Evaluation of Regional-Scale Receptor Modeling
The ability of receptor models to estimate regional contributions to fine particulate matter (PM2.5) was assessed with synthetic, speciated datasets at Brigantine National Wildlife Refuge (BRIG) in New Jersey and Great Smoky Mountains National Park (GRSM) in Tennessee. Synthetic PM2.5 chemical concentrations were generated for the summer of 2002 using the Community Multiscale Air Quality (CMAQ) model and chemically speciated PM2.5 source profiles from the U.S. Environmental Protection Agency (EPA)â€™s SPECIATE and Desert Research Instituteâ€™s source profile databases. CMAQ estimated the â€œtrueâ€ contributions of seven regions in the eastern United States to chemical species concentrations and individual source contributions to primary PM2.5 at both sites. A seven-factor solution by the positive matrix factorization (PMF) receptor model explained approximately 99% of the variability in the data at both sites. At BRIG, PMF captured the first four major contributing sources (including a secondary sulfate factor), although diesel and gasoline vehicle contributions were not separated. However, at GRSM, the resolved factors did not correspond well to major PM2.5 sources. There were no correlations between PMF factors and regional contributions to sulfate at either site. Unmix produced five- and seven-factor solutions, including a secondary sulfate factor, at both sites. Some PMF factors were combined or missing in the Unmix factors. The trajectory mass balance regression (TMBR) model apportioned sulfate concentrations to the seven source regions using Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) trajectories based on Meteorological Model Version 5 (MM5) and Eta Data Simulation System (EDAS) meteorological input. The largest estimated sulfate contributions at both sites were from the local regions; this agreed qualitatively with the true regional apportionments. Estimated regional contributions depended on the starting elevation of the trajectories and on the meteorological input data.
Regression-Based Oxides of Nitrogen Predictors for Three Nitrogen Predictors for Three
Models of diesel engine emissions such as oxides of nitrogen (NOx) are valuable when they can predict instantaneous values because they can be incorporated into whole vehicle models, support inventory predictions, and assist in developing superior engine and aftertreatment control strategies. Recent model-year diesel engines using multiple injection strategies, exhaust gas recirculation, and variable geometry turbocharging may have more transient sensitivity and demand more sophisticated modeling than for legacy engines. Emissions data from 1992, 1999, and 2004 model-year U.S. truck engines were modeled separately using a linear approach (with transient terms) and multivariate adaptive regression splines (MARS), an adaptive piece-wise regression approach that has limited prior use for emissions prediction. Six input variables based on torque, speed, power, and their derivatives were used for MARS. Emissions time delay was considered for both models. Manifold air temperature (MAT) and manifold air pressure (MAP) were further used in NOx modeling to build a plug-in model. The predictive performance for instantaneous NOx on part of the certification transient test procedure (Federal Test Procedure [FTP]) of the 2004 engine MARS was lower (R2 = 0.949) than the performance for the 1992 (R2 = 0.981) and 1999 (R2 = 0.988) engines. Linear regression performed similarly for the 1992 and 1999 engines but performed poorly (R2 = 0.896) for the 2004 engine. The MARS performance varied substantially when data from different cycles were used. Overall, the MAP and MAT plug-in model trained by MARS was the best, but the performance differences between LR and MARS were not substantial.
Review of Recent Advances in Detection of Organic Markers in Fine Particulate Matter and Their Use for Source Apportionment
Fine particulate matter is believed to be more toxic than coarse particles and to exacerbate health problems such as respiratory and cardiopulmonary diseases. Specific organic compounds within atmospheric fine particulate material can be used to differentiate specific inputs from various emissions and thus is helpful in identifying the major urban air pollution sources that contribute to these health problems. Particular marker compounds that carry signature information about different emission sources (i.e., gasoline or diesel motor vehicles, wood smoke, meat cooking, vegetative detritus, and cigarette smoke) are reviewed. Aerosol organic types (e.g., from mass spectrometry data, which can also help in elucidation of carbonaceous material sources) are also discussed. Apportionment of the primary source contributions and atmospheric processes contributing to fine particulate matter and fine particulate organic material concentrations are outlined. This review provides an overview of the latest developments in chemical characterization approaches for identification and quantification of compounds in complex organic mixtures associated with fine atmospheric particles and their use in chemical mass balance (CMB) and positive matrix factorization (PMF) source apportionment models.
Toward Effective Source Apportionment Using Positive Matrix Factorization: Experiments with Simulated PM2.5 Data
To elucidate the relationship between factors resolved by the positive matrix factorization (PMF) receptor model and actual emission sources and to refine the PMF modeling strategy, speciated PM2.5 (particulate matter with aerodynamic diameter 2.5 m) data generated from a state-of-the-art chemical transport model for two rural sites in the eastern United States are subjected to PMF analysis. In addition to 2 and R2 used to infer the quality of fitting, the interpretability of PMF factors with respect to known primary and secondary sources is evaluated using a root mean square difference analysis. For the most part, factors are found to represent imperfect combinations of sources, and the optimal number of factors should be just adequate to explain the input data (e.g., R2 0.95). Retaining more factors in the model does not help resolve minor sources, unless temporal resolution of the data is increased, thus allowing more information to be used by the model. If guided with a priori knowledge of source markers and/or special events, rotation of factors leads to more interpretable PMF factors. The choice of uncertainty weighting coefficients greatly influences the PMF modeling results, but it cannot usually be determined for simulated or real-world data. A simple test is recommended to check whether the weighting coefficients are suitable. However, uncertainties in the data divert PMF solutions even when the optimal weighting coefficients and number of factors are in place.