Mariel D. Friberg

Method for Fusing Observational Data and Chemical Transport Model Simulations To Estimate Spatiotemporally Resolved Ambient Air Pollution

Investigations of ambient air pollution health effects rely on complete and accurate spatiotemporal air pollutant estimates. Three methods are developed for fusing ambient monitor measurements and 12 km resolution chemical transport model (CMAQ) simulations to estimate daily air pollutant concentrations across Georgia. Temporal variance is determined by observations in one method, with the annual mean CMAQ field providing spatial structure. A second method involves scaling daily CMAQ simulated fields using mean observations to reduce bias. Finally, a weighted average of these results based on prediction of temporal variance provides optimized daily estimates for each 12 × 12 km grid. These methods were applied to daily metrics of 12 pollutants (CO, NO2, NOx, O3, SO2, PM10, PM2.5, and five PM2.5 components) over the state of Georgia for a seven-year period (2002-2008). Cross-validation demonstrates a wide range in optimized model performance across pollutants, with SO2 predicted most poorly due to limitations in coal combustion plume monitoring and modeling. For the other pollutants studied, 54-88% of the spatiotemporal variance (Pearson R(2) from cross-validation) was captured, with ozone and PM2.5 predicted best. The optimized fusion approach developed provides daily spatial field estimates of air pollutant concentrations and uncertainties that are consistent with observations, emissions, and meteorology.

Assessment of neighbourhood-level socioeconomic status as a modifier of air pollution–asthma associations among children in Atlanta

Background A broad literature base provides evidence of association between air pollution and paediatric asthma. Socioeconomic status (SES) may modify these associations; however, previous studies have found inconsistent evidence regarding the role of SES. Methods Effect modification of air pollution–paediatric asthma morbidity by multiple indicators of neighbourhood SES was examined in Atlanta, Georgia. Emergency department (ED) visit data were obtained for 5–18 years old with a diagnosis of asthma in 20-county Atlanta during 2002–2008. Daily ZIP Code Tabulation Area (ZCTA)-level concentrations of ozone, nitrogen dioxide, fine particulate matter and elemental carbon were estimated using ambient monitoring data and emissions-based chemical transport model simulations. Pollutant–asthma associations were estimated using a case-crossover approach, controlling for temporal trends and meteorology. Effect modification by ZCTA-level (neighbourhood) SES was examined via stratification. Results We observed stronger air pollution–paediatric asthma associations in ‘deprivation areas’ (eg, ≥20% of the ZCTA population living in poverty) compared with ‘non-deprivation areas’. When stratifying analyses by quartiles of neighbourhood SES, ORs indicated stronger associations in the highest and lowest SES quartiles and weaker associations among the middle quartiles. Conclusions Our results suggest that neighbourhood-level SES is a factor contributing vulnerability to air pollution-related paediatric asthma morbidity in Atlanta. Children living in low SES environments appear to be especially vulnerable given positive ORs and high underlying asthma ED rates. Inconsistent findings of effect modification among previous studies may be partially explained by choice of SES stratification criteria, and the use of multiplicative models combined with differing baseline risk across SES populations.

Temporally and Spatially Resolved Air Pollution in Georgia Using Fused Ambient Monitor Data and Chemical Transport Model Results

Health data geo-coded with residential coordinates are being used to investigate the relationship between ambient air quality and pediatric emergency department visits in the State of Georgia over the time period 2000–2010. Two types of ambient air quality data – observed concentrations from ambient monitors and predicted concentrations from a chemical transport model (CMAQ) – are being fused to provide spatially resolved daily metrics of five air pollutant gases (CO, NO2, NOx, SO2 and O3) and seven airborne particulate matter measures (PM10, PM2.5, and PM2.5 constituents SO4 2−, NO3 −, NH4 +, EC, OC). The observational data provide reliable temporal trends at and near monitors, but limited spatial information due to the sparse monitoring network; CMAQ data, on the other hand, provide rich spatial information but less reliable temporal information. Four data fusion techniques were applied to provide daily spatial fields of ambient air pollutant concentrations, with data withholding used to evaluate model performance. Two of the data fusion methods were combined to provide results that minimized bias and maximized correlation over time and space with withheld data. Results vary widely across pollutants. These results provide health researchers with complete temporal and spatial air pollutant fields, as well as with temporal and spatial error estimate fields that can be incorporated into health risk models. Future work will apply these methods to five cities for use in ongoing air pollution health studies and to examine strategies for incorporating land use regression variables for spatial downscaling of data fusion results.

Use of Air Quality Modeling Results in Health Effects Research

The most recent Global Burden of Disease study (Lim SS et al, Lancet 380(9859):2224–2260, 2012), for example, finds that combined exposure to ambient and indoor air pollution is one of the top five risks worldwide. Of particular concern is particulate matter (PM). Health researchers are now trying to assess how this mixture of air pollutants links to various health outcomes and how to tie the mixture components and health outcomes back to sources. This process involves the use of air quality models. As part of an EPA Clean Air Research Center, the Southeastern Center for Air Pollution and Epidemiology (SCAPE), a variety of air quality models are being developed and applied to provide enhanced temporal and spatial resolution of pollutant concentrations for use in epidemiologic analysis. Air quality models that are being further developed and used as part of the center include Bayesian-based ensemble methods and hybrid chemical transport-chemical mass balance modeling. The hybrid method uses knowledge of the emissions, modeling and measurement uncertainties, and can provide spatially and temporally complete pollutant fields.

Application and evaluation of two model fusion approaches to obtain ambient air pollutant concentrations at a fine spatial resolution (250m) in Atlanta

Abstract Epidemiologic studies rely on accurately characterizing spatiotemporal variation in air pollutant concentrations. This work presents two model fusion approaches that use publicly available chemical transport simulations, dispersion model simulations, and observations to estimate air pollutant concentrations at a neighborhood-level spatial resolution while incorporating comprehensive chemistry and emissions sources. The first method is additive and the alternative method is multiplicative. These approaches are applied to Atlanta, GA at a 250 m grid resolution to obtain daily 24-hr averaged PM2.5 and 1-hr max CO and NOx concentrations during the years 2003–2008 for use in health studies. The modeled concentrations provide comprehensive estimates with steep spatial gradients near roadways, secondary formation and loss, and effects of regional sources that can influence daily variation in ambient pollutant concentrations. Results show high temporal and spatial correlation and low biases across monitors, providing accurate pollutant concentration estimates for epidemiologic analyses.

Air Quality Model-Based Methods for Estimating Human Exposures: A Review and Comparison

Determining estimates of human exposure is increasingly relying on the use of air quality models and satellite observations to provide spatially and temporally complete pollutant concentration fields. Air quality models, in particular, are attractive as they capture the emissions and meteorological linkages. Additionally they can provide source impact information and concentration fields for a range of species not currently provided from satellite-based observations (e.g., MODIS and MAIAC), and are not subject to cloud interference. Multiple methods based on air quality modeling (including using CMAQ and/or RLINE) with and without data fusion, have been developed and are being used in health studies as part of the EPA-funded Southeastern Center for Air Pollution and Epidemiology Clean Air Research Center. The methods include CMAQ-Data Fusion where concentrations fields are blended with observations to provide spatially and temporally complete pollutant concentrations fields of PM2.5, EC, CO, and NO2. To improve the spatial resolution, this method was extended to include RLINE fields for fine scale (250 m) exposure assessment. Another method was developed to estimate spatial exposure estimates of emissions source categories using CMAQ-derived source impacts for 16–32 sources, along with observations of individual PM species. Each of these approaches have individual strengths and weaknesses. The methods that use a data fusion approach to blend observations and air quality model fields are found to best capture the spatiotemporal trends in the observations, reducing the standard error in the exposure estimates. In the past, such methods were limited by the availability of air quality model fields over long periods, but such fields are becoming more routinely available from air quality forecasting activities.

Using Air Quality Model-Data Fusion Methods for Developing Air Pollutant Exposure Fields and Comparison with Satellite AOD-Derived Fields: Application over North Carolina, USA

A data fusion approach is developed to blend ground-based observations and simulated data from the Community Multiscale Air Quality (CMAQ) model. Spatiotemporal information and finer temporal scale variations have been captured by the resulting fields that are provided by both air quality modeling and observations. The approach is applied to daily PM2.5 total mass, five major particulate species (OC, EC, SO4 2−, NO3 −, and NH4 +), and three gaseous pollutants (CO, NOx, NO2) during 2006–2008 over North Carolina (USA). Applying the data fusion method significantly reduces biases in CMAQ fields to almost zero at monitor locations. The results show improvements in capturing spatial and temporal variability with observations, which is important to health and planning studies. The correlation for the cross-validation test decreased from 0.98 (no withholding) to 0.91 (10% random data withholding) when comparing modeled results to observations. If 10% monitor-based withholding is used, the correlation is 0.91 (random 10% of monitors withheld), and the correlation is 0.88 if spatially-specific withholding is used (10% of monitors withheld are grouped spatially). Results from a satellite-retrieved aerosol optical depth (AOD) method were compared with PM2.5 total mass concentration from data fusion, and the data-fusion fields have slightly less overall error; an R2 of 0.95 compared to 0.81 (AOD). Comparing results from an application of the Integrated Mobile Source Indicator method shows that the data fusion fields can be used to estimate mobile source impacts. Overall, the data fusion approach is attractive for providing spatiotemporal pollutant fields for speciated particulate pollutants, as the demand for accurate, fused, air quality model fields is growing.

Improved Spatiotemporal Source-Based Air Pollutant Mixture Characterization for Health Studies

The growing availability of spatially resolved health data sets (i.e., resident and county level patient records) requires spatially resolved exposure or air quality metrics to investigate the impact of air pollution on health outcomes. While daily air quality data are essential in time-series epidemiologic analysis, the spatial distribution of the observations is limited. Air pollution modeling (i.e., chemical transport modeling (CTM)) addresses this by producing spatially resolved air quality predictions using terrain, emissions and meteorology inputs. However, predicted concentrations may be biased. This work incorporates unique data fusion approaches to combine air quality observations from regulatory monitoring networks (OBS) with the output from a CTM (CMAQ) to generate spatially and temporally resolved gaseous and PM species concentrations. Species concentrations alone cannot directly identify emission sources or characterize pollutant mixtures, therefore source apportionment (SA) models are required to estimate source impacts. The focus of this work is a comparison of SA results for three U.S. regions with differing air pollution sources, St. Louis, Missouri; Atlanta, Georgia; and Dallas-Fort Worth, Texas.