Airshed Model Research Articles

Three-dimensional modelling of photochemical air pollution in Switzerland and comparison with a multi-parcel Lagrangian model

The photochemical smog situation in Switzerland on 29 July, 1993 was simulated using the 3–D grid model UAM (Urban Airshed Model) and the results were compared with those of a multi–parcel Lagrangian model. Sensitivity analyses concerning initial conditions, emissions, deposition, biogenic volatile organic compounds (VOC), and emission reductions were carried out. Results were compared with field measurements. Calculations showed that although ozone (O3) formation in Switzerland is mainly NOx (NO + NO2) limited, some areas seem also to be sensitive to VOC emissions. According to the calculations, reducing anthropogenic VOC emissions by 20–80% would lead to a decrease of 5–25% in O3 levels in a few regions, whereas the calculated decrease in O3 concentrations by NOx emission reductions was 6–38% and is expected to be effective over a larger area. Impacts of emission reductions on other species, such as peroxyacetyl nitrate (PAN), hydrogen peroxide (H2O2), nitric acid (HNO3) and formaldehyde (HCHO) were investigated. A correlation was found between the sensitivity of O3 to emission reductions, and NOz (NOy–NOx, where NOy = NOx + PAN + HNO3 + other nitrates) at the time of peak O3. VOC controls seem to be effective only at NOz concentrations higher than about 2 ppb, whereas NOx controls are effective also below this range. Exclusion of biogenic VOC decreased the calculated O3 concentrations by only 5% in certain regions. PAN and H2O2 concentrations, however, decreased significantly. Differences and similarities between the results of this study and those of a multi–parcel Lagrangian model are discussed, and criteria are outlined to decide on the choice for specific problems.

Spatial inhomogeneity in pollutant concentrations, and their implications for air quality model evaluation

The Southern California Air Quality Study database provides a valuable resource with which to test urban-scale photochemical models and to achieve a better understanding of the atmospheric dynamics of pollutant formation. The CIT model was evaluated using the SCAQS database according to traditional model performance guidelines. A first application, reported previously, focused on model enhancement and application of the model to the 27–29 August 1987 episode. This study evaluates the CIT model using the 24–25 June SCAQS episode, providing further evaluation of the model. Results show that the CIT airshed model can follow the diurnal variations of reactive species and the transport for relatively unreactive species. The normalized gross error for ozone was 31 % in June compared to 38% in August. However, to fully judge model performance in proper perspective, a question arises: “How well do the measurements reflect the air quality surrounding the monitoring station, not just in that location?” This is an important but seldom quantitatively considered factor, not only in model evaluation but in the study of health effects as well. Analyses indicate that individual concentration measurements only approximately represent the true volume-averaged concentrations within a computational grid cell and that significant spatial variations exist. Thus any evaluation of models using these data sets should take these local variations into consideration. A series of tests found that the local inhomogeneities had a normalized gross error in the range of 25–45% depending on the pollutant. In this context, the performance of the CIT model is consistent with known modeling limitations such as emissions inventories and sub-grid scale variation of observations.

Inverse modeling of biogenic isoprene emissions

Using current biogenic emission estimates, Urban Airshed Model simulations for Atlanta, Georgia substantially underpredict isoprene concentrations relative to observations. In this work, an inverse method is used to determine the biogenic isoprene emissions for Atlanta that minimize the difference between the model‐simulated and the observed isoprene concentrations. The resulting isoprene emissions are 2 to 10 times higher than any of the accepted emission estimates. Overall, the concurrent ozone simulation improved when the biogenic isoprene emissions were increased. These higher isoprene emissions significantly increase the concentration of model‐simulated ozone within the plumes of nitrogen oxides emitted from large point sources. These results should be considered when developing control strategies for urban ozone.

Sensitivity of Urban Airshed Model (UAM-IV) Calculated Air Pollutant Concentrations to the Vertical Diffusion Parameterization during Convective Meteorological Situations

It is shown that Urban Airshed Model (UAM-IV) calculated air pollutant concentrations during photochemical smog episodes in Atlanta, Georgia, depend strongly on the numerical parameterization of the daytime vertical diffusivity. Results found suggest that vertical mixing is overestimated by the UAM-IV during unstable daytime conditions, as calculated vertical diffusivity values exceed measured and comparable literature values. Although deviations between measured and UAM-IV calculated air pollutant concentrations may only in part be due to the UAM-IV diffusivity parameterization, results indicate the large error potential in vertical diffusivity parameterization. Easily implemented enhancements to UAM-IV algorithms are proposed, thus improving UAM-IV modeling performance during unstable stratification.

Development of ozone-precursor relationships using VOC receptor modeling

The results of a summer 1988 VOC receptor modeling study in Detroit were used in conjunction with a photochemical model to examine the effect of VOC source contributions on predicted ambient ozone. Source hydrocarbon contributions determined from receptor modeling were input to the ozone model along with initial measured concentrations of CO, NO x, O 3 and with measured temperature, uv, pressure and relative humidity data. Wind trajectories were used to compare the predicted ozone concentration maximums against measured ozone maximums at downwind ozone monitoring stations. The r 2 value between observed and calculated downwind ozone was 0.38 and 0.52, for the 6 a.m. and 8 a.m. starting times, respectively, and the slopes of the least squares fit lines were 0.86 and 1.05, respectively, indicating unbiased results. These comparisons are well within the accuracy guidelines specified by the U.S. Environment Protection Agency for ozone prediction models, and are also within the accuracy of similar comparisons produced from Urban Airshed Model applications. Source sensitivity analysis was carried out to see the effect of reducing source emissions on ozone formation. This modeling approach demonstrates that accurate ozone predictive modeling does not have to rely on reported source emission inventories.

Comparing prognostic and diagnostic meteorological fields and their impacts on photochemical air quality modeling

Photochemical air quality models are being used increasingly to make policy decisions. Meteorological data are one of the key inputs to these models. Currently, there are two primary techniques used to generate the gridded meteorological data for input to the air quality models: diagnostic analysis and prognostic modeling. Meteorological fields developed for the Southern California region using these two techniques are compared in two ways. First, the fields are compared directly to each other and with observed data. Second, the meteorological fields are used in photochemical airshed model simulations. The second test is used to investigate how the development of meteorological fields affects photochemical model performance, and how these fields impact control strategy evaluation. The results obtained from using prognostically-derived fields as input to a photochemical model are similar to those obtained by using diagnostically-derived fields, however, they show a lower peak ozone prediction. The reasons for these results are analyzed. The viability of using prognostic modeling to develop meteorological input to the air quality models is discussed.

Preliminary sensitivity analysis of urban airshed model simulations to temporal and spatial availability of boundary layer wind measurements

The 1992 Atlanta Field Intensive of the Southern Oxidants Research Program on Ozone Non-Attainment (SORP-ONA) provided a unique data set of urban meteorological measurements. The data were used to investigate the sensitivity of Urban Airshed Model (UAM-IV) results to the spatial and temporal resolution of upper-air meteorological measurements. RMS differences and average absolute deviations for winds and model calculated NO y and O 3 values were computed for a variety of measurement strategies which differed in the spatial and temporal resolution of the upper-air data. A Lagrangian Particle Dispersion Model (LPDM) was used to study the movement of the plumes from the Atlanta urban core and an elevated power plant located northwest of Atlanta. The results show that placement of upper-air measurements at or near the center of emissions density should be highest priority of field measurements campaigns.

Effects of uncertainties in meteorological inputs on urban airshed model predictions and ozone control strategies

Although well-recognized within the photochemical modeling community, the effect of uncertainties in meteorological input on the urban airshed model (UAM) output has not been systematically evaluated. In this study, the UAM has been applied to investigate the sensitivity of ozone predictions to the choices in wind fields and mixing height profiles for the data-sparse New York metropolitan area. A set of three wind fields, in combination with spatially varying and spatially invariant mixing heights, is investigated for the July 1988 ozone episode. In general, model-predicted ozone levels were higher under the spatially varying mixing height (SVM) option than under the spatially invariant mixing height (SIM) option. SVM based UAM simulations provided better agreement between the predicted and measured ozone concentrations than SIM-based UAM simulations. However, from the regulatory standpoint, predicted ozone concentrations based on either of these mixing height options are within the range considered as acceptable. UAM simulations with emission reductions of 75% NO x and 25% VOCs (NO x -focused) reveal that the improvement in peak ozone levels under the SIM option is larger than that under the SVM option, whereas the emission reduction scenario of 25% NO x and 75% VOCs (VOC-focused) yields greater improvement in peak ozone under the SVM option than with the SIM option. Given the strong influence of mixing heights and wind fields on UAM model predictions in data-sparse areas, it is imperative that uncertainties in development of ozone abatement plans be quantified.

Air Quality Predictions of the Urban Airshed Model Containing Improved Advection and Chemistry Algorithms

We have carried out an extensive study on the effect of replacing the advection and chemistry algorithms of the Urban Airshed Model (UAM) with more accurate algorithms, using the SCAQS episode of August 26−28, 1987, and input data supplied by the California Air Resources Board. Replacing the UAM chemistry solver by the implicit−explicit hybrid solver (IEH) tends to lower the afternoon ozone concentrations somewhat, the afternoon PAN concentrations more noticeably, and the H2O2 concentrations significantly. IEH is also much less likely to have a convergence failure than the UAM solver. Replacing the Smolarkiewicz scheme by the forward Euler Taylor Galerkin scheme (FETG) yields a greater impact. The replacement increases the local peak NO concentrations and lowers the local peak ozone and CO concentrations significantly. Replacing both chemistry and advection algorithms lowers the peak ozone, PAN, and H2O2 concentrations quite noticeably. For example, the predicted peak ozone for August 28 is reduced from 234 to 200 ppb. The complicated wind field for this episode generates two high-ozone subdomains in the modelone in the east and one in the northwest. Reducing the NOx emission by 50% shifts the predicted domainwide peak ozone from the east to the northwest. If we disregard this shift, then the use of more accurate modules in the UAM leads to a 50% increase in the domainwide peak ozone while the original UAM yields a 27% increase. Both the original and improved UAM versions give a 20% increase in peak hourly ozone and a 30% increase in peak 8-h ozone in the eastern subdomain, but much greater and vastly different increases for each of the peak 1-h and peak 8-h ozone concentrations in the northwestern subdomain. The best available numerical methods should be used in modeling a complex system like air quality so that potential hidden errors may be identified and the design of air quality control may be more reliable.

Geographic Information System Framework for Modeling Mobile-Source Emissions

Because vehicle activities and the emissions associated with them can be correlated with specific points in time and space, the modeling capabilities of a geographic information system (GIS) are well suited to the modeling of mobile-source emissions. A GIS-based modeling approach can provide emissions estimates for both on-network and off-network vehicle activities on a modal basis (as a function of various vehicle operating modes that significantly affect vehicle emission rates). An entire metropolitan region can be modeled on a zone, link, and point basis. Vehicle subfleet composition can be tracked or estimated and combined with estimates of vehicle activity and characteristic operating modes to provide emissions estimates in a spatial and temporal context. Emissions from various modal activities are aggregated into grid cells to be used as input to an airshed model. Identifying spatial and temporal distributions of these activities adds to a greater understanding of emissions impacts. A research proto...

Geographic Information System Framework for Modeling Mobile-Source Emissions

Because vehicle activities and the emissions associated with them can be correlated with specific points in time and space, the modeling capabilities of a geographic information system (GIS) are well suited to the modeling of mobile-source emissions. A GIS-based modeling approach can provide emissions estimates for both on-network and off-network vehicle activities on a modal basis (as a function of various vehicle operating modes that significantly affect vehicle emission rates). An entire metropolitan region can be modeled on a zone, link, and point basis. Vehicle subfleet composition can be tracked or estimated and combined with estimates of vehicle activity and characteristic operating modes to provide emissions estimates in a spatial and temporal context. Emissions from various modal activities are aggregated into grid cells to be used as input to an airshed model. Identifying spatial and temporal distributions of these activities adds to a greater understanding of emissions impacts. A research prototype of this modeling concept was produced to illustrate its capabilities and identify difficulties that must be addressed in the development of a fully functional model. The GIS-based displays and hard-copy maps that present the spatial variability of emissions levels help improve the communication of research findings to decision makers and the public. The effort being undertaken at Georgia Institute of Technology in conjunction with the Environmental Protection Agency to develop a next-generation modal emissions model within a GIS framework is described.

Quantification of Individual VOC Reactivity Using a Chemically Detailed, Three-Dimensional Photochemical Model

Urban ozone reduction strategies targeting volatile organic compounds (VOCs) have focused primarily in controlling the total mass of VOC emitted, neglecting the variation in potential ozone formation by individual species. This variation in ozone potential, termed reactivity, is examined here using a three-dimensional Eulerian photochemical airshed model with a detailed chemical mechanism. The Eulerian modeling results are compared with those of a similar study performed using a zero-dimensional model, which is the basis for reactivity quantification for alternative fuel regulations by California Air Resources Board. The results were well correlated between models for metrics calculated at similar precursor ratios; however, notable differences in reactivity were predicted for some important species, probably due to the multi-day simulation periods and the inclusion of cloud cover by the airshed model. 21 refs., 4 figs., 5 tabs.

Effect of alternative boundary conditions on predicted ozone control strategy performance: A case study in the Los Angeles area

The purpose of this paper is to illuminate the importance of assumptions made regarding boundary and initial conditions on the predicted performance of regional ozone control strategies. A computationally efficient approach to depicting the response of an air basin to emission controls is developed. The problem of ozone isopleth generation is addressed using a large Eulerian grid-based photochemical airshed model that is distributed over a grid system of 64 different ROG and NO x control combination points that were run simultaneously on a parallel computer. This method is used for the Los Angeles area to examine the effect on predicted ozone concentrations of alternative assumptions about how boundary conditions at the edge of the air basin will change as a result of changed recirculation of pollutants from within the airshed plus emission decreases upwind. We show that an accurate forecast of the effect of future emission control programs on pollutant inflows across the boundaries of the Los Angeles modeling region is absolutely critical to selection of a successful ozone control strategy for the Los Angeles area.

Vertical sampling and analysis of nonmethane hydrocarbons for ozone control in urban North Carolina

As part of an effort by the state of North Carolina to develop a State Implementation Plan for ozone control in the Raleigh Metropolitan Statistical Area (MSA), vertical measurements of C2‐C10 hydrocarbons were made, in and above the surface inversion layer (SIL), as inputs to the urban airshed model (UAM). Three‐hour integrated ambient air samples were collected during August 1993 from 0500–0800 eastern daylight time (EDT). Additional samples were collected from 1200–1500 and 1700–2000 EDT on selected days. Vertical sampling was achieved from a 610‐m television tower located approximately 15 km southeast of the downtown area. Boundary layer wind and temperature profiles were determined by balloon soundings. For some compounds, e.g., propane, the average concentration was lower above the inversion layer (9.59 ppbC in the SIL and 2.08 ppbC above the SIL); however, other species such as 2‐methylpentane had higher concentrations above the inversion layer (1.28 ppbC in and 1.40 ppbC above the SIL). In addition, the vertical distributions of hydrocarbons within the convective boundary layer were compared to the vertical distribution estimated from calculations based on surface concentration, species reactivity, and eddy diffusivity. Compounds such as isoprene and N‐butane decreased as predicted by the equation, while others such as propane and benzene showed unexpected profiles. Calculations of propylene‐equivalent concentrations were used to estimate the effect of reactivity on the relative importance of individual hydrocarbons. The contribution of isoprene to the local hydrocarbon budget was analyzed and surface measurements were compared with data collected in Atlanta, Georgia, during the 1992 Southern Oxidants Study. Isoprene comprised more than 70% of the total propylene‐equivalent concentration in the afternoon in Raleigh but only 40% of the total in Atlanta.

Sensitivity of urban airshed model results for test fuels to uncertainties in light-duty vehicle and biogenic emissions and alternative chemical mechanisms—Auto/oil air quality improvement research program

Three sources of uncertainty in the air quality modeling performed for the Auto/Oil Air Quality Improvement: Research Program, Phase I, were investigated to assess their impact on predicted ozone for test fuels in Los Angeles in year 2010. First, quadrupling the estimated total organic gas (TOG) and tripling the CO emissions from light-duty gasoline vehicles in the air quality model increases the predicted peak ozone, as expected. The percent increase in peak ozone for the test fuels, about 25% of total ozone, is essentially the same as the percent increase in TOG emissions, about 25% of the total emissions from all sources. However, there is no important effect on the ranking of the test fuels from lowest to highest in predicted ozone formation. Second, replacing the original biogenic emission inventory with an alternative inventory having substantially lower biogenic emissions reduces the predicted peak ozone. The percent decrease in peak ozone, about 6% of total ozone, is considerably less than the percent decrease in TOG emissions, about 34% of total emissions. Fuel rankings are unchanged except for a reversal of two test gasolines in the ranking based on peak ozone. However, this reversal is not found in fuel rankings based on other measures of ozone formation. Third, replacing the Carbon Bond Mechanism version IV (CBM-IV) in the air quality model with an alternative representation of atmospheric chemistry, the Statewide Air Pollution Research Center (SAPRC) mechanism, increases the peak ozone by about 9%. There are also important changes in fuel rankings. For one research test gasoline, the contribution of light-duty gasoline vehicles to ozone is similar with both chemical mechanisms, but for another test gasoline that gives the lowest ozone with the CBM-IV, the contribution of light-duty gasoline vehicles to ozone is substantially higher with the SAPRC mechanism. With the CBM-IV mechanism, the most promising of the test gasolines studied has lower predicted ozone than any of the cases representing use of methanol fuels in prototype, flexible/variable fuel vehicles. With the SAPRC mechanism, the most promising test gasoline studied has lower predicted ozone than one methanol case and higher ozone than the other methanol case. These changes in fuel rankings are probably due to known differences in the reactivity of toluene and formaldehyde in the two mechanisms.

Development of a photochemical airshed model for Sydney, New South Wales, Australia

This paper describes the development of an airshed modelling system for the Environment Protection Authority of NSW (EPANSW). The airshed modelling system is one of the key outcomes of the Metropolitan Air Quality Study (MAQS). It will provide the tools necessary to aid in the development of an Air Quality Management Plan for photochemical smog within the Sydney, Lower Hunter, and Illawarra regions. The three dimensional airshed model was run for a 2 day generic Sydney oxidant event using historical data and some preliminary results from other MAQS projects. Predicted maximum smog concentrations obtained from the simulations appear to be comparable with the observations.

Modelling photochemical smog in the Perth region

A three-year study of the occurrence of photochemical smog in the vicinity of Perth, Western Australia is being conducted by the State Energy Commission (“SECWA”) and the W.A. Department of Environmental Protection (“DEP”). The project, termed the Perth Photochemical Smog Study (“PPSS”) includes extensive monitoring of smog components and meteorological variables, development of an inventory of emissions of smog precursors, and modelling of the combined effects of smog chemistry and meteorological transport. The Urban Airshed Model (“UAM”) has been used for initial photochemistry studies. Input data have been obtained from surface measurements and the author's own mesoscale meteorological model. While the Urban Airshed Model has proved to have limitations in its application to the Perth region, model results have highlighted a number of issues. These, and implications for further work, are described.

Ozone and aerosol productivity of reactive organics

A technique developed to determine the amount of ozone and secondary photochemical species generated by the individual organic components of a complex atmospheric organic/NO x mixture (Bowman and Seinfeld, J. geophys. Res. 99, 5309–5324 (1994a)) is used to study the individual contributions of carbonyl, aromatic, alkane, and alkene emissions to ozone and secondary organic and inorganic aerosol species in the South Coast Air Basin of California for the Southern California Air Quality Study (SCAQS) air pollution episode of 27–28 August 1987. Aldehydes exhibit the highest ozone productivity followed by alkenes, and lumped reactive aromatics. The aromatic species and formaldehyde enhance the production of secondary organic and secondary sulfate aerosol, because they are effectively OH sources. The same species, through their effect on both OH and O 3 production, are also significant precursors of HNO 3, and consequently, of nitrate aerosol. This methodology can be used in conjunction with urban airshed models to investigate alternative emission control scenarios.

Inverse air pollution modelling of urban-scale carbon monoxide emissions

A new recursive least-squares technique is developed to give spatial and temporal definition to the adjustments necessary in an emission inventory, to fit ambient concentration observations optimally. The CIT Photochemical Airshed Model is used to compute CO concentration distributions arising from 29 separate source domains in the South Coast Air Basin of California. A Kalman filter integrated within the model matches predictions with CO observations at 27 locations by superposing the computed distributions with optimal weighting factors. The filter structure allows control of the extent to which adjusted emission inventories are allowed to deviate from a base-case, which already has high spatial and temporal definition. Applied to the Southern California Air Quality Study, 27–29 August 1987, strong temporal dependence was noted in the necessary adjustment to the available CO emission inventory, with a peak factor of 3.0 at midday on weekdays. The spatial resolution of the technique revealed new high-emission zones for CO in a corridor between Pasadena and San Bernardino, in the Riverside-Corona area, and along the Pacific coast on Saturday. In this first such application to an urban environment, some success was also achieved in correcting the phasing of emissions for errors arising from the neglect of source-receptor lags in the inverse modelling technique.

An Observation-Based Model for Analyzing Ozone Precursor Relationships in the Urban Atmosphere

An Observation-Based Model (OBM) is described, which uses in-situ atmospheric observations to determine the sensitivity of ozone concentrations in an urban atmosphere to changes in the emissions of ozone precursors (i.e., volatile organic compounds and nitrogen oxides). The model is formulated following the concept of Relative Incremental Reactivity (RIR) developed by Carter and Atkinson. In the OBM, however, observed concentrations rather than emission inventories are used to drive the photochemical simulations and thereby ensure that the calculations are carried out for the proper mix of nitrogen oxides and volatile organic compounds. From these calculations, a series of sensitivity factors, or RIRs, are inferred that can be used to (1) determine whether reducing emissions of nitrogen oxide or emissions of hydrocarbons would be most effective in abating ozone in a given urban area, and (2) identify the most critical subset of hydrocarbons present in an urban atmosphere causing ozone exceedances. Because the OBM is relatively easy and inexpensive to operate and makes use of data that are increasingly available, it can be used to analyze a wide array of ozone episodes and, thus, could prove to be a relatively cost-effective tool for the analysis of ozone precursor relationships in an urban atmosphere. On the other hand, because the OBM is diagnostic rather than prognostic, it cannot be used in a predictive mode to estimate exactly how much emission reduction is needed to reduce ozone concentrations. For this reason, the OBM should be viewed as a complement to, rather than a substitute for, more sophisticated gridded, emission-based models. To illustrate the characteristics of the OBM and to demonstrate its applicability, we first compare the results of the OBM to those obtained from a series of simulations of the Atlanta metropolitan area using the Urban Airshed Model (UAM), a three-dimensional Eulerian grid model. The OBM is then used to analyze a dataset obtained from the 1990 Atlanta Ozone Study, an EPA field sampling program conducted during the summer of 1990. Because of limitations and potential flaws in the 1990 Atlanta dataset, the results of this OBM analysis are largely illustrative rather than definitive. Nevertheless, a few important issues are elucidated by the analysis. These include (1) the importance of accounting for biogenic hydrocarbons produced from urban vegetation; (2) the potential flaw in using early-morning VOC-to-NOx ratios to infer whether ozone production is limited by VOC or NOx; (3) the critical need for high-sensitivity nitrogen oxide measurements to quantify the sub-ppbv concentrations of NO during the afternoon hours; and (4) the need to consider a number of individual ozone episodes in studying an urban atmosphere because of the possibility that the degree of VOC- and NOx-limitation may vary from one episode to another.