Satellite-based PM2 Research Articles

Satellite-based ground PM2.5 estimation using a gradient boosting decision tree

Fine particulate matter with an aerodynamic diameter less than 2.5 μm (PM2.5) is one of the major air pollutants risks to human health worldwide. Satellite-based aerosol optical depth (AOD) products are an effective metric for acquiring PM2.5 information, featuring broad coverage and high resolution, which compensate for the sparse and uneven distribution of existing monitoring stations. In this study, a gradient boosting decision tree (GBDT) model for estimating ground PM2.5 concentration directly from AOD products across China in 2017, integrating human activities and various natural variables was proposed. The GBDT model performed well in estimating temporal variability and spatial contrasts in daily PM2.5 concentrations, with relatively high fitted model (10-fold cross-validation) coefficients of determination of 0.98 (0.81), low root mean square errors of 3.82 (11.57) μg/m3, and mean absolute error of 1.44 (7.45) μg/m3. Seasonal examinations revealed that summer had the cleanest air with the highest estimation accuracies, whereas winter had the most polluted air with the lowest estimation accuracies. The model successfully captured the PM2.5 distribution pattern across China in 2017, showing high levels in southwest Xinjiang, the North China Plain, and the Sichuan Basin, especially in winter. Compared with other models, the GBDT model showed the highest performance in the estimation of PM2.5 with a 3-km resolution. This algorithm can be adopted to improve the accuracy of PM2.5 estimation with higher spatial resolution, especially in summer. In general, this study provided a potential method of improving the accuracy of satellite-based ground PM2.5 estimation.

Evaluation of gap-filling approaches in satellite-based daily PM2.5 prediction models

Approximately half of satellite aerosol retrievals are missing that limits the application of satellite data in PM2.5 pollution monitoring. To obtain spatiotemporally continuous PM2.5 distributions, various gap-filling methods have been developed, but have rarely been evaluated. Here, we reviewed and summarized four types of gap-filling strategies, and applied them to a random forest PM2.5 prediction model that incorporated ground observations, chemical transport model (CTM) simulations, and satellite AOD for predicting daily PM2.5 concentrations at a 1-km resolution in 2013 in the Beijing-Tianjin-Hebei region and the Yangtze River Delta. The model out-of-bag predictions were compared with national station measurements and external measurements to assess the performance of different gap-filling methods. We also conducted a by-city cross-validation and characterized the spatial distributions of PM2.5 prediction when the AOD coverage was low. We found that the methods filling in missing data by regression, i.e. multiple imputation and decision tree, performed robustly to characterizing PM2.5 variation at a high spatial resolution and the method filling in missing PM2.5 predictions with decision tree overcame the problem of time-consuming computations. The method using spatiotemporal trends to fill in missing data, i.e. ordinary kriging and generalized additive mixed model, may be overrated in statistical evaluation tests, and predicted artificially oversmoothed PM2.5 spatial distributions. We also revealed that CTM simulations benefited the prediction of PM2.5 spatial distribution in all the models with various gap-filling strategies with higher prediction accuracy in the by-city cross-validation. We noticed that the PM2.5 prediction was not sensitive to the resolution of CTM simulations and even the 12-km resolution CTM simulations benefited the high-resolution PM2.5 prediction.

New interpretable deep learning model to monitor real-time PM2.5 concentrations from satellite data

Particulate matter with a mass concentration of particles with a diameter less than 2.5 μm (PM2.5) is a key air quality parameter. A real-time knowledge of PM2.5 is highly valuable for lowering the risk of detrimental impacts on human health. To achieve this goal, we developed a new deep learning model-EntityDenseNet to retrieve ground-level PM2.5 concentrations from Himawari-8, a geostationary satellite providing high temporal resolution data. In contrast to the traditional machine learning model, the new model has the capability to automatically extract PM2.5 spatio-temporal characteristics. Validation across mainland China demonstrates that hourly, daily and monthly PM2.5 retrievals contain the root-mean-square errors of 26.85, 25.3, and 15.34 μg/m3, respectively. In addition to a higher accuracy achievement when compared with various machine learning inversion methods (backpropagation neural network, extreme gradient boosting, light gradient boosting machine, and random forest), EntityDenseNet can “peek inside the black box” to extract the spatio-temporal features of PM2.5. This model can show, for example, that PM2.5 levels in the coastal city of Tianjin were more influenced by air from Hebei than Beijing. Further, EntityDenseNet can still extract the seasonal characteristics that demonstrate that PM2.5 is more closely related within three month groups over mainland China: (1) December, January and February, (2) March, April and May, (3) July, August and September, even without meteorological information. EntityDenseNet has the ability to obtain high temporal resolution satellite-based PM2.5 data over China in real-time. This could act as an important tool to improve our understanding of PM2.5 spatio-temporal features.

Spatial distribution of health risks associated with PM2.5 in Turkey and Iran using satellite and ground observations

PM2.5 originated from anthropogenic or natural sources is still a problem worldwide and highly associated with health impacts. In this study, mortality associated with PM2.5 was calculated using satellite retrievals and ground observations for Turkey and Iran. A high-resolution population dataset along with high resolution satellite-derived PM2.5 estimations (all compositions and dust and sea-salt removed) developed from aerosol optical depth (AOD) data were used. Health risk estimations attributable to PM2.5 from ischemic heart disease and lung cancer along with all causes were calculated by applying the cause-specific mortality rates for Turkey and Iran for the year 2016. According to dust and sea-salt removed satellite-derived PM2.5, mortality estimations of 36967 in Turkey and 34491 in Iran were estimated by all causes. Meanwhile, estimations of 40583 in Turkey and 34153 in Iran were found using ground PM2.5 observations for provinces with monitoring stations. Province-based estimations were obtained with better spatially resolved PM2.5 concentrations. Estimations using ground observations were higher than ones using satellite-derived PM2.5 for Turkey, but similar for Iran, even without national coverage. However, estimations with all composition satellite-based PM2.5 were significantly higher in Iran than Turkey. Values estimated in this study were found to be higher than global studies for both countries.

Long-Term Effects of High Exposure to Ambient Fine Particulate Matter on Coronary Heart Disease Incidence: A Population-Based Chinese Cohort Study.

Evidence of long-term effects of high exposure to ambient fine particulate matter (PM2.5) on coronary heart disease (CHD) remains limited. We incorporated the high-resolution satellite-based PM2.5 estimates with a large-scale, population-based Chinese cohort comprising 118 229 individuals, to assess the CHD risk of long-term exposure to high PM2.5. During the follow-up of 908 376 person-years, 1586 incident CHD caseswere identified. The long-term average PM2.5 concentration for study population was 64.96 μg/m3, ranging from 31.17 to 96.96 μg/m3. For an increment of 10 μg/m3 in PM2.5, the multivariate-adjusted hazard ratios (HRs) were 1.43 (95% confidence interval [CI]: 1.35-1.51) for total CHD, 1.45 (95% CI: 1.36-1.56) for nonfatal CHD, and 1.38 (95% CI: 1.25-1.53) for fatal CHD, respectively. The effects were different across specific CHD outcomes, with greater effects for unstable angina (HR, 1.71 [95% CI, 1.56-1.88]), and weaker effects for acute myocardial infarction (HR, 1.28 [95% CI, 1.19-1.39]) and other CHD (HR, 1.27 [95% CI, 1.10-1.48]). The exposure-response curve suggested that HRs increased with elevated PM2.5 concentration over the entire exposure range. Elderly and hypertensive individuals were more susceptible to PM2.5-induced CHD. Our findings demonstrate the adverse health effects of severe air pollution and highlight the potential health benefits of air quality improvement.

Estimating ground-level PM2.5 using micro-satellite images by a convolutional neural network and random forest approach

Abstract PM2.5 poses a serious threat to public health, however its spatial concentrations are not well characterized due to the sparseness of regulatory air quality monitoring (AQM) stations. This motivates novel low-cost methods to estimate ground-level PM2.5 at a fine spatial resolution so that PM2.5 exposure in epidemiological research can be better quantified. Satellite-retrieved aerosol products are widely used to estimate the spatial distribution of ground-level PM2.5. However, these aerosol products can be subject to large uncertainties due to many approximations and assumptions made in multiple stages of their retrieval algorithms. Therefore, estimating ground-level PM2.5 directly from satellites (e.g. satellite images) by skipping the intermediate step of aerosol retrieval can potentially yield lower errors because it avoids retrieval error propagating into PM2.5 estimation and is desirable compared to current ground-level PM2.5 retrieval methods. Additionally, the spatial resolutions of estimated PM2.5 are usually constrained by those of the aerosol products and are currently largely at a comparatively coarse 1 km or greater resolution. Such coarse spatial resolutions are unable to support scientific studies that thrive on highly spatially-resolved PM2.5. These limitations have motivated us to devise a computer vision algorithm for estimating ground-level PM2.5 at a high spatiotemporal resolution by directly processing the global-coverage, daily, near real-time updated, 3 m/pixel resolution, three-band micro-satellite imagery of spatial coverages significantly smaller than 1 × 1 km (e.g., 200 × 200 m) available from Planet Labs. In this study, we employ a deep convolutional neural network (CNN) to process the imagery by extracting image features that characterize the day-to-day dynamic changes in the built environment and more importantly the image colors related to aerosol loading, and a random forest (RF) regressor to estimate PM2.5 based on the extracted image features along with meteorological conditions. We conducted the experiment on 35 AQM stations in Beijing over a period of ~3 years from 2017 to 2019. We trained our CNN-RF model on 10,400 available daily images of the AQM stations labeled with the corresponding ground-truth PM2.5 and evaluated the model performance on 2622 holdout images. Our model estimates ground-level PM2.5 accurately at a 200 m spatial resolution with a mean absolute error (MAE) as low as 10.1 μg m−3 (equivalent to 23.7% error) and Pearson and Spearman r scores up to 0.91 and 0.90, respectively. Our trained CNN from Beijing is then applied to Shanghai, a similar urban area. By quickly retraining only RF but not CNN on the new Shanghai imagery dataset, our model estimates Shanghai 10 AQM stations' PM2.5 accurately with a MAE and both Pearson and Spearman r scores of 7.7 μg m−3 (18.6% error) and 0.85, respectively. The finest 200 m spatial resolution of ground-level PM2.5 estimates from our model in this study is higher than the vast majority of existing state-of-the-art satellite-based PM2.5 retrieval methods. And our 200 m model's estimation performance is also at the high end of these state-of-the-art methods. Our results highlight the potential of augmenting existing spatial predictors of PM2.5 with high-resolution satellite imagery to enhance the spatial resolution of PM2.5 estimates for a wide range of applications, including pollutant emission hotspot determination, PM2.5 exposure assessment, and fusion of satellite remote sensing and low-cost air quality sensor network information.

Satellite-based spatiotemporal trends of ambient PM2.5 concentrations and influential factors in Hubei, Central China

Accurate estimations of the concentration of ambient fine-particle matter with aerodynamic diameters of less than 2.5 μm (PM2.5) are necessary for human health studies. In this study, individual city-scale linear mixed effect models (LME) were employed to accurately estimate ground PM2.5 concentrations considering the spatiotemporal variability of the relationship between PM2.5 and atmospheric, meteorological, and land observations. The contributions of diverse influential factors including aerosol optical depth, planetary boundary layer height, relative humidity, vegetation index, and wind on local PM2.5 pollution were also determined. High correlation coefficient (R2 = 0.89) and low root mean square error (RMSE = 13.1 μg/m3) ensured satisfactory LME model performances in estimating ground-level PM2.5 concentrations. Spatiotemporal analyses of satellite-based PM2.5 showed high concentrations in eastern, southern, and northern Hubei, and low concentrations in the northwest and southeast because of unbalanced development. These analyses also displayed a mitigation trend of PM2.5 concentrations with a mean annual decline rate of 3–12% from 2016 to 2018. Moreover, from the statistical results of the model, the influential factor of aerosol optical depth was positively correlated with PM2.5 concentration, while planetary boundary layer height, relative humidity, and the normalized difference vegetation index were negatively correlated to local PM2.5 pollution. However, the winds had contradictory contributions on PM2.5 pollution; the northerly wind in western Hubei and the southerly and northeasterly winds in eastern Hubei alleviated local PM2.5 pollution, while the westerly wind in eastern Hubei facilitated PM2.5 diffusion between cities and aggravated PM2.5 pollution. The analysis of the spatiotemporal trend of local PM2.5 pollution at a city scale and the identification of the influence of wind on PM2.5 pollution provide a theoretical reference for regional pollution warnings and controls.

Observation of PM2.5 using a combination of satellite remote sensing and low-cost sensor network in Siberian urban areas with limited reference monitoring

The lack of reference ground-based PM2.5 observation leads to large gaps in air quality information, particularly in many areas of the developing world. This study investigated a new solution for urban air-quality monitoring in regions with limited reference ground-based monitoring. We developed an observation-based method by combining satellite remote-sensing techniques and a newly established low-cost sensor network to estimate long-term PM2.5 concentrations over Krasnoyarsk, a highly industrialized Siberian city. First, a physical model was developed to estimate PM2.5 concentrations using satellite remote-sensing with the aid of ground-based meteorological and radiosonde observations. Observations from the ground-based sensor network were then used to calibrate the deviations in the satellite-derived PM2.5 concentrations. The results show that the satellite-based PM2.5 concentrations obtained by our physical model were in good agreement with the sensor observations (R = 0.78 on the monthly scale). The deviation in satellite-derived annual PM2.5 concentrations resulted from data restrictions that occurred at noon and data loss in winter were identified as 20% and 30%, respectively. The regional transport of smoke from forest wildfires increased PM2.5 concentration to 150 μg/m3 in the summer 2018. The average PM2.5 concentrations in the urban districts could reach 35 μg/m3, which far exceeded the World Health Organization air quality guideline. These results underscore the good ability of our new method to determine PM2.5 concentrations in regions with limited reference ground-based monitoring. Use of sensor and meteorological observations greatly improved satellite detection of PM2.5 concentration. In addition, our method has the potential for global application to improve determination of PM2.5 concentrations, especially in sparsely monitored regions.

Associations of long-term exposure to ambient PM2.5 with mortality in Chinese adults: A pooled analysis of cohorts in the China-PAR project

BackgroundThe concentration-response relationship between mortality and long-term exposure to fine particulate matter (PM2.5) has not been fully elucidated, especially at high levels of PM2.5 concentrations. ObjectiveWe aimed to evaluate chronic effects of ambient PM2.5 exposure on deaths among Chinese adults in high-exposure settings. MethodsParticipants of the Prediction for Atherosclerotic cardiovascular disease Risk in China (China-PAR) project were included from four prospective cohorts among Chinese adults aged ≥18 years old. The overall follow-up rate of the four cohorts was 93.4% until the recent follow-up survey that ended in 2015. The average of satellite-based PM2.5 concentrations during 2000–2015 at 1-km spatial resolution was assigned to each participant according to individual residence addresses. Based on the pooled analysis of individual data from the four cohorts, a Cox proportional hazards model was used to estimate the hazard ratio (HR) and corresponding 95% confidence intervals (95% CIs) for the association of PM2.5 exposure with mortality after multivariate adjustment. ResultsA total of 116,821 participants were eligible in the final analysis. During a mean of 7.7 years of follow-up, 6,395 non-accidental deaths and 2,507 cardio-metabolic deaths occurred. The mean of PM2.5 concentration was 64.9 μg/m3 ranging from 31.2 μg/m3 to 97.0 μg/m3. For each 10 μg/m3 increment in PM2.5, the HR was 1.11 (95% CI: 1.08–1.14) for non-accidental mortality and 1.22 (95% CI: 1.16–1.27) for cardio-metabolic mortality. In addition, a weak exponential curve for the concentration-response association between mortality and PM2.5 was observed among Chinese adults. ConclusionsOur study provided important evidence of the long-term effects of PM2.5 exposure on deaths among Chinese adults. The findings expand our knowledge on concentration-response relationship in high-exposure environments, which is essential to address the urgent challenge of reducing the disease burden attributable to PM2.5 exposure in rapidly industrializing countries such as China.

Spatio-Temporal Variations of Satellite-Based PM2.5 Concentrations and Its Determinants in Xinjiang, Northwest of China.

With the aggravation of air pollution in recent years, a great deal of research on haze episodes is mainly concentrated on the east-central China. However, fine particulate matter (PM2.5) pollution in northwest China has rarely been discussed. To fill this gap, based on the standard deviational ellipse analysis and spatial autocorrelation statistics method, we explored the spatio-temporal variation and aggregation characteristics of PM2.5 concentrations in Xinjiang from 2001 to 2016. The result showed that annual average PM2.5 concentration was high both in the north slope of Tianshan Mountain and the western Tarim Basin. Furthermore, PM2.5 concentrations on the northern slope of the Tianshan Mountain increased significantly, while showing an obviously decrease in the western Tarim Basin during the period of 2001–2016. Based on the result of the geographical detector method (GDM), population density was the most dominant factor of the spatial distribution of PM2.5 concentrations (q = 0.550), followed by road network density (q = 0.423) and GDP density (q = 0.413). During the study period (2001–2016), the driving force of population density on the distribution of PM2.5 concentrations showed a gradual downward trend. However, other determinants, like DEM (Digital elevation model), NSL (Nighttime stable light), LCT (Land cover type), and NDVI (Normalized Difference Vegetation Index), show significant increased trends. Therefore, further effort is required to reveal the role of landform and vegetation in the spatio-temporal variations of PM2.5 concentrations. Moreover, the local government should take effective measures to control urban sprawl while accelerating economic development.

Long-Term Exposure to Fine Particulate Matter and Cardiovascular Disease in China

BackgroundEvidence of the effects of long-term fine particulate matter (PM2.5) exposure on cardiovascular diseases (CVDs) is rare for populations exposed to high levels of PM2.5 in China and in other countries with similarly high levels. ObjectivesThe aim of this study was to assess the CVD risks associated with long-term exposure to PM2.5 in China. MethodsA nationwide cohort study, China-PAR (Prediction for Atherosclerotic Cardiovascular Disease Risk in China), was used, with 116,972 adults without CVD in 2000 being included. Participants were followed until 2015. Satellite-based PM2.5 concentrations at 1-km spatial resolution during the study period were used for exposure assessment. A Cox proportional hazards model with time-varying exposures was used to estimate the CVD risks associated with PM2.5 exposure, adjusting for individual risk factors. ResultsAnnual mean concentrations of PM2.5 at the China-PAR sites ranged from 25.5 to 114.0 μg/m3. For each 10 μg/m3 increase in PM2.5 exposures, the multivariate-adjusted hazard ratio was 1.251 (95% confidence interval: 1.220 to 1.283) for CVD incidence and 1.164 (95% confidence interval: 1.117 to 1.213) for CVD mortality. The slopes of concentration-response functions of PM2.5 exposure and CVD risks were steeper at high PM2.5 levels. In addition, older residents, rural residents, and never smokers were more prone to adverse effects of PM2.5 exposure. ConclusionsThis study provides evidence that elevated long-term PM2.5 exposures lead to increased CVD risk in China. The effects are more pronounced at higher PM2.5 levels. These findings expand the current knowledge on adverse health effects of severe air pollution and highlight the potential cardiovascular benefits of air quality improvement in China and other low- and middle-income countries.

A Validation Approach Considering the Uneven Distribution of Ground Stations for Satellite-Based PM2.5 Estimation

Satellite remote sensing has been increasingly employed for the estimation of ground-level atmospheric PM2.5. There have been several cross-validation (CV) approaches applied for the validation of satellite-based PM2.5 estimation models. However, these validation approaches often lead to confusion, due to the unclear applicable conditions. For this, we fully analyze and assess the existing validation approaches, and provide suggestions on applicable conditions for them. Furthermore, the existing validation approaches still have limitations to disregard the uneven distribution of ground stations, and tend to overestimate the performance of the PM2.5 estimation models. To this end, a CV-based validation approach considering the uneven spatial distribution of monitoring stations (denoted as SDCV) is proposed. SDCV introduces the spatial distance between validation station and modeling station into the CV process, and evaluates the spatial performance through a strategy of excluding modeling stations within a specific distance. Meanwhile, this approach has designed reasonable evaluation indices for the model validation. Taking China as a case study, the results indicate that SDCV can yield a more complete and effective evaluation for the popular PM2.5 estimation models than the traditional validation approaches.

Quantifying PM2.5 mass concentration and particle radius using satellite data and an optical-mass conversion algorithm

Although satellite-based approaches have been developed and adopted for estimating the concentration of fine particulate matter (PM2.5) with promising accuracy, few studies have considered mass concentration and particle radius simultaneously, even though particle size is significant for human health impacts. We developed a satellite-based PM2.5 retrieval method using optical-mass relationships via aerosol microphysical characteristics. Satellite data from the MODerate resolution Imaging Spectroradiometer (MODIS) instrument, combined with parameters from meteorological reanalysis, were processed to calculate particle radii and retrieve PM2.5 mass concentrations over China in 2017. Our study is the first to identify the spatial pattern of mean PM2.5 radius over China, which was validated against observations from AERONET (RMSE = 0.11 μm). Mean particle size over eastern China is smaller than in the west, depicting a clear bifurcation across the country, especially in summertime. This finding is attributed to variations in topography, meteorology, land use and population density, which affects the properties of emitted aerosols as well as their fate and transport. A statistically significant correlation (R = 0.82) was observed between estimated and measured annual PM2.5, with RMSE = 9.25 μg/m3, MAE = 6.98 μg/m3, MBE = −1.98 μg/m3 and RPE = 17.69% (N = 1270). The spatiotemporal distributions of resulting PM2.5 are consistent with previous findings, indicating the effectiveness and applicability of our method. Our approach quantifies PM2.5 mass concentrations without introducing regionally-specific fitting parameters, which can be efficiently applied across various spatial and temporal domains.

Long-term exposure to PM2.5 and fasting plasma glucose in non-diabetic adolescents in Yogyakarta, Indonesia

BackgroundIndonesia is facing serious air pollution. However, very few studies have been conducted to examine the health risks of air pollution in Indonesia, particularly for adolescents. ObjectiveTo assess the association between long-term exposure to ambient particles with a diameter of <2.5 μm (PM2.5) and fasting plasma glucose (FPG) in adolescents. MethodsA cross-sectional study was conducted in 482 adolescents aged 14–18 years in Yogyakarta, Indonesia in 2016. We finally included 469 (97.30%) participants who had no missing data for data analysis. We collected individual data on socio-demographics, behavioral habits, and health information through standardized questionnaires. Satellite-based PM2.5 concentrations from 2013 to 2016 were assigned based on participants’ residential addresses. The association between PM2.5 and FPG was examined using a generalized linear regression model while FPG was modeled as a continuous variable. An ordered logistic regression model was used to assess the relationship between PM2.5 and FPG categories. ResultsEvery 1 μg/m³ increase in PM2.5 was associated with a 0.34 mg/dL [95 confidence interval (95% CI): 0.08 mg/dL, 0.59 mg/dL] increase in FPG levels. Comparing with the low FPG level (under 86 mg/dL), every 1 μg/m³ increase in PM2.5 was associated with a 10.20% (95% CI: 1.60%, 19.80%) increase in the odds of impaired fasting glucose (IFG) (100–125 mg/dL). Stratified analyses indicated greater effects on participants with hypertension [odds ratio (OR) = 1.30, 95% CI: 1.09, 1.57] and those had higher physical activities (OR = 1.36, 95% CI: 1.09, 1.57). Adolescents’ sex, obesity status and different cutoff points of FPG did not modify the association between the exposure to PM2.5 and FPG levels. ConclusionLong-term exposure to PM2.5 was associated with increased FPG levels in Indonesian non-diabetic adolescents.

MAPLE

S13: How Low Should We Go? New Health Research on Low-level Ambient Air Pollution, Beatrix Theater, August 28, 2019, 1:30 PM - 3:00 PM Background/Aims Fine particulate matter (PM2.5) is associated with mortality, however the range of applicable concentrations is unknown. We applied novel satellite-based estimates of PM2.5 exposure to large population-based cohorts, and characterized the shape of the mortality relationship, with emphasis on the low concentrations (<12 µg/m3) observed throughout Canada. Methods Annual satellite-based PM2.5 exposure estimates were developed at 1 km resolution across Canada for 1998-2016 and back-casted to 1981 using remote sensing, chemical transport models and ground monitoring data. Exposures were estimated for ~8.5 million subjects in three Canadian Census Health and Environment Cohort (CanCHEC) cohorts, and 540,900 Canadian Community Health Survey (CCHS) participants, based on residential histories derived from annual tax records. Mortality was linked until 2016. Sensitivity analyses evaluated adjustment for (individual/contextual/geographic) covariates and gaseous (NO2, ozone) co-pollutants; exposure time windows and spatial scale. Estimates were evaluated across strata of age, sex and immigrant status and the shape of the PM2.5 and mortality association was examined with the Shape Constrained Health Impact Function (SCHIF). Results Satellite-based PM2.5 estimates were low (mean 3-year averages of 7.4μg/m3 and 5.9 µg/m3 in CanCHEC and CCHS, respectively) and highly correlated with ground monitors (R2 = 0.82, RMSE=1.5 µg/m3). In final adjusted models non-accidental mortality hazard ratios (HR) per 10µg/m3 change in PM2.5 were 1.04 (95% CI: 1.04-1.05) in CanCHEC and 1.11 (95% CI 1.04-1.18) in CCHS. In CanCHEC the SCHIF predicted a sublinear concentration-mortality curve with little increased risk below 5 g/m3, and monotonically increased HRs above this concentration. In CCHS we estimated a supra-linear relationship to concentrations below 2µg/m3. HRs were larger for females, younger ages and non-immigrants and attenuated by gaseous pollutants. Conclusions In large population-based cohorts exposed to low levels of air pollution with extended follow-up we observed associations between PM2.5 and non-accidental mortality for concentrations as low as 5 µg/m3.

Benefits of High Resolution PM2.5 Prediction using Satellite MAIAC AOD and Land Use Regression for Exposure Assessment: California Examples.

This study estimated annual average ambient fine particulate matter (PM2.5) concentrations at 1 km resolution using satellite Multi-Angle Implementation of Atmospheric Correction (MAIAC) aerosol optical depth (AOD), land use parameters, and meteorology in California for the year 2016 (cross-validation R2 = 0.73 (site-based) and 0.81 (observation-based)). Using these high-resolution PM2.5 estimates, regionally varying urban enhancements of PM2.5 concentrations, 1.43-2.77 μg/m3 (23.9-36.2%), were identified in the densely populated air basins of San Francisco Bay Area, San Joaquin Valley, and South Coast. On the other hand, within-urban PM2.5 variability was found to be 31.4-35.6% of between-urban variability across California. However, this pattern was not consistent from region to region, even showing higher within-urban variability (e.g., San Francisco Bay Area). In addition, satellite-based PM2.5 concentrations were statistically significantly associated with demographic factors (i.e., % people of color, % poverty, and % low education) with the strongest positive association with % people of color (1.05 and 2.72 μg/m3 increases per interquartile range (IQR) and range increases, respectively). The fine-scale PM2.5 estimates enabled the assessment of long-term PM2.5 exposures for all populations particularly benefiting rural populations and socially vulnerable populations widely distributed in each urban area. This study provided evidence of regionally varying exposure misclassification that would arise without accounting for rural and within-urban exposure variabilities.

Estimating daily PM2.5 concentrations in New York City at the neighborhood-scale: Implications for integrating non-regulatory measurements

Previous PM2.5 related epidemiological studies mainly relied on data from sparse regulatory monitors to assess exposure. The introduction of non-regulatory PM2.5 monitors presents both opportunities and challenges to researchers and air quality managers. In this study, we evaluated the advantages and limitations of integrating non-regulatory PM2.5 measurements into a satellite-based daily PM2.5 model at 100 m resolution in New York City in 2015. Two separate machine learning models were developed, one using only PM2.5 data from the US Environmental Protection Agency (EPA), and the other with measurements from both EPA and the New York City Community Air Survey (NYCCAS). The EPA-only model obtained a cross-validation (CV) R2 of 0.85 while the EPA + NYCCAS model obtained a CV R2 of 0.73. With the help of the NYCCAS measurements, the EPA + NYCCAS model predicted distinctly different PM2.5 spatial patterns and more pollution hotspots compared with the EPA model, and its predictions were >15% higher than the EPA model along major roads and in densely populated areas. Our results indicated that satellite AOD and non-regulatory PM2.5 measurements can be fused together to capture neighborhood-scale PM2.5 levels and previous studies may have underestimated the disease burden due to PM2.5 in densely populated areas.

Advancing the prediction accuracy of satellite-based PM2.5 concentration mapping: A perspective of data mining through in situ PM2.5 measurements

Ground-measured PM2.5 concentration data are oftentimes used as a response variable in various satellite-based PM2.5 mapping practices, yet few studies have attempted to incorporate ground-measured PM2.5 data collected from nearby stations or previous days as a priori information to improve the accuracy of gridded PM2.5 mapping. In this study, Gaussian kernel-based interpolators were developed to estimate prior PM2.5 information at each grid using neighboring PM2.5 observations in space and time. The estimated prior PM2.5 information and other factors such as aerosol optical depth (AOD) and meteorological conditions were incorporated into random forest regression models as essential predictor variables for more accurate PM2.5 mapping. The results of our case study in eastern China indicate that the inclusion of ground-based PM2.5 neighborhood information can significantly improve PM2.5 concentration mapping accuracy, yielding an increase of out-of-sample cross validation R2 by 0.23 (from 0.63 to 0.86) and a reduction of RMSE by 7.72 (from 19.63 to 11.91) μg/m3. In terms of the estimated relative importance of predictors, the PM2.5 neighborhood information played a more critical role than AOD in PM2.5 predictions. Compared with the temporal PM2.5 neighborhood term, the spatially neighboring PM2.5 term has an even larger potential to improve the final PM2.5 prediction accuracy. Additionally, a more robust and straightforward PM2.5 predictive framework was established by screening and removing the least important predictor stepwise from each modeling trial toward the final optimization. Overall, our results fully confirmed the positive effects of ground-based PM2.5 information over spatiotemporally neighboring space on the holistic PM2.5 mapping accuracy.

Impact of China’s Air Pollution Prevention and Control Action Plan on PM2.5 chemical composition over eastern China

China promulgated the Air Pollution Prevention and Control Action Plan (the Action Plan) in 2013 and developed stringent control measures to mitigate fine particulate matter (PM2.5) pollution. Here, we investigated the PM2.5 chemical composition changes over eastern China associated with the Action Plan during 2013–2017 using satellite-based PM2.5 chemical composition data derived using CMAQ simulations and satellite inputs. The PM2.5 concentrations decreased considerably during this time as a result of the reductions in all chemical species in PM2.5. The population-weighted mean concentrations over eastern China decreased from 11.1 to 6.7 μg m–3 for SO 4 2 − , 13.8–13.1 μg m–3 for NO 3 − , 7.4–5.8 μg m–3 for NH 4 + , 9.9–8.4 μg m–3 for OM, 4.6–3.8 μg m–3 for BC and 12.9–9.6 μg m–3 for other species in PM2.5. SO 4 2 − had the largest reduction of 40%, while NO 3 − had the lowest reduction of 5%, resulting in a greater fraction of NO 3 − and a smaller fraction of SO 4 2 − in PM2.5. Among the three key regions, Beijing-Tianjin-Hebei had the largest reduction in PM2.5 and its chemical compositions. The decrease in SO 4 2 − concentrations was in line with the reduction of SO2 emissions, and the major driver of the SO2 emission reductions was the industrial sector. The decrease in NO 3 − concentrations was limited because the decrease in SO2 emissions and the stable NH3 emissions facilitated the formation of NO 3 − from HNO3, which partially offset the reduction in NO x emissions driven by the power sector. To mitigate PM2.5 pollution more effectively, future efforts are needed to reduce NH3 emissions.

Source Contributions to Ambient Fine Particulate Matter for Canada.

Understanding the sectoral contribution of emissions to fine particulate matter (PM2.5) offers information for air quality management, and for investigation of association with health outcomes. This study evaluates the contribution of different emission sectors to PM2.5 in 2013 for Canada using the GEOS-Chem chemical transport model, downscaled with satellite-based PM2.5. Despite the low population-weighted PM2.5 concentrations of 5.5 μg m-3 across Canada, we find that over 70% of population-weighted PM2.5 originates from Canadian sources followed by 30% from the contiguous United States. The three leading sectoral contributors to population-weighted PM2.5 over Canada are wildfires with 1.0 μg m-3 (17%), transportation with 0.96 μg m-3 (16%), and residential combustion with 0.91 μg m-3 (15%). The relative contribution to population-weighted PM2.5 of different sectors varies regionally with residential combustion as the leading contributor in Central Canada (19%), while wildfires dominate over Northern Canada (59%), Atlantic Canada (34%), and Western Canada (18%). The contribution from U.S. sources is larger over Central Canada (33%) than over Western Canada (17%), Atlantic Canada (17%), and Northern Canada (<2%). Sectoral trend analysis showed that the contribution from anthropogenic sources to population-weighted PM2.5 decreased from 7.1 μg m-3 to 3.4 μg m-3 over the past two decades.