Estimating PM2.5 Concentrations in the Conterminous United States Using the Random Forest Approach

Long-term spatiotemporal variation of benzo[a]pyrene in Japan: Significant decrease in ambient concentrations, human exposure, and health risk

Although Benzo[a]pyrene (BaP) is considered carcinogenic to humans, the health effects of exposure to ambient levels have not been sufficiently investigated. This study estimated the long-term spatiotemporal variation of BaP in Japan over nearly two decades at a fine spatial resolution of 1 km. This study aimed to obtain an accurate spatiotemporal distribution of BaP that can be used in epidemiological studies on the health effects of ambient BaP exposure. The annual BaP concentrations were estimated using an ensemble machine learning approach using various predictors, including the concentrations and emission intensities of the criteria air pollutants, and meteorological, land use, and traffic-related variables. The model performance, evaluated by location-based cross-validation, exhibited satisfactory accuracy (R2 of 0.693). Densely populated areas showed higher BaP levels and greater temporal reduction, whereas BaP levels remained higher in some industrial areas. The population-weighted BaP in 2018 was 0.12 ng m-3, a decrease of approximately 70% from its 2000 value of 0.44 ng m-3, which was also reflected in the estimated excess number of lung cancer incidences. Accordingly, the proportion of BaP exposure below 0.12 ng m-3, which is the BaP concentration associated with an excess lifetime cancer risk of 10-5, reached 67% in 2018. Our estimates can be used in epidemiological studies to assess the health effects of BaP exposure at ambient concentrations.

Understanding the variability of ground-level ozone and fine particulate matter over the Tibetan plateau with data-driven approach

The Tibetan Plateau, known as the "Third Pole", is susceptible to ground-level ozone (O3) and fine particulate matter (PM2.5) pollution due to its unique high-altitude environment. This study constructed random forest regression models using multi-source data from ground measurements and meteorological satellites to predict variations in ground-level O3 and PM2.5 concentrations and their influencing factors across seven major cities in the Tibetan Plateau over two-year periods. The models successfully reproduced O3 and PM2.5 levels with satisfactory R-squared values of 0.71 and 0.73, respectively. Results reveal combustion-related carbon monoxide (CO) and nitrogen dioxide (NO2) as the most substantial influences on O3 and PM2.5 concentrations. Solar radiation, geographical factors, and meteorological variables also played crucial roles in driving pollutant variations. Conversely, transport-related and human activity factors exhibited relatively lower significance. High O3 and PM2.5 pollution occurred during pre-monsoon and post-monsoon/winter seasons, driven by solar radiation and emissions, respectively. While CO consistently contributed across cities and seasons, key influencing factors varied locally. This study unveils the key driving forces governing air pollutant variations across the Tibetan Plateau, shedding light on complex atmospheric processes in this unique high-altitude region.

Characteristics and partitions of traditional and emerging organophosphate esters in soil and groundwater based on machine learning

Organophosphate esters (OPEs) pose hazards to both humans and the environment. This study applied target screening to analyze the concentrations and detection frequencies of OPEs in the soil and groundwater of representative contaminated sites in the Pearl River Delta. The clusters and correlation characteristics of OPEs in soil and groundwater were calculated by self-organizing map (SOM). The risk assessment and partitions of OPEs in industrial park soil and groundwater were conducted. The results revealed that 14 out of 23 types of OPEs were detected. The total concentrations (Σ23OPEs) ranged from 1.931 to 743.571 ng/L in the groundwater, and 0.218 to 79.578 ng/g in the soil, the former showed highly soluble OPEs with high detection frequencies and concentrations, whereas the latter exhibited the opposite trend. SOM analysis revealed that the distribution of OPEs in the soil differed significantly from that in the groundwater. In the industrial park, OPEs posed acceptable risks in both the soil and groundwater. The soil could be categorized into Zone I and II, and the groundwater into Zone I, II, and III, with corresponding management recommendations. Applying SOM to analyze the characteristics and partitions of OPEs may provide references for other new pollutants and contaminated sites.

Observational Constraints on the Aerosol Optical Depth-Surface PM2.5 Relationship during Alaskan Wildfire Seasons

Wildfire is one of the main sources of PM2.5 (particulate matter with aerodynamic diameter < 2.5 μm) in the Alaskan summer. The complexity in wildfire smokes, as well as limited coverage of ground measurements, poses a big challenge to estimate surface PM2.5 during wildfire season in Alaska. Here we aim at proposing a quick and direct method to estimate surface PM2.5 over Alaska, especially in places exposed to strong wildfire events with limited measurements. We compare the AOD-surface PM2.5 conversion factor (η = PM2.5/AOD; AOD, aerosol optical depth) from the chemical transport model GEOS-Chem (ηGC) and from observations (ηobs). We show that ηGC is biased high compared to ηobs under smoky conditions, largely because GEOS-Chem assigns the majority of AOD (67%) within the planetary boundary layer (PBL) when AOD > 1, inconsistent with satellite retrievals from CALIOP. The overestimation in ηGC can be to some extent improved by increasing the injection height of wildfire emissions. We constructed a piecewise function for ηobs across different AOD ranges based on VIIRS-SNPP AOD and PurpleAir surface PM2.5 measurements over Alaska in the 2019 summer and then applied it on VIIRS AOD to derive daily surface PM2.5 over continental Alaska in the 2021 and 2022 summers. The derived satellite PM2.5 shows a good agreement with corrected PurpleAir PM2.5 in Alaska during the 2021 and 2022 summers, suggesting that aerosol vertical distribution likely represents the largest uncertainty in converting AOD to surface PM2.5 concentrations. This piecewise function, η'obs, shows the capability of providing an observation-based, quick and direct estimation of daily surface PM2.5 over the whole of Alaska during wildfires, without running a 3-D model in real time.

Unmasking the sky: high-resolution PM2.5 prediction in Texas using machine learning techniques

BACKGROUND

Although PM2.5 (fine particulate matter with an aerodynamic diameter less than 2.5 µm) is an air pollutant of great concern in Texas, limited regulatory monitors pose a significant challenge for decision-making and environmental studies.

OBJECTIVE

This study aimed to predict PM2.5 concentrations at a fine spatial scale on a daily basis by using novel machine learning approaches and incorporating satellite-derived Aerosol Optical Depth (AOD) and a variety of weather and land use variables.

METHODS

We compiled a comprehensive dataset in Texas from 2013 to 2017, including ground-level PM2.5 concentrations from regulatory monitors; AOD values at 1-km resolution based on images retrieved from the MODIS satellite; and weather, land-use, population density, among others. We built predictive models for each year separately to estimate PM2.5 concentrations using two machine learning approaches called gradient boosted trees and random forest. We evaluated the model prediction performance using in-sample and out-of-sample validations.

RESULTS

Our predictive models demonstrate excellent in-sample model performance, as indicated by high R2 values generated from the gradient boosting models (0.94-0.97) and random forest models (0.81-0.90). However, the out-of-sample R2 values fall within a range of 0.52-0.75 for gradient boosting models and 0.44-0.69 for random forest models. Model performance varies slightly across years. A generally decreasing trend in predicted PM2.5 concentrations over time is observed in Eastern Texas.

IMPACT STATEMENT

We utilized machine learning approaches to predict PM2.5 levels in Texas. Both gradient boosting and random forest models perform well. Gradient boosting models perform slightly better than random forest models. Our models showed excellent in-sample prediction performance (R2 > 0.9).

Combining Google traffic map with deep learning model to predict street-level traffic-related air pollutants in a complex urban environment

BACKGROUND

Traffic-related air pollution (TRAP) is a major contributor to urban pollution and varies sharply at the street level, posing a challenge for air quality modeling. Traditional land use regression models combined with data from fixed monitoring stations may be unable to predict and characterize fine-scale TRAP, especially in complex urban environments influenced by various features. This study aims to estimate fine-scale (50 m) concentrations of nitrogen oxides (NO and NO₂) in Hong Kong using a deep learning (DL) structured model.

METHODS

We collected data from mobile air quality sensors on buses and crowd-sourced Google real-time traffic status as a proxy for real-time traffic emissions. Our DL model was compared with existing machine learning models to assess performance improvements. Using an interpretable machine learning method, we hierarchically evaluated the global, local, and interaction effects for different features.

RESULTS

Our DL model outperformed existing machine learning models, achieving R2 values of 0.72 for NO and 0.69 for NO₂. The incorporation of traffic status as a key predictor improved model performance by 9% to 17%. The interpretable machine learning method revealed the importance of traffic-related features and their pairwise interactions.

CONCLUSION

The results indicate that traffic-related features significantly contribute to TRAP and provide insights and guidance for urban planning. By incorporating crowd-sourced Google traffic information, we assessed traffic abatement scenarios that could inform targeted strategies for improving urban air quality.

A hybrid model for enhanced forecasting of PM2.5 spatiotemporal concentrations with high resolution and accuracy

Forecasting concentrations of PM2.5 is important due to its known impacts on public health and environment. However, PM2.5 concentrations can vary significantly over short distances and time, which can be influenced by local emissions and short-term weather patterns. This spatiotemporal variability makes accurate PM2.5 forecasting an inherently complex and challenging task. This study presented novel methodologies for short-term PM2.5 concentration forecast by combining the atmospheric chemistry transport model Community Multiscale Air Quality Modeling System (CMAQ) with data-driven machine learning methods, namely long short-term memory (LSTM) and random forest (RF) models. The combined model system forecast PM2.5 with 1 h, 1km × 1 km spatiotemporal resolution. The LSTM system forecast time-dependent PM2.5 concentrations at observation sites with a maximum root mean square error (RMSE) of 3.66 μg/m3 for 1-hr forecast and 23.75 μg/m3 for 72-hr forecast, leveraging results obtained from the atmospheric transport model with RMSE of 45.81 μg/m3. Wavelet transform in the LSTM system allowed learning and prediction of PM2.5 concentrations at different frequencies, capturing temporal variability of PM2.5 at various time scales. The RF model predicted distributions of PM2.5 concentrations by learning LSTM results and integrating crucial features such as CMAQ results, meteorological and topographical information. The feature significance of CMAQ results was the highest among the input features in RF models. Overall, the hybrid model could help with managing and mitigating the adverse effects of air pollution by enabling informed decision-making at the individual, community and policy levels.

Examining air pollution exposure dynamics in disadvantaged communities through high-resolution mapping

This study bridges gaps in air pollution research by examining exposure dynamics in disadvantaged communities. Using cutting-edge machine learning and massive data processing, we produced high-resolution (100 meters) daily air pollution maps for nitrogen dioxide (NO2), fine particulate matter (PM2.5), and ozone (O3) across California for 2012-2019. Our findings revealed opposite spatial patterns of NO2 and PM2.5 to that of O3. We also identified consistent, higher pollutant exposure for disadvantaged communities from 2012 to 2019, although the most disadvantaged communities saw the largest NO2 and PM2.5 reductions and the advantaged neighborhoods experienced greatest rising O3 concentrations. Further, day-to-day exposure variations decreased for NO2 and O3. The disparity in NO2 exposure decreased, while it persisted for O3. In addition, PM2.5 showed increased day-to-day variations across all communities due to the increase in wildfire frequency and intensity, particularly affecting advantaged suburban and rural communities.

Estimating visibility and understanding factors influencing its variations at Bangkok airport using machine learning and a game theory-based approach

In this study, six individual machine learning (ML) models and a stacked ensemble model (SEM) were used for daytime visibility estimation at Bangkok airport during the dry season (November-April) for 2017-2022. The individual ML models are random forest, adaptive boosting, gradient boosting, extreme gradient boosting, light gradient boosting machine, and cat boosting. The SEM was developed by the combination of outputs from the individual models. Furthermore, the impact of factors affecting visibility was examined using the Shapley Additive exPlanation (SHAP) method, an interpretable ML technique inspired by the game theory-based approach. The predictor variables include different air pollutants, meteorological variables, and time-related variables. The light gradient boosting machine model is identified as the most effective individual ML model. On an hourly time scale, it showed the best performance across three out of four metrics with the ρ = 0.86, MB = 0, ME = 0.48 km (second lowest), and RMSE = 0.8 km. On a daily time scale, the model performed the best for all evaluation metrics with ρ = 0.92, MB = 0.0 km, ME = 0.3 km, and RMSE = 0.43 km. The SEM outperformed all the individual models across three out of four metrics on an hourly time scale with ρ = 0.88, MB = 0.0 km, (second lowest), and RMSE = 0.75 km. On the daily scale, it performed the best with ρ = 0.93, MB = 0.02 km, ME = 0.27 km, and RMSE = 0.4 km. The seasonal average original (VISorig) and meteorologically normalized visibility (VISnorm) decrease from 2017 to 2021 but increase in 2022. The rate of decrease in VISorig is double than rate of decrease in VISnorm which suggests the effect of meteorology visibility degradation. The SHAP analysis identified relative humidity (RH), PM2.5, PM10, day of the season year (i.e., Julian day) (JD), and O3 as the most important variables affecting visibility. At low RH, visibility is not sensitive to changes in RH. However, beyond a threshold, a negative correlation between RH and visibility is found potentially due to the hygroscopic growth of aerosols. The dependence of the Shapley values of PM2.5 and PM10 on RH and the change in average visibilities under different RH intervals also suggest the effect of hygroscopic growth of aerosol on visibility. A negative relationship has been identified between visibility and both PM2.5 and PM10. Visibility is positively correlated with O3 at lower to moderate concentrations, with diminishing impact at very high concentrations. The JD is strongly negatively related to visibility during winter while weakly associated positively later in summer. Findings from this research suggest the feasibility of employing machine learning techniques for predicting visibility and understanding the factors influencing its fluctuations.

Air-sea exchange of PAHs in the Taiwan Strait: Seasonal dynamics and regulation mechanisms revealed by machine learning approach

In this study, to understand the seasonal dynamics of air-sea exchange and its regulation mechanisms, we investigated polycyclic aromatic hydrocarbons (PAHs) at the air-sea interface in the western Taiwan Strait in combination with measurements and machine learning (ML) predictions. For 3-ring PAHs and most of 4- to 6-ring, volatilization and deposition fluxes were observed, respectively. Seasonal variations in air-sea exchange flux suggest the influence of monsoon transitions. Results of interpretable ML approach (XGBoost) indicated that volatilization of 3-ring PAHs was significantly controlled by dissolved PAH concentrations (contributed 24.0 %), and the gaseous deposition of 4- to 6-ring PAHs was related to more contaminated air masses originating from North China during the northeast monsoon. Henry's law constant emerged as a secondary factor, influencing the intensity of air-sea exchange, particularly for low molecular weight PAHs. Among environmental parameters, notably high wind speed emerges as the primary factor and biological pump's depletion of PAHs in surface seawater amplifies the gaseous deposition process. The distinct dynamics of exchanges at the air-water interface for PAHs in the western TWS can be attributed to variations in primary emission intensities, biological activity, and the inconsistent pathways of long-range atmospheric transport, particularly within the context of the monsoon transition.

Predicting hospital admissions for upper respiratory tract complaints: An artificial neural network approach integrating air pollution and meteorological factors

This study uses artificial neural networks (ANNs) to examine the intricate relationship between air pollutants, meteorological factors, and respiratory disorders. The study investigates the correlation between hospital admissions for respiratory diseases and the levels of PM10 and SO2 pollutants, as well as local meteorological conditions, using data from 2017 to 2019. The objective of this study is to clarify the impact of air pollution on the well-being of the general population, specifically focusing on respiratory ailments. An ANN called a multilayer perceptron (MLP) was used. The network was trained using the Levenberg-Marquardt (LM) backpropagation algorithm. The data revealed a substantial increase in hospital admissions for upper respiratory tract diseases, amounting to a total of 11,746 cases. There were clear seasonal fluctuations, with fall having the highest number of cases of bronchitis (N = 181), sinusitis (N = 83), and upper respiratory infections (N = 194). The study also found demographic differences, with females and people aged 18 to 65 years having greater admission rates. The performance of the ANN model, measured using R2 values, demonstrated a high level of predictive accuracy. Specifically, the R2 value was 0.91675 during training, 0.99182 during testing, and 0.95287 for validating the prediction of asthma. The comparative analysis revealed that the ANN-MLP model provided the most optimal result. The results emphasize the effectiveness of ANNs in representing the complex relationships between air quality, climatic conditions, and respiratory health. The results offer crucial insights for formulating focused healthcare policies and treatments to alleviate the detrimental impact of air pollution and meteorological factors.

Leveraging physics-based and explainable machine learning approaches to quantify the relative contributions of rain and air pollutants to wet deposition

A quantitative understanding of the roles of rainfall and pollutant concentrations in wet deposition is important because they critically influence terrestrial and aquatic ecosystems. However, their relative contributions to wet deposition, which vary across regions, have not yet been identified. We propose two methods that quantitatively separate the contributions of rain and pollutant concentrations to wet deposition: one is based on simplified equations describing the wet scavenging of pollutants and the other is based on random forest models employing SHapley Additive exPlanations. Three-dimensional long-term air quality simulations from 2003 to 2019 are used as inputs for both the physics-based and machine learning models. Remarkably, the results drawn from the explainable machine learning model are consistent with those from the physics-based approach: overall, rain is a more important limiting factor than pollutant concentrations and the relative contribution of rain is larger than that of pollutants by up to a factor of 3-4 in polluted regions. In polluted regions, pollutant concentrations can remain relatively high even in the presence of precipitation owing to continuous and intense emissions; therefore, wet deposition is limited by rainfall. The contribution of rainfall is larger by 1.5-2.5 than that of pollutant concentrations in regions even with low emissions and this considerably large role of rain suggests that regional or transboundary pollutant transport plays a key role in modulating wet deposition. However, in very remote regions, once the rainfall amount exceeds a certain value, rainfall no longer contributes to increasing wet deposition because atmospheric pollutants are readily removed by rain. So, the contributions of the two factors are comparable in pristine regions. Our results can serve as a basis for explaining interannual variations in wet deposition and for future projections of wet deposition under emission control plans and climate change scenarios across regions.

Mapping nationwide concentrations of sulfate and nitrate in ambient PM2.5 in South Korea using machine learning with ground observation data

Particulate matter (PM) is a major air pollutant in Northeast Asia, with frequent high PM episodes. To investigate the nationwide spatial distribution maps of PM2.5 and secondary inorganic aerosols in South Korea, prediction models for mapping SO42- and NO3- concentrations in PM2.5 were developed using machine learning with ground-based observation data. Specifically, the random forest algorithm was used in this study to predict the SO42- and NO3- concentrations at 548 air quality monitoring stations located within the representative radii of eight intensive air quality monitoring stations. The average concentrations of PM2.5, SO42-, and NO3- across the entire nation were 17.2 ± 2.8, 3.0 ± 0.6, and 3.4 ± 1.2 μg/m3, respectively. The spatial distributions of SO42- and NO3- concentrations in 2021 revealed elevated concentrations in both the western and central regions of South Korea. This result suggests that SO42- concentrations were primarily influenced by industrial activities rather than vehicle emissions, whereas NO3- concentrations were more associated with vehicle emissions. During a high PM2.5 event (November 19-21, 2021), the concentration of SO42- was primarily influenced by SOX emissions from China, while the concentration of NO3- was affected by NOX emissions from both China and Korea. The methodology developed in this study can be used to explore the chemical characteristics of PM2.5 with high spatiotemporal resolution. It can also provide valuable insights for the nationwide mitigation of secondary PM2.5 pollution.

Novel approach for identifying VOC emission characteristics based on mobile monitoring platform data and deep learning: Application of source apportionment in a chemical industrial park

Refined volatile organic compound (VOC) emission characteristics are crucial for accurate source apportionment in chemical industrial parks. The data from mobile monitoring platforms in chemical industrial parks contain pollution information that is not intuitively displayed, requiring further excavation. A novel approach was proposed to identify VOC emission characteristics using the class activation map (CAM) technology of convolutional neural network (CNN), which was applied on the mobile monitoring platform data (MD) derived from a typical fine chemical industrial park. It converts a large amount of monitoring data with high spatiotemporal complexity into simple and interpretable characteristic maps, effectively improving the identification effect of VOC emission characteristics, supporting more accurate source apportionment of VOC pollution around the park. Using this method, the VOC emission characteristics of eight key factories were identified. VOC source apportionment in the park was conducted for one day using a positive matrix factorization (PMF) model and seven combined factor profiles (CFPs) were calculated. Based on the identified VOC emission characteristics, the main pollution sources and their contributions to surrounding schools and residential areas were determined, revealing that one pesticide factory (named LKA) had the highest contribution ratio. The source apportionment results indicated that the impact of the chemical industrial park on the surrounding areas varied from morning to afternoon, which to some extent reflected the intermittent production methods employed for fine chemicals.

Spatiotemporal distribution, light absorption characteristics, and source apportionments of black and brown carbon in China

Black carbon (BC) and brown carbon (BrC) are aerosols that absorb light and thereby contribute to climate change. In this study, the light absorption properties and spatiotemporal distributions of equivalent BC (eBC) and BrC aerosols were determined based on continuous measurements of aerosol light absorption from January to August 2017, using a seven-channel aethalometer at 49 sampling sites in China. The source apportionments of BC and BrC were identified using the BC/PM2.5, absorption Ångström exponent, the concentration-weighted trajectory method, and the random forest model. Based on the results, BC was the dominant light absorber, whereas BrC was responsible for a higher proportion of the light absorption in northern compared to southern China. The light absorption of BrC was highest in winter (34.3 Mm-1), followed by spring (19.0 Mm-1) and summer (3.6 Mm-1). The combustion of liquid fuels accounted for over 50 % of the light absorption coefficient of BC in most cities and the importance of carbon monoxide (CO) and nitrogen dioxide (NO2) was over 10 % for BC emitted by liquid fuel combustion, based on the random forest model. The contribution of solid fuel combustion to BC in the north was larger than that in the southern regions as coal combustion and crop residue burning are important emission sources of BC in most northern cities. The contribution of primary BrC to light absorption was high in some northern cities, whereas that of secondary BrC was prevalent in some southern cities. The diurnal variations in secondary BrC were affected by changes in odd oxygen and relative humidity, which promoted the photobleaching of the chromophores and aqueous-phase reactions of secondary BrC.

Residential exposure to microbial emissions from livestock farms: Implementation and evaluation of land use regression and random forest spatial models

Adverse health effects have been linked with exposure to livestock farms, likely due to airborne microbial agents. Accurate exposure assessment is crucial in epidemiological studies, however limited studies have modelled bioaerosols. This study used measured concentrations in air of livestock commensals (Escherichia coli (E. coli) and Staphylococcus species (spp.)), and antimicrobial resistance genes (tetW and mecA) at 61 residential sites in a livestock-dense region in the Netherlands. For each microbial agent, land use regression (LUR) and random forest (RF) models were developed using Geographic Information System (GIS)-derived livestock-related characteristics as predictors. The mean and standard deviation of annual average concentrations (gene copies/m3) of E. coli, Staphylococcus spp., tetW and mecA were as follows: 38.9 (±1.98), 2574 (±3.29), 20991 (±2.11), and 15.9 (±2.58). Validated through 10-fold cross-validation (CV), the models moderately explained spatial variation of all microbial agents. The best performing model per agent explained respectively 38.4%, 20.9%, 33.3% and 27.4% of the spatial variation of E. coli, Staphylococcus spp., tetW and mecA. RF models had somewhat better performance than LUR models. Livestock predictors related to poultry and pig farms dominated all models. To conclude, the models developed enable enhanced estimates of airborne livestock-related microbial exposure in future epidemiological studies. Consequently, this will provide valuable insights into the public health implications of exposure to specific microbial agents.

Photochemical oxidation of VOCs and their source impact assessment on ozone under de-weather conditions in Western Taiwan

The application of statistical models has excellent potential to provide crucial information for mitigating the challenging issue of ozone (O3) pollution by capturing its associations with explanatory variables, including reactive precursors (VOCs and NOX) and meteorology. Considering the large contribution of O3 in degrading the air quality of western Taiwan, three-year (2019-2021) hourly concentration data of VOC, NOX and O3 from 4 monitoring stations of western Taiwan: Tucheng (TC), Zhongming (ZM), Taixi (TX) and Xiaogang (XG), was evaluated to identify the effect of anthropogenic emissions on O3 formation. Owing to the high-ambient reactivity of VOCs on the underestimation of sources, photochemical oxidation was assessed to calculate the consumed VOC (VOCcons) which was followed by the source identification of their initial concentrations. VOCcons was observed to be highest in the summer season (16.7 and 22.7 ppbC) at north (TC and ZM) and in the autumn season (17.8 and 11.4 ppbC) in southward-located stations (TX and XG, respectively). Results showed that VOCs from solvents (25-27%) were the major source at northward stations whereas VOCs-industrial emissions (30%) dominated in south. Furthermore, machine learning (ML): eXtreme Gradient Boost (XGBoost) model based de-weather analysis identified that meteorological factors favor to reduce ambient O3 levels at TC, ZM and XG stations (-67%, -47% and -21%, respectively) but they have a major role in accumulating the O3 (+38%) at the TX station which is primarily transported from the upwind region of south-central Taiwan. Crucial insights using ML outputs showed that the finding of the study can be utilized for region-specific data-driven control of emission from VOCs-sources and prioritized to limit the O3-pollution at the study location-ns as well as their accumulation in distant regions.

Optimizing the development of contaminated land in China: Exploring machine-learning to identify risk markers

Often available for use, previously developed land, which includes residential and commercial/industrial areas, presents a significant challenge due to the risk to human health. China's 2018 release of health risk assessment standards for land reuse aimed to bridge this gap in soil quality standards. Despite this, the absence of representative indicators strains risk managers economically and operationally. We improved China's land redevelopment approach by leveraging a dataset of 297,275 soil samples from 352 contaminated sites, employing machine learning. Our method incorporating soil quality standards from seven countries to discern patterns for establishing a cost-effective evaluative framework. Our research findings demonstrated that detection costs could be curtailed by 60% while maintaining consistency with international soil standards (prediction accuracy = 90-98%). Our findings deepen insights into soil pollution, proposing a more efficient risk assessment system for land redevelopment, addressing the current dearth of expertise in evaluating land development in China.

BREATH-Net: a novel deep learning framework for NO2 prediction using bi-directional encoder with transformer

Air pollution poses a significant challenge in numerous urban regions, negatively affecting human well-being. Nitrogen dioxide (NO2) is a prevalent atmospheric pollutant that can potentially exacerbate respiratory ailments and cardiovascular disorders and contribute to cancer development. The present study introduces a novel approach for monitoring and predicting Delhi's nitrogen dioxide concentrations by leveraging satellite data and ground data from the Sentinel 5P satellite and monitoring stations. The research gathers satellite and monitoring data over 3 years for evaluation. Exploratory data analysis (EDA) methods are employed to comprehensively understand the data and discern any discernible patterns and trends in nitrogen dioxide levels. The data subsequently undergoes pre-processing and scaling utilizing appropriate techniques, such as MinMaxScaler, to optimize the model's performance. The proposed forecasting model uses a hybrid architecture of the Transformer and BiLSTM models called BREATH-Net. BiLSTM models exhibit a strong aptitude for effectively managing sequential data by adeptly capturing dependencies in both the forward and backward directions. Conversely, transformers excel in capturing extensive relationships over extended distances in temporal data. The results of this study will illustrate the proposed model's efficacy in predicting the levels of NO2 in Delhi. If effectively executed, this model can significantly enhance strategies for controlling urban air quality. The findings of this research show a significant improvement of RMSE = 9.06 compared to other state-of-the-art models. This study's primary objective is to contribute to mitigating respiratory health issues resulting from air pollution through satellite data and deep learning methodologies.

Predicting high-resolution air quality using machine learning: Integration of large eddy simulation and urban morphology data

Accurately predicting air pollutants, especially in urban areas with well-defined spatial structures, is crucial. Over the past decade, machine learning techniques have been widely used to forecast urban air quality. However, traditional machine learning approaches have limitations in accuracy and interpretability for predicting pollutants. In this study, we propose a convolutional neural network (CNN) model to predict the spatial distribution of CO concentration in Nanjing urban area at 10 m resolution. Our model incorporates various factors as input, such as building height, topography, emissions, and is trained against the outputs simulated by the parallelized large-eddy simulation model (PALM). The PALM model has 48 different scenarios that varied in emissions, wind speeds, and wind directions. The results display a strong consistency between the two models. Furthermore, we evaluate the performance of our model using a 10-fold cross-validation and out-of-sample cross-validation approach. This yields a robust correlation (with both R2 > 0.8) and a low RMSE between the CO predicted by the PALM and CNN models, which demonstrates the generalization capability of our CNN model. The CNN can extract crucial features from the resulted weight contribution map. This map indicates that the CO concentration at a location is more influenced by nearby buildings and emissions than distant ones. The interpretable patterns uncovered by our model are related to neighborhood effects, wind speeds, directions, and the impact of orientation on urban CO distribution. The model also shows high prediction accuracy (R > 0.8) when applied to another city. Overall, the integration of our CNN framework with the PALM model enhances the accuracy of air quality predictions, while enabling a fluid dynamic laws interpretation, providing effective tools for air quality management.

Urbanization influences dissolved organic matter characteristics but microbes affect greenhouse gas concentrations in lakes

Recognition and prediction of dissolved organic matter (DOM) properties and greenhouse gas (GHG) emissions is critical to understanding climate change and the fate of carbon in aquatic ecosystems, but related data is challenging to interpret due to covariance in multiple natural and anthropogenic variables with high spatial and temporal heterogeneity. Here, machine learning modeling combined with environmental analysis reveals that urbanization (e.g., population density and artificial surfaces) rather than geography determines DOM composition and properties in lakes. The structure of the bacterial community is the dominant factor determining GHG emissions from lakes. Urbanization increases DOM bioavailability and decreases the DOM degradation index (Ideg), increasing the potential for DOM conversion into inorganic carbon in lakes. The traditional fossil fuel-based path (SSP5) scenario increases carbon emission potential. Land conversion from water bodies into artificial surfaces causes organic carbon burial. It is predicted that increased urbanization will accelerate the carbon cycle in lake ecosystems in the future, which deserves attention in climate models and in the management of global warming.

Application of machine learning in atmospheric pollution research: A state-of-art review

Machine learning (ML) is an artificial intelligence technology that has been used in atmospheric pollution research due to their powerful fitting ability. In this review, 105 articles related to ML and the atmospheric pollution research are critically reviewed. Applications of ML in the prediction of atmospheric pollution (mainly particulate matters) are systematically described, including the principle of prediction, influencing factors and improvement measures. Researchers can improve the accuracy of the prediction model through three main aspects, namely considering the geographical features of the study area into the model, introducing the physical characteristics of pollutants, matching and optimizing ML models. And by using interpretable ML tools, researchers are able to understand the mechanism of the model and gain in-depth information. Then, the state-of-art applications of ML in the source apportionment of atmospheric particulate matter and the effect of atmospheric pollutants on human health are also described. In addition, the advantages and disadvantages of the current applications of ML in atmospheric pollution research are summarized, and the application perspective of ML in this field is elucidated. Given the scarcity of source apportionment applications and human health research, standardized research methods and specialized ML methods are required in atmospheric pollution research to connect these two disciplines.

Quantifying the contribution of environmental variables to cyclists' exposure to PM2.5 using machine learning techniques

Cyclists are particularly vulnerable to travel-related exposure to air pollution. Understanding the factors that increase exposure is crucial for promoting healthier urban environments. Machine learning models have successfully predicted air pollutant concentrations, but they tend to be less interpretable than classical statistical ones, such as linear models. This study aimed to develop a predictive model to assess cyclists' exposure to fine particulate matter (PM2.5) in urban environments. The model was generated using geo-temporally referenced data and machine learning techniques. We explored several models and found that the gradient boosting machine learning model best fitted the PM2.5 predictions, with a minimum root mean square error value of 5.62 μg m-3. The variables with greatest influence on cyclist exposure were the temporal ones (month, day of the week, and time of the day), followed by meteorological variables, such as temperature, relative humidity, wind speed, wind direction, and atmospheric pressure. Additionally, we considered relevant attributes, which are partially linked to spatial characteristics. These attributes encompass street typology, vegetation density, and the flow of vehicles on a particular street, which quantifies the number of vehicles passing a given point per minute. Mean PM2.5 concentration was lower in bicycle paths away from vehicular traffic than in bike lanes along streets. These outcomes underscore the need to thoughtfully design public transportation routes, including bus routes, concerning the network of bicycle pathways. Such strategic planning attempts to improve the air quality in urban landscapes.

High-spatiotemporal-resolution mapping of PM2.5 traffic source impacts integrating machine learning and source-specific multipollutant indicator

Traffic sources are a major contributor to fine particulate matter (PM2.5) pollution, with their emissions and diffusion exhibiting complex spatiotemporal patterns. Receptor models have limitations in estimating high-resolution source contributions due to insufficient observation networks of PM2.5 compositions. This study developed a source apportionment method that integrates machine learning and emission-based integrated mobile source indicator (IMSI) to rapidly and accurately estimate PM2.5 traffic source impacts with high spatiotemporal resolution in the Beijing-Tianjin-Hebei region. Firstly, we utilized multisource data and developed various machine learning models to optimize the traffic-related pollutant concentration fields simulated by a chemical transport model. Results demonstrated that the Extreme Gradient Boosting (XGBoost) model exhibited excellent prediction accuracy of nitrogen oxide (NO2), carbon oxide (CO), and elemental carbon (EC), with the cross-validated R values increasing to 0.87-0.92 and error indices decreasing by 50-67%. Furthermore, we estimated and predicted daily mappings of PM2.5 traffic source impacts using the IMSI method based on optimized concentration fields, which improved spatially resolved source contributions to PM2.5. Our findings reveal that PM2.5 traffic source impacts display significant spatial heterogeneity, and these hotspots can be precisely identified during the pollution processes with sharp changes. The evaluation results indicated that there is a good correlation (R of 0.79) between PM2.5 traffic source impacts by IMSI method and traffic source contributions apportioned by a receptor model at Beijing site. Our study provides deeper insights of estimating the spatiotemporal distribution of PM2.5 source-specific impacts especially in regions without PM2.5 compositions, which can provide more complete and timely guidance to implement precise air pollution management strategies.

The spatiotemporal pattern of surface ozone and its impact on agricultural productivity in China

The slowing of agricultural productivity growth globally over the past two decades has brought a new urgency to detect its drivers and potential solutions. We show that air pollution, particularly surface ozone (O3), is strongly associated with declining agricultural total factor productivity (TFP) in China. We employ machine learning algorithms to generate estimates of high-resolution surface O3 concentrations from 2002 to 2019. Results indicate that China's O3 pollution has intensified over this 18-year period. We coupled these O3 estimates with a statistical model to show that rising O3 pollution during nonwinter seasons has reduced agricultural TFP by 18% over the 2002-2015 period. Agricultural TFP is projected to increase by 60% if surface O3 concentrations were reduced to meet the WHO air quality standards. This productivity gain has the potential to counter expected productivity losses from 2°C warming.

PM2.5-Attributable Mortality Burden Variability in the Continental U.S

Epidemiologic studies have consistently observed associations between fine particulate matter (PM2.5) exposure and premature mortality. These studies use air quality concentration information from a combination of sources to estimate pollutant exposures and then assess how mortality varies as a result of differing exposures. Health impact assessments then typically use a single log-linear hazard ratio (HR) per health outcome to estimate counts of avoided human health effects resulting from air quality improvements. This paper estimates the total PM2.5-attributable premature mortality burden using a variety of methods for estimating exposures and quantifying PM2.5-attributable deaths in 2011 and 2028. We use: 1) several exposure models that apply a wide range of methods, and 2) a variety of HRs from the epidemiologic literature that relate long-term PM2.5 exposures to mortality among the U.S. population. We then further evaluate the variability of aggregated national premature mortality estimates to stratification by race and/or ethnicity or exposure level (e.g., below the current annual PM2.5 National Ambient Air Quality Standards). We find that unstratified annual adult mortality burden incidence estimates vary more (e.g., ~3-fold) by HR than by exposure model (e.g., <10%). In addition, future mortality burden estimates stratified by race/ethnicity are larger than the unstratified estimates of the entire population, and studies that stratify PM2.5-attributable mortality HRs by an exposure concentration threshold led to substantially higher estimates. These results are intended to provide transparency regarding the sensitivity of mortality estimates to upstream input choices.

Long-term mortality burden trends attributed to black carbon and PM2·5 from wildfire emissions across the continental USA from 2000 to 2020: a deep learning modelling study

BACKGROUND

Long-term improvements in air quality and public health in the continental USA were disrupted over the past decade by increased fire emissions that potentially offset the decrease in anthropogenic emissions. This study aims to estimate trends in black carbon and PM2·5 concentrations and their attributable mortality burden across the USA.

METHODS

In this study, we derived daily concentrations of PM2·5 and its highly toxic black carbon component at a 1-km resolution in the USA from 2000 to 2020 via deep learning that integrated big data from satellites, models, and surface observations. We estimated the annual PM2·5-attributable and black carbon-attributable mortality burden at each 1-km2 grid using concentration-response functions collected from a national cohort study and a meta-analysis study, respectively. We investigated the spatiotemporal linear-regressed trends in PM2·5 and black carbon pollution and their associated premature deaths from 2000 to 2020, and the impact of wildfires on air quality and public health.

FINDINGS

Our results showed that PM2·5 and black carbon estimates are reliable, with sample-based cross-validated coefficients of determination of 0·82 and 0·80, respectively, for daily estimates (0·97 and 0·95 for monthly estimates). Both PM2·5 and black carbon in the USA showed significantly decreasing trends overall during 2000 to 2020 (22% decrease for PM2·5 and 11% decrease for black carbon), leading to a reduction of around 4200 premature deaths per year (95% CI 2960-5050). However, since 2010, the decreasing trends of fine particles and premature deaths have reversed to increase in the western USA (55% increase in PM2·5, 86% increase in black carbon, and increase of 670 premature deaths [460-810]), while remaining mostly unchanged in the eastern USA. The western USA showed large interannual fluctuations that were attributable to the increasing incidence of wildfires. Furthermore, the black carbon-to-PM2·5 mass ratio increased annually by 2·4% across the USA, mainly due to increasing wildfire emissions in the western USA and more rapid reductions of other components in the eastern USA, suggesting a potential increase in the relative toxicity of PM2·5. 100% of populated areas in the USA have experienced at least one day of PM2·5 pollution exceeding the daily air quality guideline level of 15 μg/m3 during 2000-2020, with 99% experiencing at least 7 days and 85% experiencing at least 30 days. The recent widespread wildfires have greatly increased the daily exposure risks in the western USA, and have also impacted the midwestern USA due to the long-range transport of smoke.

INTERPRETATION

Wildfires have become increasingly intensive and frequent in the western USA, resulting in a significant increase in smoke-related emissions in populated areas. This increase is likely to have contributed to a decline in air quality and an increase in attributable mortality. Reducing fire risk via effective policies besides mitigation of climate warming, such as wildfire prevention and management, forest restoration, and new revenue generation, could substantially improve air quality and public health in the coming decades.

FUNDING

National Aeronautics and Space Administration (NASA) Applied Science programme, NASA MODIS maintenance programme, NASA MAIA satellite mission programme, NASA GMAO core fund, National Oceanic and Atmospheric Administration (NOAA) GEO-XO project, NOAA Atmospheric Chemistry, Carbon Cycle, and Climate (AC4) programme, and NOAA Educational Partnership Program with Minority Serving Institutions.

Machine Learning in Environmental Research: Common Pitfalls and Best Practices

Machine learning (ML) is increasingly used in environmental research to process large data sets and decipher complex relationships between system variables. However, due to the lack of familiarity and methodological rigor, inadequate ML studies may lead to spurious conclusions. In this study, we synthesized literature analysis with our own experience and provided a tutorial-like compilation of common pitfalls along with best practice guidelines for environmental ML research. We identified more than 30 key items and provided evidence-based data analysis based on 148 highly cited research articles to exhibit the misconceptions of terminologies, proper sample size and feature size, data enrichment and feature selection, randomness assessment, data leakage management, data splitting, method selection and comparison, model optimization and evaluation, and model explainability and causality. By analyzing good examples on supervised learning and reference modeling paradigms, we hope to help researchers adopt more rigorous data preprocessing and model development standards for more accurate, robust, and practicable model uses in environmental research and applications.

Predicting Atmospheric Water-Soluble Organic Mass Reversibly Partitioned to Aerosol Liquid Water in the Eastern United States

Water-soluble organic matter (WSOM) formed through aqueous processes contributes substantially to total atmospheric aerosol, however, the impact of water evaporation on particle concentrations is highly uncertain. Herein, we present a novel approach to predict the amount of evaporated organic mass induced by sample drying using multivariate polynomial regression and random forest (RF) machine learning models. The impact of particle drying on fine WSOM was monitored during three consecutive summers in Baltimore, MD (2015, 2016, and 2017). The amount of evaporated organic mass was dependent on relative humidity (RH), WSOM concentrations, isoprene concentrations, and NOx/isoprene ratios. Different models corresponding to each class were fitted (trained and tested) to data from the summers of 2015 and 2016 while model validation was performed using summer 2017 data. Using the coefficient of determination (R2) and the root-mean-square error (RMSE), it was concluded that an RF model with 100 decision trees had the best performance (R2 of 0.81) and the lowest normalized mean error (NME < 1%) leading to low model uncertainties. The relative feature importance for the RF model was calculated to be 0.55, 0.2, 0.15, and 0.1 for WSOM concentrations, RH levels, isoprene concentrations, and NOx/isoprene ratios, respectively. The machine learning model was thus used to predict summertime concentrations of evaporated organics in Yorkville, Georgia, and Centerville, Alabama in 2016 and 2013, respectively. Results presented herein have implications for measurements that rely on sample drying using a machine learning approach for the analysis and interpretation of atmospheric data sets to elucidate their complex behavior.

Generating High Spatial Resolution Exposure Estimates from Sparse Regulatory Monitoring Data

Random Forest algorithms have extensively been used to estimate ambient air pollutant concentrations. However, the accuracy of model-predicted estimates can suffer from extrapolation problems associated with limited measurement data to train the machine learning algorithms. In this study, we developed and evaluated two approaches, incorporating low-cost sensor data, that enhanced the extrapolating ability of random-forest models in areas with sparse monitoring data. Rochester, NY is the area of a pregnancy-cohort study. Daily PM2.5 concentrations from the NAMS/SLAMS sites were obtained and used as the response variable in the model, with satellite data, meteorological, and land-use variables included as predictors. To improve the base random-forest models, we used PM2.5 measurements from a pre-existing low-cost sensors network, and then conducted a two-step backward selection to gradually eliminate variables with potential emission heterogeneity from the base models. We then introduced the regression-enhanced random forest method into the model development. Finally, contemporaneous urinary 1-hydroxypyrene was used to evaluate the PM2.5 predictions generated from the two approaches. The two-step approach increased the average external validation R2 from 0.49 to 0.65, and decreased the RMSE from 3.56 μg/m3 to 2.96 μg/m3. For the regression-enhanced random forest models, the average R2 of the external validation was 0.54, and the RMSE was 3.40 μg/m3. We also observed significant and comparable relationships between urinary 1-hydroxypyrene levels and PM2.5 predictions from both improved models. This PM2.5 model estimation strategy could improve the extrapolating ability of random forest models in areas with sparse monitoring data.

Prediction and model evaluation for space-time data

Evaluation metrics for prediction error, model selection and model averaging on space-time data are understudied and poorly understood. The absence of independent replication makes prediction ambiguous as a concept and renders evaluation procedures developed for independent data inappropriate for most space-time prediction problems. Motivated by air pollution data collected during California wildfires in 2008, this manuscript attempts a formalization of the true prediction error associated with spatial interpolation. We investigate a variety of cross-validation (CV) procedures employing both simulations and case studies to provide insight into the nature of the estimand targeted by alternative data partition strategies. Consistent with recent best practice, we find that location-based cross-validation is appropriate for estimating spatial interpolation error as in our analysis of the California wildfire data. Interestingly, commonly held notions of bias-variance trade-off of CV fold size do not trivially apply to dependent data, and we recommend leave-one-location-out (LOLO) CV as the preferred prediction error metric for spatial interpolation.

Application of satellite remote sensing data and random forest approach to estimate ground-level PM2.5 concentration in Northern region of Thailand

Numerous epidemiological studies have shown that particulate matter with aerodynamic diameter up to 2.5 μm (PM2.5) is associated with many health consequences, where PM2.5 concentration obtained from the monitoring station was normally applied as the exposure level, so that the concentration of PM2.5 in unmonitored areas has not been captured. The satellite-derived aerosol optical depth (AOD) product is then used to spatially predict ground truth of PM2.5 concentration that covers the locations with no air quality monitoring station, but this method has seldom been developed in Thailand. This study aimed at estimating ground-level PM2.5 concentration at 3 km × 3 km spatial resolution over Northern region of Thailand in 2021 using the random forest model integrating the Moderate Resolution Imaging Spectroradiometer (MODIS) AOD products from Terra and Aqua satellites, meteorological factors, and land use data. A random forest model contained 100 decision trees was utilized to train the model, and 10-fold cross-validation approach was implemented to validate the model performance. The good consistency between actual (observed) and predicted concentrations of PM2.5 in Northern region of Thailand was observed, where a coefficient of determination (R2) and root mean square error (RMSE) of the model fitting were 0.803 and 14.30 μg/m3, respectively, and those of 10-fold cross-validation approach were 0.796 and 14.64 μg/m3, respectively. The three most important predictors for estimating the ground-level concentrations of PM2.5 in this study were normalized difference vegetation index (NDVI), relative humidity, and number of fire hotspot, respectively. Findings from this study revealed that integrating the MODIS AOD, meteorological variables, and land use data into the random forest model precisely and accurately estimated ground-level PM2.5 concentration over Northern region of Thailand that can be further used to investigate the effects of PM2.5 exposure on health consequences, even in unmonitored locations, in epidemiological studies.

The application of machine learning to air pollution research: A bibliometric analysis

Machine learning (ML) is an advanced computer algorithm that simulates the human learning process to solve problems. With an explosion of monitoring data and the increasing demand for fast and accurate prediction, ML models have been rapidly developed and applied in air pollution research. In order to explore the status of ML applications in air pollution research, a bibliometric analysis was made based on 2962 articles published from 1990 to 2021. The number of publications increased sharply after 2017, comprising approximately 75% of the total. Institutions in China and United States contributed half of all publications with most research being conducted by individual groups rather than global collaborations. Cluster analysis revealed four main research topics for the application of ML: chemical characterization of pollutants, short-term forecasting, detection improvement and optimizing emission control. The rapid development of ML algorithms has increased the capability to explore the chemical characteristics of multiple pollutants, analyze chemical reactions and their driving factors, and simulate scenarios. Combined with multi-field data, ML models are a powerful tool for analyzing atmospheric chemical processes and evaluating the management of air quality and deserve greater attention in future.

Deep learning mapping of surface MDA8 ozone: The impact of predictor variables on ozone levels over the contiguous United States

The limited number of ozone monitoring stations imposes uncertainty in various applications, calling for accurate approaches to capturing ozone values in all regions, particularly those with no in-situ measurements. This study uses deep learning (DL) to accurately estimate daily maximum 8-hr average (MDA8) ozone and examines the spatial contribution of several factors on ozone levels over the contiguous U.S. (CONUS) in 2019. A comparison between in-situ observations and DL-estimated MDA8 ozone values shows a correlation coefficient (R) of 0.95, an index of agreement (IOA) of 0.97, and a mean absolute bias (MAB) of 2.79 ppb, highlighting the promising performance of the deep convolutional neural network (Deep-CNN) at estimating surface MDA8 ozone. Spatial cross-validation also confirms the high spatial accuracy of the model, which obtains an R of 0.91, and IOA of 0.96 and an MAB of 3.46 ppb when it is trained and tested on separate stations. To interpret the black-box nature of our DL model, we use Shapley additive explanations (SHAP) to generate a spatial feature contribution map (SFCM), the results of which confirm an advanced ability of Deep-CNN to capture the interactions between most predictor variables and ozone. For instance, the model shows that solar radiation (SRad) SFCM, with higher values, enhances the formation of ozone, particularly in the south and southwestern CONUS. As SRad triggers ozone precursors to produce ozone via photochemical reactions, it increases ozone concentrations. The model also shows that humidity, with its low values, increases ozone concentrations in the western mountainous regions. The negative correlation between humidity and ozone levels can be attributed to factors such as higher ozone decomposition resulting from increased levels of humidity and OH radicals. This study is the first to introduce the SFCM to investigate the spatial role of predictor variables on changes in estimated MDA8 ozone levels.

Spatiotemporally continuous estimates of daily 1-km PM2.5 concentrations and their long-term exposure in China from 2000 to 2020

Monitoring long-term variations in fine particulate matter (PM_2.5) is essential for environmental management and epidemiological studies. While satellite-based statistical/machine-learning methods can be used for estimating high-resolution ground-level PM_2.5 concentration data, their applications have been hindered by limited accuracy in daily estimates during years without PM_2.5 measurements and massive missing values due to satellite retrieval data. To address these issues, we developed a new spatiotemporal high-resolution PM_2.5 hindcast modeling framework to generate the full-coverage, daily, 1-km PM_2.5 data for China for the period 2000-2020 with improved accuracy. Our modeling framework incorporated information on changes in observation variables between periods with and without monitoring data and filled gaps in PM_2.5 estimates induced by satellite data using imputed high-resolution aerosol data. Compared to previous hindcast studies, our method achieved superior overall cross-validation (CV) R² and root-mean-square error (RMSE) of 0.90 and 12.94 μg/m³ and significantly improved the model performance in years without PM_2.5 measurements, raising the leave-one-year-out CV R² [RMSE] to 0.83 [12.10 μg/m³] at a monthly scale (0.65 [23.29 μg/m³] at a daily scale). Our long-term PM_2.5 estimates show a sharp decline in PM_2.5 exposure in recent years, but the national exposure level in 2020 still exceeded the first annual interim target of the 2021 World Health Organization air quality guidelines. The proposed hindcast framework represents a new strategy to improve air quality hindcast modeling and can be applied to other regions with limited air quality monitoring periods. These high-quality estimates can support both long- and short-term scientific research and environmental management of PM_2.5 in China.

Forecasting and Optimizing Dual Media Filter Performance via Machine Learning

Four different machine learning algorithms, including Decision Tree (DT), Random Forest (RF), Multivariable Linear Regression (MLR), Support Vector Regressions (SVR), and Gaussian Process Regressions (GPR), were applied to predict the performance of a multi-media filter operating as a function of raw water quality and plant operating variables. The models were trained using data collected over a seven year period covering water quality and operating variables, including true colour, turbidity, plant flow, and chemical dose for chlorine, KMnO₄, FeCl₃, and Cationic Polymer (PolyDADMAC). The machine learning algorithms have shown that the best prediction is at a 1-day time lag between input variables and unit filter run volume (UFRV). Furthermore, the RF algorithm with grid search using the input metrics mentioned above with a 1-day time lag has provided the highest reliability in predicting UFRV with a RMSE and R² of 31.58 and 0.98, respectively. Similarly, RF with grid search has shown the shortest training time, prediction accuracy, and forecasting events using a ROC-AUC curve analysis (AUC over 0.8) in extreme wet weather events. Therefore, Random Forest with grid search and a 1-day time lag is an effective and robust machine learning algorithm that can predict the filter performance to aid water treatment operators in their decision makings by providing real-time warning of the potential turbidity breakthrough from the filters.

7Be, 210Pb, airborne particulate matter and PM10 concentrations in relation to meteorological conditions in southern Poland in 1998-2016

An analysis of the concentration of ⁷Be in aerosol samples collected in one of the most polluted areas in Europe (Katowice and Krakow in southern Poland) indicated seasonal variability, with a maximum in the summer months. The average concentrations of ⁷Be were 4616.1 μBq m^-3 in Katowice and 3259.4 μBq m^-3 in Krakow, respectively, and they are among the highest values recorded in Poland in the studied period (1998-2016). These cities are also characterised by Poland's highest concentrations of ²¹⁰Pb (547.8 μBq m^-3 and 513.2 μBq m^-3). The highest radioactive concentrations of ²¹⁰Pb were observed in the winter and autumn, since in the case of these industrial areas, the combustion processes related to heating in the cold season of the year are an additional source of this isotope, next to its natural origin. The airborne particulate matter concentrations at both locations correlate with the concentrations of ²¹⁰Pb. The average values of PM10 concentrations (71.1 μg m^-3 in Krakow to 45.0 μg m^-3 in Katowice), were 2-3 times higher than the average ones recorded in northern Poland. It has been proven that air temperature is the key parameter affecting the transport of isotopes, especially in the warm season of the year, when its increase causes increased thermal convection, leading to intense vertical mixing and exchange in the troposphere. Analyses using the machine learning method allowed for an indication of the correlation between relative humidity and atmospheric precipitation, as well as higher wind speed and concentrations of ⁷Be which is inversely proportional. Geographical factors (the latitude of the station and the land elevation) have no impact on near-surface concentrations of ⁷Be in Poland.

Algal community structure prediction by machine learning

The algal community structure is vital for aquatic management. However, the complicated environmental and biological processes make modeling challenging. To cope with this difficulty, we investigated using random forests (RF) to predict phytoplankton community shifting based on multi-source environmental factors (including physicochemical, hydrological, and meteorological variables). The RF models robustly predicted the algal communities composed by 13 major classes (Bray-Curtis dissimilarity = 9.2 ± 7.0%, validation NRMSE mostly <10%), with accurate simulations to the total biomass (validation R² > 0.74) in Norway's largest lake, Lake Mjosa. The importance analysis showed that the hydro-meteorological variables (Standardized MSE and Node Purity mostly >0.5) were the most influential factors in regulating the phytoplankton. Furthermore, an in-depth ecological interpretation uncovered the interactive stress-response effect on the algal community learned by the RF models. The interpretation results disclosed that the environmental drivers (i.e., temperature, lake inflow, and nutrients) can jointly pose strong influence on the algal community shifts. This study highlighted the power of machine learning in predicting complex algal community structures and provided insights into the model interpretability.

Scientometric and multidimensional contents analysis of PM2.5 concentration prediction

The foundation for the environmental department to take suitable measures and make a significant contribution towards improving air quality is the precise and dependable prediction of PM_2.5 concentration. It is essential to review the development process and hotspots of PM_2.5 concentration prediction studies over the past 20 years (2000-2021) comprehensively and quantitatively. This study used detailed bibliometric methods and CiteSpace software to visually analyze the PM_2.5 pollution level. The outcomes found that the prediction research phases of PM_2.5 can be broadly divided into three phases and enter the rapid growth phase after 2017. Five categories of keywords are clustered, and the forecasting data and forecasting methods are typical cluster representatives. Then, the construction and processing of PM_2.5 concentration prediction datasets, the prediction methods and technical processes, and the determination of the prediction spatial-temporal scales are the main content of the analysis. In the future, it is necessary to concentrate on multi-source data fusion for PM_2.5 concentration prediction at multiple spatial-temporal scales and focus on technology integration and innovative applications in forecasting models, especially the optimal use of deep machine learning methods to improve prediction accuracy and practical application conversion.

Accounting for the aerosol type and additional satellite-borne aerosol products improves the prediction of PM2.5 concentrations

Fine airborne particles (diameter <2.5 μm; PM_2.5) are recognized as a major threat to human health due to their physicochemical properties: composition, size, shape, etc. However, normally only size-fraction-specific particle concentrations are monitored. Interestingly, although the aerosol type is reported as part of the aerosol optical depth retrieval from satellite observations, it has not been utilized, to date, as an auxiliary information/co-variate for PM_2.5 prediction. We developed Random Forest (RF) and eXtreme Gradient Boosting (XGBoost) models that account for this information when predicting surface PM_2.5. The models take as input only widely available data: satellite aerosol products with full cover and surface meteorological data. Distinct models were developed for AOD of specific aerosol types. Both the RF and XGBoost models performed well, showing moderate-to-high cross-validated adjusted R² (RF: 0.753-0.909; XGBoost: 0.741-0.903), depending on the aerosol type and other covariates. The weighted performance of the specific aerosol-type models was higher than of the RF and XGBoost baseline models, where all the AOD retrievals were used together (the common practice). Our approach can provide improved risk estimates due to exposure to PM_2.5, better resolved radiative forcing calculations, and tailored abatement surveillance of specific pollutants/sources.

Graph Neural Tree: A novel and interpretable deep learning-based framework for accurate molecular property predictions

Determining various properties of molecules is a critical step in drug discovery. Recently, with the improvement of large heterogeneous datasets and the development of deep learning approaches, more and more scientists have turned their attention to neural network-based virtual preliminary screening to reduce the time and monetary cost of drug discovery. However, the poor interpretability of deep learning masks causality, so models' conclusions are often beyond the comprehension of human users, which reduces the credibility of the model and makes it difficult for chemists to further narrow the huge chemical space based on models' results. Thus, this study develops a novel framework consisting of Graph Neural Networks for feature extraction, Curriculum-Based Learning Strategies for optimization, and a Learning Binary Neural Tree (LBNT) for prediction, to improve the performance of neural networks and reveal their decision-making process to chemists. The framework encodes molecular graph data with graph neural networks (GNNs), then retrains the encoder with curriculum-based learning strategies to reduce uncertainty and improve accuracy, and finally uses LBNT as the predictor, which joint retrains with the encoder after independently training, for prediction and visualization. The framework is validated on the public datasets and compared to single GNNs with normal training strategies as well as GNN encoders with common machine learning predictors instead of the LBNT predictor. The result reveals that the proposed framework enhances the point prediction accuracy of the completely trained GNN and reduces its uncertainty through curriculum-based learning, and further improves the accuracy by combining LBNT. Besides, compared with common machine learning tools, the LBNT predictor generally has the best performance because of joint retraining with the GNN encoder. The decision-making process of LBNT is also better and easier to explain than that of other models.

Investigating the long-term trends in aerosol optical depth and its association with meteorological parameters and enhanced vegetation index over Turkey

Aerosol optical depth (AOD) provides useful information on particulate matter pollution at both regional and global levels. In this study, the long-term datasets of aerosols, meteorological parameters, and enhanced vegetation index (EVI) were used from September 2002 to December 2021 over Turkey. This study examined the spatiotemporal distribution of aerosols and their association with meteorological parameters (temperature (Temp), relative humidity (RH), wind speed (WS)), and EVI over Turkey from 2002 to 2021. Moreover, this study also performed a comparison of AOD retrieved from Aqua with other satellites (Terra, SeaWiFS, and MISR) and ground-based (AERONET) products. The higher mean seasonal AOD (> 0.3) was observed over Southeastern Anatolia Region due to the dust transport from the Saharan Desert and Arabian Peninsula. Moreover, AOD was positively correlated with Temp and WS in the east of Turkey, while negative correlations were observed in the coastal regions. The correlation between AOD and RH was also observed negative in most parts of Turkey. Furthermore, in the coastal region, the correlation between AOD and EVI was found to be positive, whereas a negative correlation was seen over less vegetative areas. The multi-seasonal AOD averages were calculated as 0.187, 0.183, 0.138, and 0.104 for the spring, summer, autumn, and winter seasons, respectively. The most important result of this study is the regional differences in AOD over Turkey. For new studies, AOD should be observed separately for coastal areas and the eastern part of Turkey.

Modeling spatiotemporal carbon emissions for two mega-urban regions in China using urban form and panel data analysis

Spatiotemporal monitoring of urban CO₂ emissions is crucial for developing strategies and actions to mitigate climate change. However, most spatiotemporal inventories do not adopt urban form data and have a coarse resolution of over 1 km, which limits their implications in intra-city planning. This study aims to model the spatiotemporal carbon emissions of the two largest mega-urban regions in China, the Yangtze River Delta and the Pearl River Delta, using urban form data from the Local Climate Zone scheme and landscape metrics, nighttime light images, and a year-fixed effects model at a fine resolution from 2012 to 2016. The panel data model has an R² value of 0.98. This study identifies an overall fall in carbon emissions in both regions since 2012 and a slight elevation of emissions from 2015 to 2016. In addition, urban compaction and integrated natural landscapes are found to be related to low emissions, whereas scattered low-rise buildings are associated with rising carbon emissions. Furthermore, this study more accurately extracts urban areas and can more clearly identify intra-urban variations in carbon emissions than other datasets. The open data supported methodology, regression models, and results can provide accurate and quantifiable evidence at the community level for achieving a carbon-neutral built environment.

Exploring the detailed spatiotemporal characteristics of PM2.5: Generating a full-coverage and hourly PM2.5 dataset in the Sichuan Basin, China

Fine particulate matter (PM_2.5) has received worldwide attention due to its threat to public health. In the Sichuan Basin (SCB), PM_2.5 is causing heavy health burdens due to its high concentrations and population density. Compared with other heavily polluted areas, less effort has been made to generate a full-coverage PM_2.5 dataset of the SCB, in which the detailed PM_2.5 spatiotemporal characteristics remain unclear. Considering commonly existing spatiotemporal autocorrelations, the top-of-atmosphere reflectance (TOAR) with a high coverage rate and other auxiliary data were employed to build commonly used random forest (RF) models to generate accurate hourly PM_2.5 concentration predictions with a 0.05° × 0.05° spatial resolution in the SCB in 2016. Specifically, with historical concentrations predicted from a spatial RF (S-RF) and observed at stations, an alternative spatiotemporal RF (AST-RF) and spatiotemporal RF (ST-RF) were built in grids with stations (type 1). The predictions from the AST-RF in grids without stations (type 2) and observations in type 1 formed the PM_2.5 dataset. The LOOCV R², RMSE and MAE were 0.94/0.94, 8.71/8.62 μg∕m³ and 5.58/5.57 μg∕m³ in the AST-RF/ST-RF, respectively. Using the produced dataset, spatiotemporal analysis was conducted for a detailed understanding of the spatiotemporal characteristics of PM_2.5 in the SCB. The PM_2.5 concentrations gradually increased from the edge to the center of the SCB in spatial distribution. Two high-concentration areas centered on Chengdu and Zigong were observed throughout the year, while another high-concentration area centered on Dazhou was only observed in winter. The diurnal variation had double peaks and double valleys in the SCB. The concentrations were high at night and low in daytime, which suggests that characterizing the relationship between PM_2.5 and adverse health outcomes by daily means might be inaccurate with most human activities conducted in daytime.

Nowcasting Applications of Geostationary Satellite Hourly Surface PM2.5 Data

The mass concentration of fine particulate matter (PM2.5; diameters less than 2.5 μm) estimated from geostationary satellite aerosol optical depth (AOD) data can supplement the network of ground monitors with high temporal (hourly) resolution. Estimates of PM2.5 over the United States (US) were derived from NOAA's operational geostationary satellites Advanced Baseline Imager (ABI) AOD data using a geographically weighted regression with hourly and daily temporal resolution. Validation versus ground observations shows a mean bias of -21.4% and -15.3% for hourly and daily PM2.5 estimates, respectively, for concentrations ranging from 0 to 1000 μg/m3. Because satellites only observe AOD in the daytime, the relation between observed daytime PM2.5 and daily mean PM2.5 was evaluated using ground measurements; PM2.5 estimated from ABI AODs were also examined to study this relationship. The ground measurements show that daytime mean PM2.5 has good correlation (r > 0.8) with daily mean PM2.5 in most areas of the US, but with pronounced differences in the western US due to temporal variations caused by wildfire smoke; the relation between the daytime and daily PM2.5 estimated from the ABI AODs has a similar pattern. While daily or daytime estimated PM2.5 provides exposure information in the context of the PM2.5 standard (> 35 μg/m3), the hourly estimates of PM2.5 used in Nowcasting show promise for alerts and warnings of harmful air quality. The geostationary satellite based PM2.5 estimates inform the public of harmful air quality ten times more than standard ground observations (1.8 vs. 0.17 million people per hour).

Impacts of traffic-related particulate matter pollution on semen quality: A retrospective cohort study relying on the random forest model in a megacity of South China

BACKGROUND

Emerging evidence shows the detrimental impacts of particulate matter (PM) on poor semen quality. High-resolution estimates of PM concentrations are conducive to evaluating accurate associations between traffic-related PM exposure and semen quality.

METHODS

In this study, we firstly developed a random forest model incorporating meteorological factors, land-use information, traffic-related variables, and other spatiotemporal predictors to estimate daily traffic-related PM concentrations, including PM_2.5, PM₁₀, and PM₁. Then we enrolled 1310 semen donors corresponding to 4912 semen samples during the study period from January 1, 2019, and December 31, 2019 in Guangzhou city, China. Linear mixed models were employed to associate individual exposures to traffic-related PM during the entire (0-90 lag days) and key periods (0-37 and 34-77 lag days) with semen quality parameters, including sperm concentration, sperm count, progressive motility and total motility.

RESULTS

The results showed that decreased sperm concentration was associated with PM₁₀ exposures (β: -0.21, 95 % CI: -0.35, -0.07), sperm count was inversely related to both PM_2.5 (β: -0.19, 95 % CI: -0.35, -0.02) and PM₁₀ (β: -0.19, 95 % CI: -0.33, -0.05) during the 0-90 days lag exposure window. Besides, PM_2.5 and PM₁₀ might diminish sperm concentration by mainly affecting the late phase of sperm development (0-37 lag days). Stratified analyses suggested that PBF and drinking seemed to modify the associations between PM exposure and sperm motility. We did not observe any significant associations of PM₁ exposures with semen parameters.

CONCLUSION

Our results indicate that exposure to traffic-related PM_2.5 and PM₁₀ pollution throughout spermatogenesis may adversely affect semen quality, especially sperm concentration and count. The findings provided more evidence for the negative associations between traffic-related PM exposure and semen quality, highlighting the necessity to reduce ambient air pollution through environmental policy.

Natural and human factors influencing urban particulate matter concentrations in central heating areas with long-term wearable monitoring devices

In northern China, central heating, as an important source of urban particulate matter (UPM), causes more than half of the air pollution during the heating season and has significant spatial-temporal heterogeneity. Owing to the limitations of stationary air monitoring networks, few studies distinguish between heating/non-heating seasons and few have been conducted in urban areas. However, fixed monitoring cannot accurately capture the dynamic exposure of residents to UPM, and there is a lack of comprehensive evaluation of the factors affecting UPM. Therefore, this study used wearable Sniffer 4D equipment to monitor the concentrations of UPM (PM₁, PM_2.5, and PM₁₀) in selected typical areas of Shenyang City from March 2019 to February 2020. A random forest model was combined with land use and point-of-interest data to analyze the contributions and marginal effects of multiple influences on UPM, in both heating and non-heating seasons. The results showed that in the eastern part of the study area, UPM showed completely opposite spatial distribution characteristics during the two seasons. The concentrations of UPM were higher during the heating season than during the non-heating season. The results indicated that temperature and humidity were important factors in diffusing UPM. The production and operation of boilers were important for the production of UPM. In two-dimensional landscape pattern indices, the percentage of forest and Shannon diversity index were the first and second most important factors, respectively. The three-dimensional pattern of buildings had important effects on the transport and diffusion of UPM (landscape height range >100, floor area ratio >1.3, and landscape volume density >5). Wearable devices could monitor the real situation of residents' exposure to UPM and quantify the factors influencing the spatial-temporal distribution of UPM in an ecological sense. These results provide a scientific basis for urban planning and for health risk reduction for residents.

Data-driven predictive modeling of PM2.5 concentrations using machine learning and deep learning techniques: a case study of Delhi, India

The present study intends to use machine learning (ML) and deep learning (DL) models to forecast PM_2.5 concentration at a location in Delhi. For this purpose, multi-layer feed-forward neural network (MLFFNN), support vector machine (SVM), random forest (RF) and long short-term memory networks (LSTM) have been applied. The air pollutants, e.g., CO, Ozone, PM₁₀, NO, NO₂, NO_x, NH₃, SO₂, benzene, toluene, as well as meteorological parameters (temperature, wind speed, wind direction, rainfall, evaporation, humidity, pressure, etc.), have been used as inputs in the present study. Moreover, this is one of the first papers that employ aerodynamic roughness coefficient as an input parameter for the prediction of PM_2.5 concentration. The result of the study shows that the LSTM model with index of agreement (IA) 0.986, root mean square error (RMSE) 21.510, Nash-Sutcliffe efficiency index (NSE) 0.945, (coefficient of determination)R² 0.945, and (correlation coefficient)R 0.972 is the best performing technique for the prediction of PM_2.5 followed by MLFFNN, SVM, and RF models. The sensitivity analysis for the LSTM model reported that PM₁₀, wind speed, NH₃, and benzene are the key influencing parameters for the estimation of PM_2.5. The findings in this work suggest that the LSTM could advance in PM_2.5 forecasting and thus would be useful for developing fine-scale, state-of-the-art air pollution forecasting models.