly with reduced OCSB emissions, while MDA8 O-3 concentrations exhibit a non-linear trend due to the complex responses to its precursors. Implementing the clean disposal of straw residue during crop harvest presents an effective strategy to improve the air quality in Hunan.

Atmospheric aerosols influence clouds and precipitation by aerosol-radiation interactions (ARIs) and aerosol-cloud interactions (ACIs). In this study, the synergetic effect of ARIs and ACIs on the development and precipitation of a mesoscale convective system (MCS) that occurred in the Guanzhong Basin (GZB) of central China have been examined using a cloud-resolving fully coupled weather research and forecasting model with chemistry (WRF-Chem). The model reasonably reproduces the temporal variation and spatial distribution of air pollutants, the hourly rain rate, and daily precipitation distribution against observations in the GZB. Sensitivity simulations are conducted under different aerosol scenarios by adjusting the anthropogenic emissions. When the ARI effect is not considered, the daily precipitation does not show an increasing trend with increasing aerosols in the GZB. This primarily reflects the effects of ACIs due to competition among convective clouds to available water vapor in the development of the MCS. ARIs exert two opposite effects on convection: a stabilizing effect to suppress convection and a lifting effect to foster convection, which counteract each other. When the lifting effect outweighs stabilizing effect, the updraft is enhanced, which increases precipitation in the GZB. However, the synergetic effect of ARIs and ACIs significantly suppress precipitation when the particulate-matter (PM) pollution is severe. Note that the synergetic effect consistently decreases the precipitation in the whole domain with increasing aerosols, but ARIs play a more important role in the decreasing trend of the precipitation with deterioration of PM pollution.

The Northeast Plain in China ranks among the top five regions that have been significantly impacted by haze pollution. To effectively control pollution, it is crucial to accurately assess the effects of emission reduction measures. In this study, we analyzed surveillance data and found substantial decreases (ranging from 19.0 % to 50.1 %) in average annual mass concentrations of key pollutants (such as CO, SO2, NO2, and PM2.5) in the Northeast Plain from 2016 to 2020. To precisely determine the contributions of meteorological conditions and emission reductions to the improvement of air quality in the Northeast Plain, we conducted three scenario simulations. By comparing source emissions in December 2016 and 2020 using the WRF-Chem model (except for SO2), we observed significant reductions of 21.3 %, 8.8 %, and 9.8 % in mass concentrations of PM2.5, NO2, and CO, respectively, from 2016 to 2020. This highlights the essential role that meteorological conditions play in determining air quality in the Northeast Plain. Moreover, further reducing source emissions by 30 % in December 2016 resulted in subsequent reductions of 25.3 %, 29.0 %, 4.5 %, and 30.3 % in mass concentrations of PM2.5, SO2, NO2, and CO, respectively, under the same meteorological conditions. Notably, source emission reduction was effective for PM2.5, SO2, and CO, but not for NO2. The improvement in air quality in the Northeast Plain from 2016 to 2020 can be attributed to the combined effects of improved meteorological conditions and reduced pollution sources.

The drawdown areas of the Three Gorges Reservoir (TGR) have emerged as significant greenhouse gas (GHG) hotspots, yet their contribution remains poorly quantified due to the difficulty of monitoring large, dynamic, and frequently cloud-covered landscapes. To address these challenges, we developed a remote sensing-based framework to generate the first high-resolution (30 m, monthly) estimates of CO2 fluxes from the TGR drawdown area over 2015-2023. By integrating all-weather Sentinel-1 SAR imagery with digital elevation models (DEM) and hydrological data, we successfully mapped the dynamic boundary between submerged and exposed lands, achieving an overall accuracy of 92.3 % (Kappa >0.8). Our results show that the drawdown area, covering only 21.7 % of the reservoir surface, contributes 392.2 ± 19.5 Gg of CO2 annually-accounting for 44.4 % of total emissions from the reservoir surface. Critically, we identify the seasonally exposed terrestrial phase as the overwhelming driver, responsible for 82.4 % of the drawdown area's total flux. A strong inverse correlation (r = -0.91) between CO2 emissions and reservoir water levels demonstrates that the anthropogenic operational regime for flood control and power generation-not natural seasonality-is the definitive control on these emissions. These findings establish that drawdown area emissions are a critical and previously under-accounted component of hydropower's carbon footprint. This work provides robust remote sensing-driven insights and actionable recommendations for sustainable reservoir operations and informed carbon accounting frameworks. We propose that adaptive reservoir management may mitigate these emissions.

Anthropogenic carbon dioxide (CO2) emissions are one of the most important source of greenhouse gases that driver climate change. However, there are still large uncertainties persist in the anthropogenic CO2 emissions and and no global observing system exists to monitor emissions from localized CO2 sources with sufficient accuracy. Combined nitrogen oxides (NOx) and CO2 observations showed the potential to constrain the identification of the locations and strength of anthropogenic CO2 emissions. In this study, a four-dimensional variational assimilation (4Dvar) system was developed to estimate high-resolution inversion of CO2 fluxes based on a regional chemical transport model WRF-Chem combined multiple observations. The ratios of NOx-to-CO2 from the TROPOspheric Monitoring Instrument (TROPOMI) and Orbiting Carbon Observatory 2 (OCO-2) satellite observations were applied to distinguish the enhancements of CO2 concentration from anthropogenic CO2 emissions. The gridded data set of monthly anthropogenic emissions (GCP-GridFED version 2022.2) was considered as the priori emission. The posterior CO2 emissions over China during April and September 2021 were estimated to reduce the influence of terrestrial ecosystem carbon sources and sinks. The posterior CO2 anthropogenic emissions reproduced the daily and hourly variation effectively, demonstrating the ability to fully absorb observations and the potential of applying NOx-to-CO2 ratio to estimate CO2 emissions. Two sets of forecast experiments were conducted to evaluate the prior and posterior CO2 simulations with independent surface observation data. Our results provide the basic data for carbon emissions reduction policies and may be helpful for accurate estimating carbon sinks, achieving carbon neutrality. The advanced data assimilation systems are beneficial to better quantify and characterize uncertainty for large carbon sources and sinks.

The distribution of insolation over time and space is a significant driver of climate change on orbital timescales. However, the influence of insolation on long-term climate evolution remains poorly understood due to the absence of a discernible long-term trend regulated by Earth's orbit. In this study, we present a sea surface temperature anomaly (SSTA) stack spanning the past 2 Myr, compiled from 26 millennial-resolved records obtained from the global ocean. The global average sea surface temperature (SST) reveals a 405-kyr cycle, as well as a gradual decrease of 2.34 f 1.05 degrees C (1 sigma) from 2000 ka to 940 ka, followed by a period of relative stability. We introduce an index named the integral of annual mean insolation anomaly (IAMIA), which quantifies the continuous departure of annual mean insolation (AMI) from its "normal" cycle over a specific time interval. We find that the SST leads the variations in AMI and IAMIA at the 405-kyr band, intimating that the cycle evident in global SST does not originate from the changes of eccentricity but rather stems from harmonic or combination tones. Notably, IAMIA exhibits a fundamental shift at 935 ka, coinciding with the "900-ka event" observed in the SST. Modeling results support that the "900-ka event" could be driven by the change of cumulative insolation. Additionally, IAMIA underscores the important role of insolation in Pleistocene climate change on the long-term trend through the cumulative response of ocean heat content (OHC) to successive small step-wise insolation changes. Furthermore, we hypothesize that the fundamental changes of insolation around 935 ka, transitioning from a positive state characterized by substantial amplitude to a negative state typified by diminished amplitude, facilitated the onset and progression of the mid-Pleistocene transition (MPT). This investigation provides invaluable insights into the role of insolation in long-term climate evolution.

Nuclear explosions and accidents release large amounts of radionuclides that harm human health and the environment. Accurate forecasting of nuclide pollutants and assessment of the ramifications of nuclear incidents are necessary for the emergency response and disaster assessment of nuclide pollution. In this study, we developed a three-dimensional variational (3Dvar) system to assimilate ^137 Cs based on the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) model. The distribution of ^137 Cs after the Fukushima nuclear accident in Japan on 15 March 2011 was analysed. The ^137 Cs background field at 06:00 UTC was assimilated using a 3Dvar system and surface observational data to optimise the ^137 Cs analysis field. Compared with the background field, the root mean square error (RMSE) and mean bias in the ^137 Cs analysis field decreased by 98% and 94%, respectively. The average fraction of predictions within factors of 2 (FAC2), 5 (FAC5), and 10 (FAC10) increased from 0.67, 0.72, and 0.72 to 0.90, 1.00, and 1.00, respectively. This substantial enhancement indicated the effectiveness of the 3DVar system in mitigating the uncertainty associated with the background field. Two 12 h forecast experiments were conducted to gauge the advancement in ^137 Cs forecasting facilitated by data assimilation (DA). The control experiment was conducted without DA, whereas the assimilation experiment was conducted with DA. Compared with the control experiment, the average FAC2, FAC5, and FAC10 in the assimilation experiment increased by 28%, 30%, and 29%, respectively. The average RMSE decreased by 33%. The mean bias and correlation coefficient increased by 41% and 36%, respectively. These results indicated that the 3Dvar method improves the forecast accuracy of ^137 Cs concentration.

During the 2023 Chinese New Year (CNY), many city governments temporarily relaxed firework restrictions, leading to increased sulfur dioxide (SO2) emissions from the combustion of sulfur-containing fireworks. This study employed the four-dimensional variational (4DVar) assimilation system to examine variations in SO2 emissions in China by assimilating hourly ground-based observations. Two experiments were conducted during CNY in 2022 and 2023 to quantify the variations in SO2 emissions. On CNY’s Eve in 2023, following the relaxation of the firework ban, SO2 emissions surged by 8.22 Gg nationwide compared to the previous day with significant increases in the Energy Golden Triangle (2.037 Gg), the North China Plain (1.709 Gg), and northeast China (0.945 Gg). Emissions peaked on CNY’s Eve and rapidly declined in the following two days but remained elevated compared to the pre-CNY period, indicating lingering effects of firework burning. Compared to the forecasts using the prior emissions, the optimized emissions markedly improved the model forecasts of SO2 during the 2023 CNY period, with an increase in the correlation coefficient (R) from 0.13 to 0.64 and a reduction in the root mean square error (RMSE) by 49.2%, demonstrating the effectiveness of the optimized emissions. These findings will be useful for local governments in formulating strategies for firework burning during CNY.

Abstract. Haze events across the North China Plain (NCP) during the COVID-19 lockdown have highlighted the complexities of air quality management in the face of reduced human activity. While previous studies have focused primarily on the atmospheric chemistry processes under anomalous weather conditions, interactions between air pollutants, atmospheric chemistry, and their responses to emissions and meteorological factors remain underexplored. Here, we utilized the WRF-Chem model to assess the impact of abrupt emission reductions and meteorological conditions on PM2.5 levels across the NCP. By comparing simulations sensitive to meteorological conditions with climatology averaged over 2015–2019 and considering the sudden decrease in anthropogenic emissions due to the lockdown, we identified significant regional disparities. In the Northern NCP (NNCP), adverse meteorological conditions negated the benefits of emission reductions, leading to a net increase in PM2.5 levels by 30 to 60 μg m-3 during haze episodes. Conversely, the Southern NCP (SNCP) experienced a decrease in PM2.5 levels attributed to favourable meteorological conditions combined with emission reductions, with decreases ranging from 20 to 40 μg m-3 during the same periods. Our results highlight the critical role of meteorological conditions in modulating the effects of emission reductions, particularly in regions like the NNCP, where adverse weather can significantly counteract the benefits of reduced emissions. This study provides valuable insights into the complex interactions between emissions, meteorology, and air quality, underscoring the necessity of integrated approaches that address emissions and atmospheric dynamics.

To study the long-term variation in ozone (O3) pollution in Sichuan Basin,the spatiaotemporal distribution of O3 concentrations during 2017 to 2020 was analyzed using ground-level O3 concentration data and meteorological observation data from 18 cities in the basin. The dominant meteorological factors affecting the variation in O3 concentration were screened out,and a prediction model between meteorological factors and O3 concentration was constructed based on a random forest model. Finally,a prediction analysis of O3 pollution in the Sichuan Basin urban agglomeration during 2020 was carried out. The results showed that:① O3 concentrations displayed a fluctuating trend during the period from 2017 to 2020,with a downward trend in 2019 and a rebound in 2020. ② The fluctuating trend of O3 concentration was significantly influenced by relative humidity,daily maximum temperature,and sunshine hours,whereas wind speed,air pressure,and precipitation had less impact. The linear relationships between meteorological factors were different. Air pressure was negatively correlated with other meteorological factors,whereas the remaining meteorological factors had a positive correlation. ③ The goodness of fit statistics (R2) between the predicted and actual values of the O3 prediction model constructed based on random forest demonstrated a strong predictive performance and ability to accurately forecast the long-term daily variations in O3 concentration. The random forest O3 prediction model exhibited excellent stability and generalization capability. ④ The prediction analysis of O3 concentrations in 18 cities in the basin showed that the explanation rate of variables in the prediction model reached over 80% in all cities (except Ya'an),indicating that the random forest model predicted the trend of O3 concentration accurately.