Xuefei Hu

Estimating ground-level PM2.5 in China using satellite remote sensing.

Estimating ground-level PM2.5 from satellite-derived aerosol optical depth (AOD) using a spatial statistical model is a promising new method to evaluate the spatial and temporal characteristics of PM2.5 exposure in a large geographic region. However, studies outside North America have been limited due to the lack of ground PM2.5 measurements to calibrate the model. Taking advantage of the newly established national monitoring network, we developed a national-scale geographically weighted regression (GWR) model to estimate daily PM2.5 concentrations in China with fused satellite AOD as the primary predictor. The results showed that the meteorological and land use information can greatly improve model performance. The overall cross-validation (CV) R(2) is 0.64 and root mean squared prediction error (RMSE) is 32.98 μg/m(3). The mean prediction error (MPE) of the predicted annual PM2.5 is 8.28 μg/m(3). Our predicted annual PM2.5 concentrations indicated that over 96% of the Chinese population lives in areas that exceed the Chinese National Ambient Air Quality Standard (CNAAQS) Level 2 standard. Our results also confirmed satellite-derived AOD in conjunction with meteorological fields and land use information can be successfully applied to extend the ground PM2.5 monitoring network in China.

Satellite-Based Spatiotemporal Trends in PM2.5 Concentrations: China 2004-2013

Background Three decades of rapid economic development is causing severe and widespread PM2.5 (particulate matter ≤ 2.5 μm) pollution in China. However, research on the health impacts of PM2.5 exposure has been hindered by limited historical PM2.5 concentration data. Objectives We estimated ambient PM2.5 concentrations from 2004 to 2013 in China at 0.1° resolution using the most recent satellite data and evaluated model performance with available ground observations. Methods We developed a two-stage spatial statistical model using the Moderate Resolution Imaging Spectroradiometer (MODIS) Collection 6 aerosol optical depth (AOD) and assimilated meteorology, land use data, and PM2.5 concentrations from China’s recently established ground monitoring network. An inverse variance weighting (IVW) approach was developed to combine MODIS Dark Target and Deep Blue AOD to optimize data coverage. We evaluated model-predicted PM2.5 concentrations from 2004 to early 2014 using ground observations. Results The overall model cross-validation R2 and relative prediction error were 0.79 and 35.6%, respectively. Validation beyond the model year (2013) indicated that it accurately predicted PM2.5 concentrations with little bias at the monthly (R2 = 0.73, regression slope = 0.91) and seasonal (R2 = 0.79, regression slope = 0.92) levels. Seasonal variations revealed that winter was the most polluted season and that summer was the cleanest season. Analysis of predicted PM2.5 levels showed a mean annual increase of 1.97 μg/m3 between 2004 and 2007 and a decrease of 0.46 μg/m3 between 2008 and 2013. Conclusions Our satellite-driven model can provide reliable historical PM2.5 estimates in China at a resolution comparable to those used in epidemiologic studies on the health effects of long-term PM2.5 exposure in North America. This data source can potentially advance research on PM2.5 health effects in China. Citation Ma Z, Hu X, Sayer AM, Levy R, Zhang Q, Xue Y, Tong S, Bi J, Huang L, Liu Y. 2016. Satellite-based spatiotemporal trends in PM2.5 concentrations: China, 2004–2013. Environ Health Perspect 124:184–192; http://dx.doi.org/10.1289/ehp.1409481

10-year spatial and temporal trends of PM 2.5 concentrations in the southeastern US estimated using high-resolution satellite data

Long-term PM2.5 exposure has been associated with various adverse health outcomes. However, most ground monitors are located in urban areas, leading to a potentially biased representation of true regional PM2.5 levels. To facilitate epidemiological studies, accurate estimates of the spatiotemporally continuous distribution of PM2.5 concentrations are important. Satellite-retrieved aerosol optical depth (AOD) has been increasingly used for PM2.5 concentration estimation due to its comprehensive spatial coverage. Nevertheless, previous studies indicated that an inherent disadvantage of many AOD products is their coarse spatial resolution. For instance, the available spatial resolutions of the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Multiangle Imaging SpectroRadiometer (MISR) AOD products are 10 and 17.6 km, respectively. In this paper, a new AOD product with 1 km spatial resolution retrieved by the multi-angle implementation of atmospheric correction (MAIAC) algorithm based on MODIS measurements was used. A two-stage model was developed to account for both spatial and temporal variability in the PM2.5–AOD relationship by incorporating the MAIAC AOD, meteorological fields, and land use variables as predictors. Our study area is in the southeastern US centered at the Atlanta metro area, and data from 2001 to 2010 were collected from various sources. The model was fitted annually, and we obtained model fitting R2 ranging from 0.71 to 0.85, mean prediction error (MPE) from 1.73 to 2.50 μg m−3, and root mean squared prediction error (RMSPE) from 2.75 to 4.10 μg m−3. In addition, we found cross-validation R2 ranging from 0.62 to 0.78, MPE from 2.00 to 3.01 μgm−3, and RMSPE from 3.12 to 5.00 μgm−3, indicating a good agreement between the estimated and observed values. Spatial trends showed that high PM2.5 levels occurred in urban areas and along major highways, while low concentrations appeared in rural or mountainous areas. Our time-series analysis showed that, for the 10-year study period, the PM2.5 levels in the southeastern US have decreased by ∼20 %. The annual decrease has been relatively steady from 2001 to 2007 and from 2008 to 2010 while a significant drop occurred between 2007 and 2008. An observed increase in PM2.5 levels in year 2005 is attributed to elevated sulfate concentrations in the study area in warm months of 2005.

Calibrating MODIS aerosol optical depth for predicting daily PM2.5 concentrations via statistical downscaling

There has been a growing interest in the use of satellite-retrieved aerosol optical depth (AOD) to estimate ambient concentrations of PM2.5 (particulate matter <2.5 μm in aerodynamic diameter). With their broad spatial coverage, satellite data can increase the spatial–temporal availability of air quality data beyond ground monitoring measurements and potentially improve exposure assessment for population-based health studies. This paper describes a statistical downscaling approach that brings together (1) recent advances in PM2.5 land use regression models utilizing AOD and (2) statistical data fusion techniques for combining air quality data sets that have different spatial resolutions. Statistical downscaling assumes the associations between AOD and PM2.5 concentrations to be spatially and temporally dependent and offers two key advantages. First, it enables us to use gridded AOD data to predict PM2.5 concentrations at spatial point locations. Second, the unified hierarchical framework provides straightforward uncertainty quantification in the predicted PM2.5 concentrations. The proposed methodology is applied to a data set of daily AOD values in southeastern United States during the period 2003–2005. Via cross-validation experiments, our model had an out-of-sample prediction R2 of 0.78 and a root mean-squared error (RMSE) of 3.61 μg/m3 between observed and predicted daily PM2.5 concentrations. This corresponds to a 10% decrease in RMSE compared with the same land use regression model without AOD as a predictor. Prediction performances of spatial–temporal interpolations to locations and on days without monitoring PM2.5 measurements were also examined.

Assessing Intra-Urban Surface Energy Fluxes Using Remotely Sensed ASTER Imagery and Routine Meteorological Data: A Case Study in Indianapolis, U.S.A.

The seasonal and spatial variability of surface heat fluxes is crucial to the understanding of urban heat island phenomenon and dynamics. To estimate energy fluxes, remote sensing derived biophysical variables need to be integrated with surface atmospheric parameters measured in meteorological stations or in situ field measurements. In this study, based on the two-source energy balance algorithm, we applied a method to estimate surface energy fluxes by combined use of multispectral ASTER images and routine meteorological data, and applied it to the City of Indianapolis, United States, aiming at in-depth understanding of the spatial patterns of energy fluxes. By computing the fluxes by land use and land cover (LULC) type, we further investigated the spatial variability of heat fluxes. Results show that the energy fluxes possessed a strong seasonality, with the highest net radiation in summer, followed by spring, fall and winter. Sensible heat flux tended to change largely with surface temperature, while latent heat was largely modulated by the change in vegetation abundance and vigor and the accompanying moisture condition. The fluctuation in all heat fluxes tended to be high in the summer months and low in the winter months. Sensible and latent heat fluxes showed a stronger spatial variability than net radiation and ground heat flux. The variations of net radiation among the land use/cover types were mainly attributable to surface albedo and temperature, while the within-class variability in the turbulent heat fluxes was more associated with the changes in vegetation, water bodies, and other surface factors.

Pediatric Emergency Visits and Short-Term Changes in PM2.5 Concentrations in the U.S. State of Georgia.

Background: Associations between pediatric emergency department (ED) visits and ambient concentrations of particulate matter ≤ 2.5 μm in diameter (PM2.5) have been reported in previous studies, although few were performed in nonmetropolitan areas. Objective: We estimated associations between daily PM2.5 concentrations, using a two-stage model that included land use parameters and satellite aerosol optical depth measurements at 1-km resolution, and ED visits for six pediatric conditions in the U.S. state of Georgia by urbanicity classification. Methods: We obtained pediatric ED visits geocoded to residential ZIP codes for visits with nonmissing PM2.5 estimates and admission dates during 1 January 2002–30 June 2010 for 2- to 18-year-olds for asthma or wheeze (n = 189,816), and for 0- to 18-year-olds for bronchitis (n = 76,243), chronic sinusitis (n = 15,745), otitis media (n = 237,833), pneumonia (n = 52,946), and upper respiratory infections (n = 414,556). Daily ZIP code–level estimates of 24-hr average PM2.5 were calculated by averaging concentrations within ZIP code boundaries. We used time-stratified case-crossover models stratified on ZIP code, year, and month to estimate odds ratios (ORs) between ED visits and same-day and previous-day PM2.5 concentrations at the ZIP code level, and we investigated effect modification by county-level urbanicity. Results: A 10-μg/m3 increase in same-day PM2.5 concentrations was associated with ED visits for asthma or wheeze (OR = 1.013; 95% CI: 1.003, 1.023) and upper respiratory infections (OR = 1.015; 95% CI: 1.008, 1.022); associations with previous-day PM2.5 concentrations were lower. Differences in the association estimates across levels of urbanicity were not statistically significant. Conclusion: Pediatric ED visits for asthma or wheeze and for upper respiratory infections were associated with PM2.5 concentrations in Georgia. Citation: Strickland MJ, Hao H, Hu X, Chang HH, Darrow LA, Liu Y. 2016. Pediatric emergency visits and short-term changes in PM2.5 concentrations in the U.S. state of Georgia. Environ Health Perspect 124:690–696; http://dx.doi.org/10.1289/ehp.1509856

Statistical data fusion of multi-sensor AOD over the Continental United States

This article illustrates two techniques for merging daily aerosol optical depth (AOD) measurements from satellite and ground-based data sources to achieve optimal data quality and spatial coverage. The first technique is a traditional Universal Kriging (UK) approach employed to predict AOD from multi-sensor aerosol products that are aggregated on a reference grid with AERONET as ground truth. The second technique is spatial statistical data fusion (SSDF); a method designed for massive satellite data interpolation. Traditional kriging has computational complexity O(N3), making it impractical for large datasets. Our version of UK accommodates massive data inputs by performing kriging locally, while SSDF accommodates massive data inputs by modelling their covariance structure with a low-rank linear model. In this study, we use aerosol data products from two satellite instruments: the moderate resolution imaging spectrometer and the geostationary operational environmental satellite, covering the Continental United States.

Maternal exposure to traffic-related air pollution and birth defects in Massachusetts.

Exposures to particulate matter with diameter of 2.5µm or less (PM2.5) may influence risk of birth defects. We estimated associations between maternal exposure to prenatal traffic-related air pollution and risk of cardiac, orofacial, and neural tube defects among Massachusetts births conceived 2001 through 2008. Our analyses included 2729 cardiac, 255 neural tube, and 729 orofacial defects. We used satellite remote sensing, meteorological and land use data to assess PM2.5 and traffic-related exposures (distance to roads and traffic density) at geocoded birth addresses. We calculated adjusted odds ratios (OR) and confidence intervals (CI) using logistic regression models. Generalized additive models were used to assess spatial patterns of birth defect risk. There were positive but non-significant associations for a 10µg/m(3) increase in PM2.5 and perimembranous ventricular septal defects (OR=1.34, 95% CI: 0.98, 1.83), patent foramen ovale (OR=1.19, 95% CI: 0.92, 1.54) and patent ductus arteriosus (OR=1.20, 95% CI: 0.95, 1.62). There was a non-significant inverse association between PM2.5 and cleft lip with or without palate (OR=0.76, 95% CI: 0.50, 1.10), cleft palate only (OR=0.89, 95% CI: 0.54, 1.46) and neural tube defects (OR=0.77, 95% CI: 0.46, 1.05). Results for traffic related exposure were similar. Only ostium secundum atrial septal defects displayed significant spatial variation after accounting for known risk factors.

Improving satellite‐driven PM2.5 models with Moderate Resolution Imaging Spectroradiometer fire counts in the southeastern U.S.

Multiple studies have developed surface PM2.5 (particle size less than 2.5 µm in aerodynamic diameter) prediction models using satellite-derived aerosol optical depth as the primary predictor and meteorological and land use variables as secondary variables. To our knowledge, satellite-retrieved fire information has not been used for PM2.5 concentration prediction in statistical models. Fire data could be a useful predictor since fires are significant contributors of PM2.5. In this paper, we examined whether remotely sensed fire count data could improve PM2.5 prediction accuracy in the southeastern U.S. in a spatial statistical model setting. A sensitivity analysis showed that when the radius of the buffer zone centered at each PM2.5 monitoring site reached 75 km, fire count data generally have the greatest predictive power of PM2.5 across the models considered. Cross validation (CV) generated an R2 of 0.69, a mean prediction error of 2.75 µg/m3, and root-mean-square prediction errors (RMSPEs) of 4.29 µg/m3, indicating a good fit between the dependent and predictor variables. A comparison showed that the prediction accuracy was improved more substantially from the nonfire model to the fire model at sites with higher fire counts. With increasing fire counts, CV RMSPE decreased by values up to 1.5 µg/m3, exhibiting a maximum improvement of 13.4% in prediction accuracy. Fire count data were shown to have better performance in southern Georgia and in the spring season due to higher fire occurrence. Our findings indicate that fire count data provide a measurable improvement in PM2.5 concentration estimation, especially in areas and seasons prone to fire events.

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

PM2.5 concentration fields that correspond to just meeting national ambient air quality standards (NAAQS) are useful for characterizing exposure in regulatory assessments. Computationally efficient methods that incorporate predictions from photochemical grid models (PGM) are needed to realistically project baseline concentration fields for these assessments. Thorough cross validation (CV) of hybrid spatial prediction models is also needed to better assess their predictive capability in sparsely monitored areas. In this study, a system for generating, evaluating, and projecting PM2.5 spatial fields to correspond with just meeting the PM2.5 NAAQS is developed and demonstrated. Results of ten-fold CV based on standard and spatial cluster withholding approaches indicate that performance of three spatial prediction models improves with decreasing distance to the nearest neighboring monitor, improved PGM performance, and increasing distance from sources of PM2.5 heterogeneity (e.g., complex terrain and fire). An air quality projection tool developed here is demonstrated to be effective for quickly projecting PM2.5 spatial fields to just meet NAAQS using realistic spatial response patterns based on air quality modeling. PM2.5 tends to be most responsive to primary PM2.5 emissions in urban areas, whereas response patterns are relatively smooth for NOx and SO2 emission changes. On average, PM2.5 is more responsive to changes in anthropogenic primary PM2.5 emissions than NOx and SO2 emissions in the contiguous U.S.To estimate PM2.5 concentrations, many parametric regression models have been developed, while nonparametric machine learning algorithms are used less often and national-scale models are rare. In this paper, we develop a random forest model incorporating aerosol optical depth (AOD) data, meteorological fields, and land use variables to estimate daily 24 h averaged ground-level PM2.5 concentrations over the conterminous United States in 2011. Random forests are an ensemble learning method that provides predictions with high accuracy and interpretability. Our results achieve an overall cross-validation (CV) R2 value of 0.80. Mean prediction error (MPE) and root mean squared prediction error (RMSPE) for daily predictions are 1.78 and 2.83 μg/m3, respectively, indicating a good agreement between CV predictions and observations. The prediction accuracy of our model is similar to those reported in previous studies using neural networks or regression models on both national and regional scales. In addition, the incorporation of convolutional layers for land use terms and nearby PM2.5 measurements increase CV R2 by ∼0.02 and ∼0.06, respectively, indicating their significant contributions to prediction accuracy. A pair of different variable importance measures both indicate that the convolutional layer for nearby PM2.5 measurements and AOD values are among the most-important predictor variables for the training process.