A comparison study of three urban air pollution models

The predictions of three urban air pollution models with varying degrees of mathematical and computational complexities are compared against the hourly SO₂ ground-level concentrations observed on 10 winter nights of the RAPS experiment in St. Louis. The emphasis in this study is on the prediction of urban area source concentrations. Statistics for the paired comparison of predictions of each model with the observations are presented. The RAM and the ATDL model with stable diffusion coefficients overestimated the observed night-time concentrations. The results show that the performance of the ATDL model with near-neutral diffusion coefficients is comparable to the more sophisticated 3-D grid numerical model.     

Catalytic Cigarette Filter for Carbon Monoxide Reduction

A comparative study was performed in order to determine the relative accuracy of a gaussian dispersion model. The U.S. EPA’s RAM (Urban) model was chosen to estimate 24-hour average sulfur dioxide concentrations in the Cleveland, Ohio area. Point and area source emissions, along with a background concentration were included in the modeling effort. Projections from the model made at the ambient air stations were compared to measured sulfur dioxide concentrations. A total of 3020 comparisons were performed at 33 monitoring sites. An analysis of the results illustrates that, on a daily basis, the predictions of the model did not reflect actual air quality. The correlation coefficients of the 24-hour comparisons at the monitoring sites varied from a low of —0.121 to a high of 0.541. When the highest and second highest modeled concentrations were evaluated with respect to the highest and second highest measured concentrations, over a period of a year, a more favorable comparison was observed.     

A Comparative Validation of the RAM and PTMTP Models for Short-Term SO2 Concentrations in Two Urban Areas

A comprehensive and comparative model validation of two EPA models for short-term SO₂ concentrations was performed. The two models tested were RAM (Urban version) and PTMTP (Terrain version). Both are multiple source, multiple receptor gaussian plume models, recommended in the EPA Guideline On Air Quality Models. ¹ The principal difference between the two models is in their use of empirical dispersion coefficients. It was because of the potential for markedly different predicted maximum SO₂ concentrations, and the absence of any testing data on the RAM model, that the validation analysis was undertaken. The current study utilized a full year of air quality data from monitoring sites in two Indiana cities, Michigan City and Indianapolis. Cumulative frequency distributions for each site and model were prepared and comparisons made. The results indicate that the RAM (Urban) model was highly inaccurate in predicting maximum short-term SO₂ concentrations. The PTMTP model, although conservative in its estimates, produces results which more closely resemble the distribution of observed SO₂ concentrations. The body of information presented in this paper is directed to environmental scientists responsible for air quality modeling, and to those persons who set policy on the use of models in air quality studies.     

Comparisons of WRF/Chem simulated O3 concentrations in Mexico City with ground-based RAMA measurements during the MILAGRO period

This work compares the WRF/Chem (Weather Research and Forecasting – Chemistry) simulated O₃ concentrations in the Mexico City Metropolitan Area (MCMA) with measurements from the ground-based RAMA network during the MILAGRO (Megacity Initiative: Local and Global Research Observations) period. The model resolves the observations reasonably well in terms of diurnal cycle and mean magnitude as reflected by high correlation coefficients and low root-mean-square errors. Stations located in the center of the MCMA generally exhibit higher correlation coefficients and lower model biases than those stations located in the peripheral of the MCMA. Large temporal variations in the observed and simulated O₃ concentrations are noted from station to station during the MILAGRO period. Sensitivity analyses of O₃ concentrations to changes in NO_x and VOC emissions rates suggest that O₃ production in the MCMA is VOC-sensitive, and VOC-emissions reduction appears to be an effective strategy for reducing high surface O₃ concentrations in the MCMA.     

Spatial and temporal variability of outdoor coarse particulate matter mass concentrations measured with a new coarse particle sampler during the Detroit Exposure and Aerosol Research Study

The Detroit Exposure and Aerosol Research Study (DEARS) provided data to compare outdoor residential coarse particulate matter (PM_10–2.5) concentrations in six different areas of Detroit with data from a central monitoring site. Daily and seasonal influences on the spatial distribution of PM_10–2.5 during Summer 2006 and Winter 2007 were investigated using data collected with the newly developed coarse particle exposure monitor (CPEM). These data allowed the representativeness of the community monitoring site to be assessed for the greater Detroit metro area. Multiple CPEMs collocated with a dichotomous sampler determined the precision and accuracy of the CPEM PM_10–2.5 and PM_2.5 data.CPEM PM_2.5 concentrations agreed well with the dichotomous sampler data. The slope was 0.97 and the R² was 0.91. CPEM concentrations had an average 23% negative bias and R² of 0.81. The directional nature of the CPEM sampling efficiency due to bluff body effects probably caused the negative CPEM concentration bias.PM_10–2.5 was observed to vary spatially and temporally across Detroit, reflecting the seasonal impact of local sources. Summer PM_10–2.5 was 5 μg m^?3 higher in the two industrial areas near downtown than the average concentrations in other areas of Detroit. An area impacted by vehicular traffic had concentrations 8 μg m^?3 higher than the average concentrations in other parts of Detroit in the winter due to the suspected suspension of road salt. PM_10–2.5 Pearson Correlation Coefficients between monitoring locations varied from 0.03 to 0.76. All summer PM_10–2.5 correlations were greater than 0.28 and statistically significant (p-value < 0.05). Winter PM_10–2.5 correlations greater than 0.33 were statistically significant (p-value < 0.05). The PM_10–2.5 correlations found to be insignificant were associated with the area impacted by mobile sources during the winter. The suspected suspension of road salt from the Southfield Freeway, combined with a very stable atmosphere, caused concentrations to be greater in this area compared to other areas of Detroit. These findings indicated that PM_10–2.5, although correlated in some instances, varies sufficiently across a complex urban airshed that that a central monitoring site may not adequately represent the population's exposure to PM_10–2.5.     

A neural network-based approach for the prediction of urban SO2 concentrations in the Istanbul metropolitan area

A three-layer Artificial Neural Network (ANN) model was developed to forecast air pollution levels. The subsequent SO₂ concentration (24-hour averaged) being the output parameter of this study was estimated by seven input parameters such as preceding SO₂ concentrations (24-hour averaged), average daily temperature, sea-level pressure, relative humidity, cloudiness, average daily wind speed and daily dominant wind direction. After Backpropagation training combined with Principal Component Analysis (PCA), the proposed model predicted subsequent SO₂ values based on measured data. ANN testing outputs were proven to be satisfactory with correlation coefficients of about 0.770, 0.744 and 0.751 for the winter, summer and overall data, respectively.     

Participant-based monitoring of indoor and outdoor nitrogen dioxide,volatile organic compounds,and polycyclic aromatic hydrocarbons among MICA-Air households

The Mechanistic Indicators of Childhood Asthma (MICA) study in Detroit, Michigan introduced a participant-based approach to reduce the resource burden associated with collection of indoor and outdoor residential air sampling data. A subset of participants designated as MICA-Air conducted indoor and outdoor residential sampling of nitrogen dioxide (NO₂), volatile organic compounds (VOCs), and polycyclic aromatic hydrocarbons (PAHs). This participant-based methodology was subsequently adapted for use in the Vanguard phase of the U.S. National Children’s Study. The current paper examines residential indoor and outdoor concentrations of these pollutant species among health study participants in Detroit, Michigan.Pollutants measured under MICA-Air agreed well with other studies and continuous monitoring data collected in Detroit. For example, NO₂ and BTEX concentrations reported for other Detroit area monitoring were generally within 10–15% of indoor and outdoor concentrations measured in MICA-Air households. Outdoor NO₂ concentrations were typically higher than indoor NO₂ concentration among MICA-Air homes, with a median indoor/outdoor (I/O) ratio of 0.6 in homes that were not impacted by environmental tobacco smoke (ETS) during air sampling. Indoor concentrations generally exceeded outdoor concentrations for VOC and PAH species measured among non-ETS homes in the study. I/O ratios for BTEX species (benzene, toluene, ethylbenzene, and m/p- and o-xylene) ranged from 1.2 for benzene to 3.1 for toluene. Outdoor NO₂ concentrations were approximately 4.5 ppb higher on weekdays versus weekends. As expected, I/O ratios pollutants were generally higher for homes impacted by ETS.These findings suggest that participant-based air sampling can provide a cost-effective alternative to technician-based approaches for assessing indoor and outdoor residential air pollution in community health studies. We also introduced a technique for estimating daily concentrations at each home by weighting 2- and 7-day integrated concentrations using continuous measurements from regulatory monitoring sites. This approach may be applied to estimate short-term daily or hourly pollutant concentrations in future health studies.     

Near-road air pollutant concentrations of CO and PM2.5: A comparison of MOBILE6.2/CALINE4 and generalized additive models

The contribution of vehicular traffic to air pollutant concentrations is often difficult to establish. This paper utilizes both time-series and simulation models to estimate vehicle contributions to pollutant levels near roadways. The time-series model used generalized additive models (GAMs) and fitted pollutant observations to traffic counts and meteorological variables. A one year period (2004) was analyzed on a seasonal basis using hourly measurements of carbon monoxide (CO) and particulate matter less than 2.5 μm in diameter (PM_2.5) monitored near a major highway in Detroit, Michigan, along with hourly traffic counts and local meteorological data. Traffic counts showed statistically significant and approximately linear relationships with CO concentrations in fall, and piecewise linear relationships in spring, summer and winter. The same period was simulated using emission and dispersion models (Motor Vehicle Emissions Factor Model/MOBILE6.2; California Line Source Dispersion Model/CALINE4). CO emissions derived from the GAM were similar, on average, to those estimated by MOBILE6.2. The same analyses for PM_2.5 showed that GAM emission estimates were much higher (by 4–5 times) than the dispersion model results, and that the traffic-PM_2.5 relationship varied seasonally. This analysis suggests that the simulation model performed reasonably well for CO, but it significantly underestimated PM_2.5 concentrations, a likely result of underestimating PM_2.5 emission factors. Comparisons between statistical and simulation models can help identify model deficiencies and improve estimates of vehicle emissions and near-road air quality.     

NMOC,ozone, and organic aerosol in the southeastern United States, 1999–2007: 2. Ozone trends and sensitivity to NMOC emissions in Atlanta,Georgia

Non-methane organic carbon (NMOC) measurements made in Atlanta, Georgia from 1999–2007 are used with nitrogen oxide (NO_x or NO_y) and ozone (O₃) data to investigate relationships between O₃ precursors and peak 8-hour O₃ concentrations in the city. Data from a WNW-to-ENE transect of sites illustrate that the mean urban peak 8-hour O₃ excess constitutes about 20% of the peak 8-hour O₃ measured at the area-wide maximum O₃ site when air-mass movement is from the northwest quadrant; local influence is potentially greater on days with more stagnation or recirculation. The peak 8-hour O₃ concentrations in Atlanta increase as (1) surface temperature (T), ambient NMOC and NO_y concentrations, and previous-day peak O₃ concentrations increase, and as (2) relative humidity, surface wind speeds, and ratios of NMOC-to-NO_y decrease. An observation-based statistical model is introduced to relate area-wide peak 8-hour O₃ concentrations to ambient NMOC and NO_y concentrations, while accounting for the non-linear dependences of peak 8-hour O₃ concentrations on meteorological factors. On the majority of days when the area-wide peak 8-hour O₃ exceeds 75 ppbv, meteorologically-adjusted peak 8-hour O₃ concentrations increase as ambient NMOC concentrations increase (NMOC sensitive) and ambient NO_y concentrations decrease. This result contrasts with regional conditions in which O₃ formation appears to be NO_x-sensitive in character. The results offer observationally-based information of relevance to O₃ management strategies in the Atlanta area, potentially contributing to “weight-of-evidence” assessments.     

Pragmatic estimation of a spatio-temporal air quality model with irregular monitoring data

Statistical analyses of health effects of air pollution have increasingly used GIS-based covariates for prediction of ambient air quality in “land use” regression models. More recently these spatial regression models have accounted for spatial correlation structure in combining monitoring data with land use covariates. We present a flexible spatio-temporal modeling framework and pragmatic, multi-step estimation procedure that accommodates essentially arbitrary patterns of missing data with respect to an ideally complete space by time matrix of observations on a network of monitoring sites. The methodology incorporates a model for smooth temporal trends with coefficients varying in space according to Partial Least Squares regressions on a large set of geographic covariates and nonstationary modeling of spatio-temporal residuals from these regressions. This work was developed to provide spatial point predictions of PM_2.5 concentrations for the Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air) using irregular monitoring data derived from the AQS regulatory monitoring network and supplemental short-time scale monitoring campaigns conducted to better predict intra-urban variation in air quality. We demonstrate the interpretation and accuracy of this methodology in modeling data from 2000 through 2006 in six U.S. metropolitan areas and establish a basis for likelihood-based estimation.     

The CCMS air pollution model intercomparison study

Four air pollution transport models were tested and compared in an area of ~ 400 × 400 km². Three models were Eulcrian grid models, the fourth a Lagrangian trajectory model. The data base (emissions and meteorological observations) were essentially the same for all models. Differences in model output could only be a result of the different (numerical) structure of the models and of the differences in processing of the meteorological data. It turned out that the latter was the major source of differences in model results. Generally there was a satisfactory correlation between model results and observed concentrations. Mainly due to the negligence of transport of pollutants into the modelling region, predicted concentrations were considerably lower than the observed.     

Numerical simulation of air pollution in urban areas: model performance

The numerical model developed in the first part of this investigation is applied to assess the behavior of sulfur dioxide and sulfate concentration distributions in an urban area using the St Louis Regional Air Pollution Study (RAPS) data. Statistical techniques chosen to determine the accuracy and uncertainty associated with the numerical model results include paired analysis and resampling analysis. The results of the numerical model are also compared with those of RAM, a Gaussian plume model. Finally, the behavior of point and area emission sources in an urban area is assessed to provide an insight into the complex interrelationships between the emissions and meteorological conditions which determine the distribution of ground level concentrations.     

Meteorological Factors of Ozone Predictability at Houston,Texas

ABSTRACT

Several ozone modeling approaches were investigated to determine if uncertainties in the meteorological data would be sufficiently large to limit the application of physically realistic ozone (O₃) forecast models. Three diagnostic schemes were evaluated for the period of May through September 1997 for Houston, TX. Correlations between measured daily maximum and model calculated O₃ air concentrations were found to be 0.70 using a linear regression model, 0.65 using a non-advective box model, and 0.49 using a three-dimensional (3-D) transport and dispersion model. Although the regression model had the highest correlation, it showed substantial underestimates of the highest concentrations. The box model results were the most similar to the regression model and did not show as much underestimation. The more complex 3-D modeling approach yielded the worst results, likely resulting from O₃ maxima that were driven by local factors rather than by the transport of pollutants from outside of the Houston domain. The highest O₃ concentrations at Houston were associated with light winds and meandering trajectories. A comparison of the gridded meteorological data used by the 3-D model to the observations showed that the wind direction and speed values at Houston differed most on those days on which the O₃ underestima-tions were the greatest. These periods also tended to correspond with poor precipitation and temperature estimates. It is concluded that better results are not just obtained through additional modeling complexity, but there needs to be a comparable increase in the accuracy of the meteorological data.     

Predicting ozone levels

This paper presents a statistical model that is capable of predicting ozone levels from precursor concentrations and meteorological conditions during daylight hours in the Shuaiba Industrial Area (SIA) of Kuwait. The model has been developed from ambient air quality data that was recorded for one year starting from December 1994 using an air pollution mobile monitoring station. The functional relationship between ozone level and the various independent variables has been determined by using a stepwise multiple regression modelling procedure. The model contains two terms that describe the dependence of ozone on nitrogen oxides (NOx) and nonmethane hydrocarbon precursor concentrations, and other terms that relate to wind direction, wind speed, sulphur dioxide (SO₂) and solar energy. In the model, the levels of the precursors are inversely related to ozone concentration, whereas SO₂ concentration, wind speed and solar radiation are positively correlated. Typically, 63 % of the variation in ozone levels can be explained by the levels of NOx. The model is shown to be statistically significant and model predictions and experimental observations are shown to be consistent. A detailed analysis of the ozone-temperature relationship is also presented; at temperatures less than 27 °C there is a positive correlation between temperature and ozone concentration whereas at temperatures greater than 27 °C a negative correlation is seen. This is the first time a non-monotonic relationship between ozone levels and temperature has been reported and discussed.     

COMPLEX I and II Model Performance Evaluation in Nevada and New Mexico

The COMPLEX I and COMPLEX II Gaussian dispersion models for complex terrain applications have been made available by EPA. Various terrain treatment options under IOPT(25) can be selected for a particular application, one of which [IOPT(25) = 1] is an algorithm similar to that of the VALLEY model. A model performance evaluation exercise involving three of the available options with both COMPLEX models was carried out using SF⁶ tracer measurements taken during worst-case stable impaction conditions in complex terrain at the Harry Allen Plant site in southern Nevada. The models did not reproduce observed concentrations on an event by event basis, as correlation coefficients for 1-h concentrations of 0-0.3 were exhibited. When observed and calculated cumulative frequency distributions for 1-h and 3-h concentrations were compared, a close correspondence between observations and concentrations calculated with COMPLEX I, IOPT(25) = 2 or 3 was noted; both options consistently overestimated observed concentrations. With IOPT(25) = 1, upper percentile (maximum) values in the calculated frequency distribution exceeded the corresponding IOPT(25) = 2 or 3 value by roughly a factor of 2, and observed values by 2.5-5. COMPLEX II typically produced maximum values 2-4 times as great as COMPLEX I for the same terrain treatment option. From these results it is concluded that: 1) the physically unrealistic sector-spread approach used in VALLEY and COMPLEX I under stable impaction conditions is a surrogate for wind direction variation, and 2) the doubling of the plume centerline concentration due to ground reflection under terrain impingement conditions that is included in IOPT(25) = 1 is inappropriate.

These findings were found to be consistent with an analysis of noncurrent observed and calculated SO₂ χ/Q frequency distributions for 1, 3, and 24 hours near the Four Corners Plant in New Mexico. The comparison involved a four-year calculated χ/Q data set and a two-year observed χ/Q data set at the worst-case high terrain impact location near the plant.     

Regression modelling of hourly NOx and NO2 concentrations in urban air in London

Based on hourly measurements of NO_x NO₂ and O₃ and meteorological data, an ordinary least squares (OLS) model and a first-order autocorrelation (AR) model were developed to analyse the regression and prediction of NO_x and NO₂ concentrations in London. Primary emissions and wind speed are the most important factors influencing NO_x concentrations; in addition to these two, reaction of NO with O₃ is also a major factor influencing NO₂ concentrations. The AR model resulted in high correlation coefficients (R > 0.95) for the NO_x and NO₂ regression based on a whole year's data, and is capable of predicting NO₂ (R = 0.83) and NO_x (R = 0.65) concentrations when the explanatory variables were available. The analysis of the structure of regression models by Principal Component Analysis (PCA) indicates that the regression models are stable. The results of the OLS model indicate that there was an exceptional NO₂ source, other than primary emission and reaction of NO with O₃, in the air pollution episode in London in December 1991.     

Comparison of Predicted and Measured Concentrations for 58 Alternative models of Plume Transport in Complex Terrain

The purpose of this study was to evaluate alternative prediction models for the SO₂ concentrations produced in the vicinity of the Ohio Edison Company Sammis Power Plant. The plant is situated in the northeastern portion of the Ohio River Valley in complex terrain. Comparisons of the 16 highest predicted and measured short-term SO₂ concentrations were conducted for a one year period for 58 alternative models. Several models were found to predict reasonably accurately the 16 highest measured 24-hour SO₂ concentrations. Each of these models requires an upward adjustment in the plume centerline location as the plume is transported downwind in rising terrain. These same models overpredict by substantial margins the 16 highest measured 3-hour SO₂ concentrations. Improvements in emissions inventory data and improvements in the prediction models used are believed necessary to increase prediction accuracy further.     

Modeling Inorganic Aerosols and Their Response to Changes in Precursor Concentration in Mexico City

Abstract

Based on data from the 1997 Investigación sobre Materia Particulada y Deterioro Atmosférico-Aerosol and Visibility Evaluation Research (IMADA-EVER) campaign and the inorganic aerosol model ISORROPIA, the response of inorganic aerosols to changes in precursor concentrations was calculated. The aerosol behavior is dominated by the abundance of ammonia and thus, changes in ammonia concentration are expected to have a small effect on particle concentrations. Changes in sulfate and nitrate are expected to lead to proportional reductions in inorganic fine particulate matter (PM_2.5). Comparing the predictions of ISORROPIA with the observations, the lowest bias and error are achieved when the aerosols are assumed to be in the efflorescence branch. Including crustal species reduces the bias and error for nitrate but does not improve overall model performance. The estimated response of inorganic PM_2.5 to changes in precursor concentrations is affected by the inclusion of crustal species in some cases, although average responses are comparable with and without crustal species. Observed concentrations of particle chloride suggest that gas phase concentrations of hydrogen chloride may not be negligible, and future measurement campaigns should include observations to test this hypothesis. Our ability to model aerosol behavior in Mexico City and, thus, design control strategies, is constrained primarily by a lack of observations of gas phase precursors. Future campaigns should focus in particular on better understanding the temporal and spatial distribution of ammonia concentrations. In addition, gas phase observations of nitric acid are needed, and a measure of particle water content will allow stable versus metastable aerosol behavior to be distinguished.     

Lead Concentrations in Detroit,New York,and Los Angeles Air

Lead concentrations in air were measured at 12 sites in Detroit, New York and Los Angeles as part of a program to relate automobile emissions and polynuclear aromatic hydrocarbons in air. The information on lead is reported separately because of the current interest in lead as an air pollutant. Sampling was conducted by means of a large “absolute” filter and equipment contained in a step-van truck. A portion of the filter was macerated in nitric acid and the lead determined spectrographically. The combined annual average lead concentration for four sites in metropolitan Los Angeles was approximately 40% higher than the combined averages of either the five sites in metropolitan New York or the three sites in metropolitan Detroit. Concentrations ranged from 0.4 ug/M³ at Santa Monica, to 18.4 ug/M³ at a Los Angeles Freeway Interchange. Concentrations were generally highest in freeway areas, intermediate in commercial areas, and lowest in residential areas. They were about 40% higher in daytime than at night. Average lead concentrations were highest during autumn in New York and winter in Los Angeles reflecting an inverse relationship with wind speed. Correlation coefficients between lead and carbon monoxide, at all sites, were statistically non-zero with 99% confidence and varied from 0.75 to 0.96. Lead concentrations in this study were higher than concentrations reported by others for Detroit, New York, and Los Angeles, presumably because sampling in this study was closer to traffic. However, concentrations in this study were lower than in-traffic concentrations given in the literature.     

Neural network forecasting of air pollutants hourly concentrations using optimised temporal averages of meteorological variables and pollutant concentrations

The new method for the forecasting hourly concentrations of air pollutants is presented in the paper. The method was developed for a site in urban residential area in city of Zagreb, Croatia, for four air pollutants (NO₂, O₃, CO and PM₁₀). Meteorological variables and concentrations of the respective pollutant were taken as predictors. A novel approach, based on families of univariate regression models, was employed in selecting the averaging intervals for input variables. For each variable and each averaging period between 1 and 97 h, a separate model was built. By inspecting values of the coefficient of correlation between measured and modelled concentrations, optimal averaging periods for each variable were selected. A new dataset for building the forecasting model was then calculated as temporal moving averages (running means) of former variables. A multi-layer perceptron type of neural networks is used as the forecasting model. Index of agreement, calculated for the entire dataset including the data for model building, ranged from 0.91 to 0.97 for the respective pollutants. As suggested by the analysis of the relative importance of the input variables, different agreements for different pollutants are likely due to different sources and production mechanisms of investigated pollutants. A comparison of the new method with more traditional method, which takes hourly averages of the forecast hour as input variables, showed similar or better performance. The model was developed for the purpose of public-health-oriented air quality forecasting, aiming to use a numerical weather forecast model for the prediction of the part of input data yet unknown at the forecasting time. It is to expect that longer term averages used as inputs in the proposed method will contribute to smaller input errors and the greater accuracy of the model.