Development of a model for particulate matter pollution in Australia with implications for other satellite-based models

Climatology and landscape determinants of AOD, SO2 and NO2 over Indo-Gangetic Plain

Indo-Gangetic Plains (IGP) experiences high loading of particulate and gaseous pollutants all year around and is considered to be the most polluted regions of India. Understanding the effect of landscape determinants on air pollution in IGP regions is crucial to make its environment sustainable. We examined satellite retrievals of OMI NO₂ and SO₂, and MODIS AOD to analyse the long-term trend, spatio-seasonal pattern and dynamics of aerosols, NO₂ and SO₂ over three IGP regions, namely Upper Indo-Gangetic plain (UIGP), Middle Indo-Gangetic plain (MIGP) and Lower Indo-Gangetic plain (LIGP) over the period 2005-2019. IGP experienced an overall increment in AOD (R² = 0.63) and SO₂ (R² = 0.67) values, with LIGP (AOD, R² = 0.8 & SO₂, R² = 0.8) experiencing the largest rate of enhancement. The levels of NO₂ (R² = 0.2) experienced a decrement after 2012 (owing to implementation of vehicle emission policy) except in MIGP, with UIGP (R² = 0.23) exhibiting the largest rate of decrement. Seasonal heterogeneity in the nature of sources was observed over IGP regions. AOD (0.61 ± 0.1) and NO₂ value (3.82 ± 0.98 × 10¹⁵ molecules/cm²) were found highest during post-monsoon in UIGP owing to crop residue burning activity. The value of NO₂ (3.8 ± 1.4 × 10¹⁵ molecules/cm²) in MIGP was found highest during pre-monsoon due to high consumption of coal in power plants for summer cooling demand. The highest SO₂ level (0.09 ± 0.06 DU) was observed during post-monsoon in UIGP, as a large number of brick kilns are fired during this period. Correlations among landscape determinants and pollutants revealed that topography is the dominant variable that affect the spatial pattern of AOD compared to vegetation and land use. Lower elevation tends to have high AOD values compared to higher elevation. Vegetation-AOD relationship showed an inverse association in IGP regions and is influenced by factors such as seasonal meteorology and size of the airborne particles. Vegetation possesses positive relationship with SO₂ and NO₂, implying no pollution abatement effect on SO₂ and NO₂ pollutants. Built-up change has deteriorating effect as well as quenching effect on pollutants. Increase in built terrain have deteriorated the air quality in UIGP whereas it favored in suppressing the aerosol level in LIGP.

Deep Ensemble Machine Learning Framework for the Estimation of PM2.5 Concentrations

BACKGROUND

Accurate estimation of historical PM2.5 (particle matter with an aerodynamic diameter of less than 2.5μm) is critical and essential for environmental health risk assessment.

OBJECTIVES

The aim of this study was to develop a multiple-level stacked ensemble machine learning framework for improving the estimation of the daily ground-level PM2.5 concentrations.

METHODS

An innovative deep ensemble machine learning framework (DEML) was developed to estimate the daily PM2.5 concentrations. The framework has a three-stage structure: At the first stage, four base models [gradient boosting machine (GBM), support vector machine (SVM), random forest (RF), and eXtreme gradient boosting (XGBoost)] were used to generate a new data set of PM2.5 concentrations for training the next-stage learners. At the second stage, three meta-models [RF, XGBoost, and Generalized Linear Model (GLM)] were used to estimate PM2.5 concentrations using a combination of the original data set and the predictions from the first-stage models. At the third stage, a nonnegative least squares (NNLS) algorithm was employed to obtain the optimal weights for PM2.5 estimation. We took the data from 133 monitoring stations in Italy as an example to implement the DEML to predict daily PM2.5 at each 1km×1km grid cell from 2015 to 2019 across Italy. We evaluated the model performance by performing 10-fold cross-validation (CV) and compared it with five benchmark algorithms [GBM, SVM, RF, XGBoost, and Super Learner (SL)].

RESULTS

The results revealed that the PM2.5 prediction performance of DEML [coefficients of determination (R2)=0.87 and root mean square error (RMSE)=5.38μg/m3] was superior to any benchmark models (with R2 of 0.51, 0.76, 0.83, 0.70, and 0.83 for GBM, SVM, RF, XGBoost, and SL approach, respectively). DEML displayed reliable performance in capturing the spatiotemporal variations of PM2.5 in Italy.

DISCUSSION

The proposed DEML framework achieved an outstanding performance in PM2.5 estimation, which could be used as a tool for more accurate environmental exposure assessment. https://doi.org/10.1289/EHP9752.

The comparison of AOD-based and non-AOD prediction models for daily PM2.5 estimation in Guangdong province, China with poor AOD coverage

The large amount of missing values has challenged the application of satellite-retrieved aerosol optical depth (AOD) in mapping surface PM_2.5 concentrations. In this study, we developed a non-AOD random forest model to estimate daily concentrations of PM_2.5 in Guangdong Province, China, where more than 80% of AOD data were missing. The predictive ability of the non-AOD model was compared with that of a AOD-based model. Daily ground-based measurements of PM_2.5 were obtained from 148 ground-monitoring sites in Guangdong with a 60 km rectangle buffer from January 2016 to December 2018. Daily MODIS MAIAC AOD were downloaded from NASA at a resolution of approximately 1 km. Two random forest models were developed to predict ground-level PM_2.5 with one included AOD as a predictor and the other one without AOD. The two random forest models were developed using the same dataset and their predictive abilities were compared. The results of 10-fold cross validation (CV) showed that the non-AOD and AOD-based random forest models yielded similar performance. The CV R² (RMSE) for the non-AOD model in 2016-2018 were 0.82 (8.4 μg/m³), 0.81 (9.5 μg/m³) and 0.78 (9.4 μg/m³), and those for AOD-based model were 0.83 (8.2 μg/m³), 0.82 (9.2 μg/m³) and 0.80 (9.0 μg/m³), respectively. Higher consistency of estimated PM_2.5 were observed in areas close to monitoring sites than those far away from sites, and in southern coastal than northern areas. As the non-AOD random forest model is not affected by AOD missingness, it can be used for epidemiological studies to estimate individual-level exposure to PM_2.5 at a high resolution without spatial or temporal gaps.

A novel method for identifying hotspots and forecasting air quality through an adaptive utilization of spatio-temporal information of multiple factors

Air pollution exerts serious impacts on human health and sustainable development. The accurate forecasting of air quality can guide the formulation of mitigation strategies and reduce exposure to air pollution. It is beneficial to explicitly consider both spatial and temporal information of multiple factors, e.g., the meteorological data, in the forecasting of air pollutant concentrations. The temporal information of relevant factors collected at a location should be considered for forecasting. In addition, these factors recorded at other locations may also provide useful information. Existing methods utilizing the spatio-temporal information of these relevant factors are usually based on some very complicated frameworks. In this study, we propose a novel and simple spatial attention-based long short-term memory (SA-LSTM) that combines LSTM and a spatial attention mechanism to adaptively utilize the spatio-temporal information of multiple factors for forecasting air pollutant concentrations. Specifically, the SA-LSTM employs gated recurrent connections to extract temporal information of multiple factors at individual locations, and the spatial attention mechanism to spatially fuse the temporal information extracted at these locations. This method is effective and applicable to forecast any air pollutant concentrations when spatio-temporal information of relevant factors has to be utilized. To validate the effectiveness of the proposed SA-LSTM, we apply it to forecast the daily air quality in Hong Kong, a high density city with peculiar cityscapes, by using the air quality and meteorological data. Empirical results demonstrate that the proposed SA-LSTM outperforms the conventional models with respect to one-day forecast accuracy, especially for extreme values. Moreover, the attention weights learned by the SA-LSTM can identify hotspots of the air pollution process for reducing computational complexity of forecasting and provide a better understanding of the underlying mechanism of air pollution.

Particulate matter (PM2.5 and PM10) generation map using MODIS Level-1 satellite images and deep neural network

Most studies about particulate matter (PM) estimation have been done based on satellite-derived optical depth aerosol (AOD) products. But, the use of AOD products having coarse resolution is not possible for PM map generation in small spatial coverage such as local cities. To solve this issue, a PM estimation framework is proposed in this work which accepts the original calibrated radiance of MODIS-Level 1 images as input. There are no intermediate computations for atmospheric reflectance or aerosol thickness calculation. A deep neural network consisting of recurrent layers is proposed to extract the relationship between the grey level values of the satellite image bands and the PM measurements in different days and locations. Two individual networks are trained for PM_2.5 and PM₁₀ concentrations. The PM_2.5 map and PM₁₀ map of Tehran city are generated. The performance of the proposed method is compared with several recently published air pollution studies. The results show that the proposed method is a simple, low cost and efficient approach for PM generation of small-scaled coverage using free available Moderate Resolution Imaging Spectroradiometer (MODIS) images.

Assessment of kitchen emissions using a backpropagation neural network model based on urinary hydroxy polycyclic aromatic hydrocarbons

Kitchen emissions are mixed indoor air pollutants with adverse health effects, but the large-scale assessment is limited by costly equipment and survey methods. This study aimed to discuss the application of backpropagation (BP) neural network models in the assessment of kitchen emissions based on the exposure marker. A total of 3686 participants were recruited for the kitchen survey, and their sleep quality was measured by the Pittsburgh sleep quality index (PSQI). After excluding the confounders, 365 participants were selected to assess their urinary hydroxy polycyclic aromatic hydrocarbons (OH-PAHs) concentrations by ultra-high-performance liquid chromatography/tandem mass spectrometry. Two BP neural network models were then set up using the survey and detection data from the 365 participants and used to predict the total urinary OH-PAHs concentrations of all participants. The total urinary OH-PAHs and 1-hydroxy-naphthalene (1-OHNap) concentrations were significantly higher among the 365 participants with poor sleep quality (global PSQI score > 5; P < 0.05). Results from internal and external validation showed that our model has high credibility (model 2). Further, the participants with higher predicted total urinary OH-PAHs concentrations were associated with the global PSQI score of >5 (odds ratio (OR) = 1.284, 95% confidence interval (CI) = 1.082-1.525 for participants with predicted total urinary OH-PAHs concentrations of over 1.897 μg/mmol creatinine in model 1, and OR = 1.467, 95% CI = 1.240-1.735 for participants with predicted total urinary OH-PAHs concentrations of over 2.253 μg/mmol creatinine in model 2) after adjusting for the confounders. Findings suggest that the BP neural network model is suitable for assessing kitchen emissions, and the urinary OH-PAHs concentrations can be taken as the model outlay.

A hybrid air quality early-warning framework: An hourly forecasting model with online sequential extreme learning machines and empirical mode decomposition algorithms

Modelling air quality with a practical tool that produces real-time forecasts to mitigate risk to public health continues to face significant challenges considering the chaotic, non-linear and high dimensional nature of air quality predictor variables. The novelty of this research is to propose a hybrid early-warning artificial intelligence (AI) framework that can emulate hourly air quality variables (i.e., Particulate Matter 2.5, PM_2.5; Particulate Matter 10, PM₁₀ and lower atmospheric visibility, VIS), the atmospheric variables associated with increased respiratory induced mortality and recurrent health-care cost. Firstly, hourly air quality data series (January-2015 to December-2017) are demarcated into their respective intrinsic mode functions (IMFs) and a residual sub-series that reveal patterns and resolve data complexity characteristics, followed by partial autocorrelation function applied to each IMF and residual sub-series to unveil historical changes in air quality. To design the prescribed hybrid model, the data is partitioned into training (70%), validation (15%) and testing (15%) sub-sets. The online sequential-extreme learning machine (OS-ELM) algorithm integrated with improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is designed as a data pre-processing system to robustly extract predictive patterns and fine-tune the model generalization to a near-optimal global solution, which represents modelled air quality at hourly forecast horizons. The resulting early warning AI-based framework denoted as ICEEMDAN-OS-ELM model, is individually constructed by forecasting each IMF and residual sub-series, with hourly PM_2.5, PM₁₀, and VIS obtained by the aggregated sum of forecasted IMFs and residual sub-series. The results are benchmarked with many competing predictive approaches; e.g., hybrid ICEEMDAN-multiple-linear regression (MLR), ICEEMDAN-M5 model tree and standalone versions: OS-ELM, MLR, M5 model tree. Statistical metrics including the root-mean-square error (RMSE), mean absolute error (MAE), Willmott's Index (WI), Legates & McCabe's Index (E_LM) and Nash-Sutcliffe coefficients (E_NS) are used to evaluate the model's accuracy. Both visual and statistical results show that the proposed ICEEMDAN-OS-ELM model registers superior results, outperforming alternative comparison approaches. For instance, for PM_2.5,E_LM values ranged from 0.65-0.82 vs. 0.59-0.77 for ICEEMDAN-M5 tree, 0.59-0.74 for ICEEMDAN-MLR, 0.28-0.54 for OS-ELM, 0.27-0.54 for M5 tree and 0.25-0.53 for the MLR model. For remaining air quality variables (i.e., PM₁₀ & VIS), the objective model (ICEEMDAN-OS-ELM) outperformed the comparative models. In particular, ICEEMDAN-OS-ELM registered relatively low RMSE/MAE, ranging from approximately 0.7-1.03 μg/m³(MAE), 1.01-1.47 μg/m³(RMSE) for PM_2.5 whereas for PM₁₀, these metrics registered a value of 1.29-3.84 μg/m³(MAE), 3.01-7.04 μg/m³(RMSE) and for Visibility, they were 0.01-3.72 μg/m³ (MAE (Mm^-1)), 0.04-5.98 μg/m³ (RMSE (Mm^-1)). Visual analysis of forecasted and observed air quality through a Taylor diagram illustrates the objective model's preciseness, confirming the versatility of early warning AI-model in generating air quality forecasts. The excellent performance ascertains the hybrid model's potential utility for air quality monitoring and subsequent public health risk mitigation.

International Mind, Activities and Urban Places (iMAP) study: methods of a cohort study on environmental and lifestyle influences on brain and cognitive health

INTRODUCTION

Numerous studies have found associations between characteristics of urban environments and risk factors for dementia and cognitive decline, such as physical inactivity and obesity. However, the contribution of urban environments to brain and cognitive health has been seldom examined directly. This cohort study investigates the extent to which and how a wide range of characteristics of urban environments influence brain and cognitive health via lifestyle behaviours in mid-aged and older adults in three cities across three continents.

METHODS AND ANALYSIS

Participants aged 50-79 years and living in preselected areas stratified by walkability, air pollution and socioeconomic status are being recruited in Melbourne (Australia), Barcelona (Spain) and Hong Kong (China) (n=1800 total; 600 per site). Two assessments taken 24 months apart will capture changes in brain and cognitive health. Cognitive function is gauged with a battery of eight standardised tests. Brain health is assessed using MRI scans in a subset of participants. Information on participants' visited locations is collected via an interactive web-based mapping application and smartphone geolocation data. Environmental characteristics of visited locations, including the built and natural environments and their by-products (e.g., air pollution), are assessed using geographical information systems, online environmental audits and self-reports. Data on travel and lifestyle behaviours (e.g., physical and social activities) and participants' characteristics (e.g., sociodemographics) are collected using objective and/or self-report measures.

ETHICS AND DISSEMINATION

The study has been approved by the Human Research Ethics Committee of the Australian Catholic University, the Institutional Review Board of the University of Hong Kong and the Parc de Salut Mar Clinical Research Ethics Committee of the Government of Catalonia. Results will be communicated through standard scientific channels. Methods will be made freely available via a study-dedicated website.

TRIAL REGISTRATION NUMBER

ACTRN12619000817145.

A Clean Air Plan for Sydney: An Overview of the Special Issue on Air Quality in New South Wales

This paper presents a summary of the key findings of the special issue of Atmosphere on Air Quality in New South Wales and discusses the implications of the work for policy makers and individuals. This special edition presents new air quality research in Australia undertaken by (or in association with) the Clean Air and Urban Landscapes hub, which is funded by the National Environmental Science Program on behalf of the Australian Government’s Department of the Environment and Energy. Air pollution in Australian cities is generally low, with typical concentrations of key pollutants at much lower levels than experienced in comparable cities in many other parts of the world. Australian cities do experience occasional exceedances in ozone and PM2.5 (above air pollution guidelines), as well as extreme pollution events, often as a result of bushfires, dust storms, or heatwaves. Even in the absence of extreme events, natural emissions play a significant role in influencing the Australian urban environment, due to the remoteness from large regional anthropogenic emission sources. By studying air quality in Australia, we can gain a greater understanding of the underlying atmospheric chemistry and health risks in less polluted atmospheric environments, and the health benefits of continued reduction in air pollution. These conditions may be representative of future air quality scenarios for parts of the Northern Hemisphere, as legislation and cleaner technologies reduce anthropogenic air pollution in European, American, and Asian cities. However, in many instances, current legislation regarding emissions in Australia is significantly more lax than in other developed countries, making Australia vulnerable to worsening air pollution in association with future population growth. The need to avoid complacency is highlighted by recent epidemiological research, reporting associations between air pollution and adverse health outcomes even at air pollutant concentrations that are lower than Australia’s national air quality standards. Improving air quality is expected to improve health outcomes at any pollution level, with specific benefits projected for reductions in long-term exposure to average PM2.5 concentrations.

Applying linear and nonlinear models for the estimation of particulate matter variability

In this study, data collected from an urban air quality monitoring network are being used for the purpose of evaluating various methodologies used for spatial interpolation in the context of proposing an effective yet simple to apply scheme for PM spatial point estimations. The examined methods are the Inverse Distance Weighting, two linear regression models, the Multiple Linear Regression and the Linear Mixed Model, along with a Feed Forward Neural Network (FFNN) model. These schemes utilize daily PM₁₀ and PM_2.5 concentrations collected from five and three air quality monitoring sites respectively. In order to obtain the resulted estimations, the leave-one-out cross-validation methodology is used for all methods. The evaluation of their predictive ability is performed by using a combination of difference and correlation statistical measures, scatter plots and statistical tests. The results indicate the usefulness of FFNNs as they are found to be statistically significantly superior for modelling the particulate matter spatial variability. The model performance statistics show that in most cases the error values are considerably lower for the FFNN model. Additionally, the rank and Wilcoxon rank tests reveal that the null hypothesis for equal predictive accuracy is rejected for the majority of monitoring sites and schemes (values lower than the critical t-value). According to the comparison results, the FFNN model is selected for forecasting air quality limit exceedances set by the European Union and World Health Organization air quality standards. For two monitoring sites in which the largest number of exceedances occurred, the probability of detection is high while the probability of false detection is very low, further establishing the neural networks' predictive ability.

Satellite-Based Land-Use Regression for Continental-Scale Long-Term Ambient PM2.5 Exposure Assessment in Australia

Australia has relatively diverse sources and low concentrations of ambient fine particulate matter (<2.5 μm, PM_2.5). Few comparable regions are available to evaluate the utility of continental-scale land-use regression (LUR) models including global geophysical estimates of PM_2.5, derived by relating satellite-observed aerosol optical depth to ground-level PM_2.5 ("SAT-PM_2.5"). We aimed to determine the validity of such satellite-based LUR models for PM_2.5 in Australia. We used global SAT-PM_2.5 estimates (∼10 km grid) and local land-use predictors to develop four LUR models for year-2015 (two satellite-based, two nonsatellite-based). We evaluated model performance at 51 independent monitoring sites not used for model development. An LUR model that included the SAT-PM_2.5 predictor variable (and six others) explained the most spatial variability in PM_2.5 (adjusted R² = 0.63, RMSE (μg/m³ [%]): 0.96 [14%]). Performance decreased modestly when evaluated (evaluation R² = 0.52, RMSE: 1.15 [16%]). The evaluation R² of the SAT-PM_2.5 estimate alone was 0.26 (RMSE: 3.97 [56%]). SAT-PM_2.5 estimates improved LUR model performance, while local land-use predictors increased the utility of global SAT-PM_2.5 estimates, including enhanced characterization of within-city gradients. Our findings support the validity of continental-scale satellite-based LUR modeling for PM_2.5 exposure assessment in Australia.