Grantee Research Project Results
2010 Progress Report: Spatial temporal analysis of health effects associated with sources and speciation of fine PM
EPA Grant Number: R833863Title: Spatial temporal analysis of health effects associated with sources and speciation of fine PM
Investigators: Fuentes, Montserrat , Frey, H. Christopher , Bell, Michelle L. , Reich, Brian , Dominici, Francesca , Zhang, Yang
Institution: North Carolina State University , The Johns Hopkins University , Yale University
EPA Project Officer: Chung, Serena
Project Period: December 1, 2008 through November 30, 2012 (Extended to November 30, 2013)
Project Period Covered by this Report: December 1, 2009 through November 30,2010
Project Amount: $893,439
RFA: Innovative Approaches to Particulate Matter Health, Composition, and Source Questions (2007) RFA Text | Recipients Lists
Research Category: Particulate Matter , Air
Objective:
The overall objectives of this proposed nationwide spatiotemporal analysis are to investigate the adverse health outcomes associated with population exposure to fine particulate matter (PM2.5) and speciation and to characterize geographic differences, sources, and population heterogeneity in the putatively PM2.5 mediated health effects, combining different sources of data with atmospheric models.
We aim to answer the following research questions:
What is the recommended framework to integrate atmospheric models with monitoring data and other sources of information to obtain a better spatial and temporal characterization of fine PM components and sources? Can we improve the PM component-based epidemiologic studies by using atmospheric and exposure models? How to integrate the atmospheric models in this epidemiologic framework, while characterizing uncertainties in the epidemiological and numerical models? How to use source apportionment approaches in national epidemiologic studies, while characterizing different sources of uncertainty in the models and the data?
Progress Summary:
During Year 2, we have accomplished most of the objectives that are critical to achieve our final goals. Our major findings and their significances are summarized below.
- The CMAQ model evaluation results show that the model performs well for O3 predictions, with NMBs of -8.2% to 7.2% and NMEs from 12.3 – 22.7%. Large biases exist in 24-hr average PM2.5 concentrations and its components (e.g., overpredicted in January with NMBs of 18.8 – 52.1% but underpredicted in July with NMBs of -39.2% to -26.3% for PM2.5). Possible reasons for biases in simulated O3 and PM2.5 include uncertainties in emissions of precursor species (e.g., SO2, NH3, and NOx), biases in meteorological predictions (e.g, wind speed, precipitation), and uncertainties in their boundary conditions. Despite these biases, the overall model performance for PM and its composition is consistent with current PM model performance reported in the literature (e.g., Eder and Yu, 2006; Zhang et al., 2006b, 2009) and is considered to be satisfactory based on recommended criteria in these papers.
- Using the CMAQ/Brute Force Method (BFM) for source apportionment, we found that biomass burning is the most important source domainwide in January with a contribution of 13.3% to surface monthly-mean PM2.5. POA is the species contributing the largest (7.4%) to the overall PM2.5 contribution from biomass burning. Miscellaneous area sources and coal combustion are the 2 next largest sources in January with contributions of 11.8% and 10.8%, respectively. Coal combustion is the most important source domainwide in July, with a contribution of 30.8% to surface monthly-mean PM2.5; SO42- contributes to 25.8% of the overall PM2.5 contribution from coal combustion. Miscellaneous area sources and industrial processes are the 2 next largest sources in July with contributions of 8.9% and 6.9%, respectively.
- Source apportionment results obtained using the BFM are subject to inherent limitations, most notably its assumption that the contributions from each source category are linear and additive. Such an assumption may not be valid for non-linear processes in the atmosphere. In addition, the computational demand of the BFM is usually high when conducting source sensitivity simulations for several source categories. Separate simulations are required for each source category for each month.
- We have developed methods for exposure apportionment and demonstrated these for a case study of Bronx and Queens Counties in New York City in 2002. For all three sites, the mean contribution from secondary sulfate contribution is significantly greater in the summer than in winter. Conversely, the mean secondary nitrate contribution is significantly smaller in summer than in winter. Secondary sulfate and nitrate are influenced by primary SO2 emissions from coal combustion and primary NOx emissions from fuel combustion. Thus, contributions to ambient PM2.5 for secondary sulfate are likely to be representative of high sulfur emission sources such as coal power plants and furnaces. Source apportionment to nitrate is likely to represent NOx emission sources such as vehicles, power plants, and other combustion sources.
- Overall source apportionment of total PM2.5 exposure was conducted for New York city in 2002, integrating the results of ambient source apportionment and total source exposure source apportionment by SHEDS-PM. For both smokers and non-smokers exposed to ETS, home smoking contributes at least 69.5% of total exposure for all seasons. Home cooking contributes about 5% for smokers and 4% for non-smokers exposed to ETS. For non-smokers not exposed to ETS, home cooking contributes 16% to 38%. For all population groups, motor vehicle contributes the largest share of total exposure after home cooking and/or home smoking. For non-smokers not exposed to ETS, motor vehicle contributes as high as 33% of the total exposure. Secondary sulfate and secondary nitrate contribute a notable amount. All other sources contribute less than 10% of the total exposure. The seasonal variability of non-ambient exposure source apportionment results from air exchange rates variability. High air exchange rates cause lower non-ambient exposure.
- To better understand the certainty of CMAQ estimates, basic model evaluation was performed comparing model results to monitored (observed) concentrations using several metrics. Both individual and spatially-aggregated (county-level) exposure estimates for PM2.5 and O3 were calculated for all counties for year 2002. The advantages and limitations of the various modeling approaches were assessed. Results indicate that exposure estimates generated from CMAQ provided greater spatial coverage as well as improved spatial and temporal resolution, compared to exposures from ambient monitors. For county-level exposure estimates, the modeling approach covers 150% more population than the monitoring approach for both PM2.5 and O3. Generally, CMAQ provided reasonable estimates of PM2.5 concentrations, with a tendency to under-estimate PM2.5 at low observed concentrations (e.g., <10 μg/m3) (mean fraction bias (MFB) = -1.47% and mean fractional error (MFE) = 25.1%). The model was less successful at modeling observed O3 concentrations, with the model consistently over-estimating O3 levels compared to monitored concentrations (MFB = 36.0% and MFE= 37.4%). Correlation (and R2) values between modeled and observed pollutant concentrations, matched by day and location, were 0.62 (0.38) and 0.76 (0.52) for PM2.5 and O3, respectively. Counties with monitors tend to be more urban than counties without monitors, and there were also significant differences in percentage of black residents, college graduates, population below the poverty line, as well as differences in median income and modeled pollutant concentrations. These findings are relevant to epidemiological studies of the human health consequences of air pollution because they provide insight into how air quality modeling can be used to address limitations in the traditional method of exposure assessment (i.e., use of ambient air quality monitors).
- We studied the impact of speciated PM2.5 exposure on premature mortality, using daily mortality data at the county-level nationwide for years 1999-2005, and monitoring data. Overall a 10-μg/m3 increase in PM2.5 at lag 1 is associated with 0.25 (95 percent interval -0.08, 0.58) in all-cause mortality. We conducted similar analysis for the PM components, the organic carbon and the silicon were the components that appeared to have a significant impact on mortality. We also evaluated the impact of different metrics to characterize exposure on the estimated risk, using air quality model (CMAQ), monitoring data, and combining CMAQ with monitoring data. To characterize how the effect of PM2.5 and ozone affect the risk of mortality jointly, we allowed for non-linear interaction. We found that the metric for PM2.5 can have a substantial impact on the results. For example, using monitoring data in North Carolina we found a 1.5% increase in risk for a 10 μg/m3increase in PM2:5 (z-score 2.7) and when using the CMA-monitoring combined surfaces the estimated increase is 3.2% (z-score 6.0). Also the use of the general interaction between PM2.5 and ozone shows dramatic interactions that are not observed in the linear main effects analysis.
Future Activities:
During Year 3, we plan to conduct the following tasks:
- Conduct detailed source apportionment analysis at individual sites representative of urban, rural, and coastal ambient conditions;
- Intercompare CMAQ/BFM results with published results of source apportionment using other approaches
- Conduct source apportionment using the Particulate Source Apportionment Technology (PSAT) in the Comprehensive Air Quality Model with Extensions (CAMx) (ENVIRON, 2006), if resources and time permitted.
- Conduct exposure apportionment case studies for Harris County, TX, and a six county region in NC along the I-40 corridor. These case studies will enable us to assess seasonal, climatic, and regional differences in exposure apportionment.
- With respect to the health data, we are currently recreating hospital admissions datasets for the area covering the spatial domain with information by cause of death, age at death, etc. for persons ages 65 years and older from Medicare billing information, we will continue this work. This is but one of the health datasets that will be available for this project. We also plan to continue our analysis using mortality data, linking mortality to apportioned PM. We continue to refine the analytical plan to combine regional air quality modeling exposure data with health data. The study designs and health datasets have been presented at conference calls with the research team and EPA in order to gain feedback and keep members of the research team and EPA informed of different portions of the project.
- In addition, we will continue to disseminate our research results at national/international conferences/workshops and prepare manuscripts for publications in peer-reviewed journals.
Journal Articles on this Report : 8 Displayed | Download in RIS Format
Other project views: | All 90 publications | 49 publications in selected types | All 49 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Anderson GB, Bell ML. Does one size fit all? The suitability of standard ozone exposure metric conversion ratios and implications for epidemiology. Journal of Exposure Science and Environmental Epidemiology 2010;20(1):2-11. |
R833863 (2009) R833863 (2010) R833863 (Final) |
Exit Exit |
|
Burr MJ, Zhang Y. Source apportionment of fine particulate matter over the Eastern U.S. Part I: source sensitivity simulations using CMAQ with the brute force method. Atmospheric Pollution Research 2011;2(3):300-317. |
R833863 (2010) R833863 (2011) R833863 (Final) |
Exit Exit Exit |
|
Cao Y, Frey HC. Assessment of interindividual and geographic variability in human exposure to fine particulate matter in environmental tobacco smoke. Risk Analysis 2011;31(4):578-591. |
R833863 (2010) R833863 (2011) R833863 (Final) |
Exit |
|
Chang HH, Zhou J, Fuentes M. Impact of climate change on ambient ozone level and mortality in Southeastern United States. International Journal of Environmental Research and Public Health 2010;7(7):2866-2880. |
R833863 (2010) R833863 (Final) |
Exit Exit |
|
Dennis R, Fox T, Fuentes M, Gilliland A, Hanna S, Hogrefe C, Irwin J, Rao ST, Scheffe R, Schere K, Steyn D, Venkatram A. A framework for evaluating regional-scale numerical photochemical modeling systems. Environmental Fluid Mechanics 2010;10(4):471-489. |
R833863 (2009) R833863 (2010) R833863 (Final) |
Exit Exit |
|
Liu X, Frey HC, Cao Y. Estimating in-vehicle concentration of and exposure to fine particulate matter: near-roadway ambient air quality and variability in vehicle operation. Transportation Research Record 2010;2158:105-112. |
R833863 (2010) R833863 (2011) R833863 (Final) |
Exit Exit |
|
Reich BJ, Fuentes M, Herring AH, Evenson KR. Bayesian variable selection for multivariate spatially varying coefficient regression. Biometrics 2010;66(3):772-782. |
R833863 (2009) R833863 (2010) R833863 (Final) |
Exit Exit |
|
Reich BJ, Fuentes M, Dunson DB. Bayesian spatial quantile regression. Journal of the American Statistical Association 2011;106(493):6-20. |
R833863 (2010) R833863 (Final) |
Exit Exit |
Supplemental Keywords:
Bayesian inference, epidemiology,public health data, particulate matter, pollution exposure, risk assessment, statistical modelling,Relevant Websites:
http://www4.stat.ncsu.edu/~fuentes Exit
http://environment.yale.edu/profile/bell Exit
http://www4.ncsu.edu/~frey/ Exit
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.