Bin Chen

Long‐term changes in aerosol optical depth (AOD) and its vertical structure are important for understanding aerosol climate effects. However, how aerosols from different sources contribute to AOD changes across tropospheric layers remains unclear. Using CALIPSO observations from 2006 to 2021, together with time‐series decomposition, aerosol‐type analysis, and concentration‐weighted trajectory analysis, this study investigates AOD changes in the lower (0–4), middle (4–8), and upper (8–12 km) troposphere, cross‐validated by MODIS and MERRA‐2 data sets. Results show that global AOD changes exhibit pronounced vertical and regional differences. Lower‐tropospheric AOD decreased significantly over densely populated regions such as China and eastern U.S., mainly because of reductions in local anthropogenic‐influenced aerosols. Over West Asia, lower‐ and middle‐tropospheric dust weakened markedly, and the seasonal peak of lower‐tropospheric AOD declined by 41% after 2017, associated with weaker Arabian Peninsula dust emissions and suppressed dust transport under anomalous spring–summer circulation patterns. Across 30°N–60°N, dust and polluted dust declined widely in the middle and upper troposphere. After 2015, the trend components of middle‐ and upper‐tropospheric AOD over the northwestern Pacific decreased by 40% and 27%, respectively, consistent with weakened dust uplift and reduced eastward long‐range transport. Intensified wildfire activity after 2012 altered lower‐tropospheric AOD seasonality over central Russia from bimodal to trimodal and increased the frequency of middle‐ and upper‐tropospheric AOD anomalies over North America and western Europe. These major variations are generally consistent with MODIS and MERRA‐2 observations, highlighting the importance of vertical aerosol structures and transport processes in understanding global aerosol evolution.

Accurate monitoring of ground-level PM_2.5 is essential for air quality management, yet sparse in-situ observations limit large-scale surveillance. Although satellite-based retrieval has become indispensable, existing methods often struggle to simultaneously achieve high resolution, high accuracy, and physical interpretability under spatiotemporally heterogeneous conditions. To address these challenges, we propose STaR-PM, an interpretable spatiotemporal adaptive framework for PM_2.5 retrieval. The core of STaR-PM is an attention-augmented improved Kolmogorov–Arnold Network (iKAN), which represents the input–output relationship through variable-wise spline-based nonlinear response functions. This structure enables direct inspection of feature-specific effects and reveals regime-dependent, physically interpretable relationships between satellite observations, meteorological drivers, and PM_2.5 concentrations. To enhance robustness and generalization across regions and seasons, spatiotemporal transfer learning is integrated, allowing shared low-level representations to adapt to diverse spatiotemporal regimes. STaR-PM achieves superior performance, with an R² of 0.89 at 1-hour/0.05◦ resolution and 0.84 at 10-minute/0.02◦ resolution, outperforming multiple baseline models. Sensitivity analyses further demonstrate stable behavior under extreme pollution levels, varying cloud conditions, and across daytime and nighttime illumination regimes. Interpretability analyses based on iKAN response functions and interaction diagnostics identify key physical drivers, and elucidate their regime-dependent impacts on PM_2.5. Overall, STaR-PM effectively reconciles accuracy, resolution, generalization, and interpretability, providing a practical and physically grounded solution for large-scale, near-real-time air quality monitoring. Our source codes are available at https://github.com/HaoMingxiang/STaR-PM.

Understanding the urban-rural patterns and driving drivers behind the recent decrease in particulate matter (PM) pollution across eastern China is essential for assessing the efficacy of environmental policies and ensuring equitable health co-benefits. By employing an interpretable, end-to-end machine learning framework integrating satellite observations, meteorological factors, and auxiliary datasets, this study reveals changes in urban and rural PM pollution and the underlying drivers. During the period 2015–2023, the average decrease rates of PM₁₀ and PM_2.5 in eastern China were −4.02 ± 1.29 and −2.41 ± 0.91 µg m⁻³ yr⁻¹, respectively. The rate of decrease in urban areas was higher than that in rural areas, which played a dominant role in PM reduction. Significant reductions in PM concentrations were observed in urban core areas, suburbs, towns and regions with high agricultural pressure. The interpretability analysis showed that temperature and interannual variability were the main drivers of PM pollution reduction. However, only interannual variability showed a significant decreasing trend in its effect on PM pollution, while other driving factors showed periodic variations. Furthermore, there were differences in the drivers of PM reduction between urban and rural areas, particularly with interannual variability in particular contributing to PM pollution reduction in urban areas, but having a lesser impact in most rural areas. This study reveals the urban-rural patterns of PM pollution reduction in eastern China, and highlights the need for differentiated air pollution control strategies in urban and rural areas.

Urban-rural patterns of particulate matter (PM) pollution reduction in China remain poorly understood. Using an interpretable end-to-end machine learning model framework from original satellite data, we identified changes in urban and rural PM pollution and the underlying drivers. During the period 2015–2023, the average decrease rates of PM10 and PM2.5 in eastern China were -4.1±1.1 μg/m³/month and -2.4±0.8 μg/m³/month, respectively. The rate of decrease in urban areas was higher than that in rural areas, which played a dominant role in PM reduction. Significant reductions in PM concentrations were observed in urban core areas, suburbs, towns and regions with high agricultural pressure. The interpretability analysis showed that temperature and interannual variability were the main drivers of PM pollution reduction. However, only interannual variability showed a significant decreasing trend in its effect on PM pollution, while other driving factors showed periodic variations. Furthermore, there were differences in the drivers of PM reduction between urban and rural areas, particularly with interannual variability in particular contributing to PM pollution reduction in urban areas, but having a lesser impact in most rural areas. This study reveals the urban-rural patterns of PM pollution reduction in eastern China, and highlights the need for differentiated air pollution control strategies in urban and rural areas.

Urban-rural patterns of particulate matter (PM) pollution reduction in China remain poorly understood. Using an interpretable end-to-end machine learning model framework from original satellite data, we identified changes in urban and rural PM pollution and the underlying drivers. During the period 2015–2023, the average decrease rates of PM10 and PM2.5 in eastern China were -4.1±1.1 μg/m³/month and -2.4±0.8 μg/m³/month, respectively. The rate of decrease in urban areas was higher than that in rural areas, which played a dominant role in PM reduction. Significant reductions in PM concentrations were observed in urban core areas, suburbs, towns and regions with high agricultural pressure. The interpretability analysis showed that temperature and interannual variability were the main drivers of PM pollution reduction. However, only interannual variability showed a significant decreasing trend in its effect on PM pollution, while other driving factors showed periodic variations. Furthermore, there were differences in the drivers of PM reduction between urban and rural areas, particularly with interannual variability in particular contributing to PM pollution reduction in urban areas, but having a lesser impact in most rural areas. This study reveals the urban-rural patterns of PM pollution reduction in eastern China, and highlights the need for differentiated air pollution control strategies in urban and rural areas.

Urban-rural patterns of particulate matter (PM) pollution reduction in China remain poorly understood. Using an interpretable end-to-end machine learning model framework from original satellite data, we identified changes in urban and rural PM pollution and the underlying drivers. During the period 2015–2023, the average decrease rates of PM10 and PM2.5 in eastern China were -4.1±1.1 μg/m³/month and -2.4±0.8 μg/m³/month, respectively. The rate of decrease in urban areas was higher than that in rural areas, which played a dominant role in PM reduction. Significant reductions in PM concentrations were observed in urban core areas, suburbs, towns and regions with high agricultural pressure. The interpretability analysis showed that temperature and interannual variability were the main drivers of PM pollution reduction. However, only interannual variability showed a significant decreasing trend in its effect on PM pollution, while other driving factors showed periodic variations. Furthermore, there were differences in the drivers of PM reduction between urban and rural areas, particularly with interannual variability in particular contributing to PM pollution reduction in urban areas, but having a lesser impact in most rural areas. This study reveals the urban-rural patterns of PM pollution reduction in eastern China, and highlights the need for differentiated air pollution control strategies in urban and rural areas.

Urban-rural patterns of particulate matter (PM) pollution reduction in China remain poorly understood. Using an interpretable end-to-end machine learning model framework from original satellite data, we identified changes in urban and rural PM pollution and the underlying drivers. During the period 2015–2023, the average decrease rates of PM10 and PM2.5 in eastern China were -4.1±1.1 μg/m³/month and -2.4±0.8 μg/m³/month, respectively. The rate of decrease in urban areas was higher than that in rural areas, which played a dominant role in PM reduction. Significant reductions in PM concentrations were observed in urban core areas, suburbs, towns and regions with high agricultural pressure. The interpretability analysis showed that temperature and interannual variability were the main drivers of PM pollution reduction. However, only interannual variability showed a significant decreasing trend in its effect on PM pollution, while other driving factors showed periodic variations. Furthermore, there were differences in the drivers of PM reduction between urban and rural areas, particularly with interannual variability in particular contributing to PM pollution reduction in urban areas, but having a lesser impact in most rural areas. This study reveals the urban-rural patterns of PM pollution reduction in eastern China, and highlights the need for differentiated air pollution control strategies in urban and rural areas.

Urban-rural patterns of particulate matter (PM) pollution reduction in China remain poorly understood. Using an interpretable end-to-end machine learning model framework from original satellite data, we identified changes in urban and rural PM pollution and the underlying drivers. During the period 2015–2023, the average decrease rates of PM10 and PM2.5 in eastern China were -4.1±1.1 μg/m³/month and -2.4±0.8 μg/m³/month, respectively. The rate of decrease in urban areas was higher than that in rural areas, which played a dominant role in PM reduction. Significant reductions in PM concentrations were observed in urban core areas, suburbs, towns and regions with high agricultural pressure. The interpretability analysis showed that temperature and interannual variability were the main drivers of PM pollution reduction. However, only interannual variability showed a significant decreasing trend in its effect on PM pollution, while other driving factors showed periodic variations. Furthermore, there were differences in the drivers of PM reduction between urban and rural areas, particularly with interannual variability in particular contributing to PM pollution reduction in urban areas, but having a lesser impact in most rural areas. This study reveals the urban-rural patterns of PM pollution reduction in eastern China, and highlights the need for differentiated air pollution control strategies in urban and rural areas.

By utilizing top-of-atmosphere radiation (TOAR) data from China’s new generation of geostationary satellites (FY-4A and FY-4B) along with interpretable machine learning models, near-surface particulate matter concentrations in China were estimated, achieving hourly temporal resolution, 4 km spatial resolution, and 100% spatial coverage. First, the cloudless TOAR data were matched and modeled with the solar radiation products from the ERA5 dataset to construct and estimate a fully covered TOAR dataset under assumed clear-sky conditions, which increased coverage from 20–30% to 100%. Subsequently, this dataset was applied to estimate particulate matter. The analysis demonstrated that the fully covered TOAR dataset (R² = 0.83) performed better than the original cloudless dataset (R² = 0.76). Additionally, using feature importance scores and SHAP values, the impact of meteorological factors and air mass trajectories on the increase in PM₁₀ and PM_2.5 during dust events were investigated. The analysis of haze events indicated that the main meteorological factors driving changes in particulate matter included air pressure, temperature, and boundary layer height. The particulate matter concentration products obtained using fully covered TOAR data exhibit high coverage and high spatiotemporal resolution. Combined with data-driven interpretable machine learning, they can effectively reveal the influencing factors of particulate matter in China.