Use of the Kit Fox field data to analyze dense gas dispersion modeling issues

The 1995 Kit Fox dense gas field data set consists of 52 trials where short-duration CO₂ gas releases were made at ground level over a rough surface during neutral to stable conditions. The experiments were intended to demonstrate the effects on dense gas clouds of relatively large roughnesses typical of industrial process plants. Fast response concentration observations were made by 80 samplers located on four downwind lines (25, 50, 100, and 225 m), including profile observations on three towers on each of the closest three arcs. Detailed meteorological measurements were made on several towers within and outside of the roughness arrays. The data analysis emphasized the variation of maximum concentration with surface roughness, the dependence of cloud advection speed on cloud depth, the variation of the three components of dispersion with ambient turbulence, and the dependence of vertical entrainment rate on ambient friction velocity and cloud Richardson number. The Kit Fox data were used to evaluate a specific dense gas dispersion model (HEGADAS 3+), with emphasis on whether it would be able to account for the increased roughness. The model was able to satisfactorily simulate the observed concentrations, with a mean bias of about 5% and with about 90% of the predictions within a factor of two of the observations.     

Application of circuit model for Taipei City PM10 simulation

This study simulated PM₁₀ concentrations in Taipei City from September 1990 to June 1992, by using a circuit model developed by Tsuang and Chao. Simulation results demonstrated that the scheme performs better than a box model. The correlation coefficient between observed and predicted values was 0.61 with a standard error of 60 μg m^-3. The mean treatment score of successful predictions of high PM₁₀ concentrations (exceeding 125 μg m^-3) was 58%. Fine particulates were found to be the major contributor toward hazardous air quality, under a stable, low wind speed and rainless atmosphere. In addition, the ill-mixed factor was found to be zero. The circuit model is close to a box model equation under an unstable atmosphere where the mixing height is much higher than the surface layer.     

An analytical model for crosswind integrated concentrations released from a continuous source in a finite atmospheric boundary layer

An analytical model for the crosswind integrated concentrations released from a continuous source in a finite atmospheric boundary layer is formulated by considering the wind speed as a power law profile of vertical height above the ground and eddy diffusivity as an explicit function of downwind distance from the source and vertical height. A closed form analytical solution of the resulting advection–diffusion equation for these profiles of wind speed and eddy diffusivity with the physically relevant boundary conditions is derived using the separation of variables technique that leads to a Sturm–Liouville eigen value problem. Various particular cases of the model are deduced.The model is evaluated with the observations obtained from Prairie Grass experiment in various stability classes varying from very unstable to neutral and stable conditions and Hanford diffusion experiment in stable conditions. The agreement is found to be good between the computed and observed concentrations in both the diffusion experiments. For Prairie Grass experiment, the model is predicting 78% cases with in a factor of two and gives a value of NMSE as 0.075. On the other hand, for Hanford observations in stable conditions, it predicts 70% cases with in factor of two. An extensive analysis of statistical measures with the downwind distances from the source reveals that the model is performing well close to the source.     

A simple urban dispersion model tested with tracer data from Oklahoma City and Manhattan

A simple urban dispersion model is tested that is based on the Gaussian plume model and modifications to the Briggs urban dispersion curves. An initial dispersion coefficient (σ_o) of 40 m is assumed to apply in built-up downtown areas, and the stability is assumed to be slightly unstable during the day and slightly stable during the night. Observations from tracer experiments during the Joint Urban 2003 (JU2003) field study in Oklahoma City and the Madison Square Garden 2005 (MSG05) field study in Manhattan are used for model testing. The tracer SF₆ was released during JU2003 near ground level in the downtown area and concentrations were observed at over 100 locations within 4 km from the source. Six perfluorocarbon tracer (PFT) gases were released near ground level during MSG05 and sampled by about 20 samplers at the surface and on building roofs. The evaluations compare concentrations normalized by source release rate, C/Q, for each sampler location and each tracer release, where data were used only if both the observed and predicted concentrations exceeded threshold levels. At JU2003, for all samplers and release times, the fractional mean bias (FB) is about 0.2 during the day (20% mean underprediction) and 0.0 during the night. About 45 –50% of the predictions are within a factor of two (FAC2) of the observations day and night at JU2003. The maximum observed C/Q is about two times the maximum predicted C/Q both day and night. At MSG05, for all PFTs, surface samplers, and release times, FB is 0.14 and FAC2 is about 45%. The overall 60 min-averaged maximum C/Q is underpredicted by about 40% for the surface samplers and is overpredicted by about 25% for the building-roof samplers.     

Analysis of air quality data near roadways using a dispersion model

We used a dispersion model to analyze measurements made during a field study conducted by the U.S. EPA in July–August 2006, to estimate the impact of traffic emissions on air quality at distances of tens of meters from an eight-lane highway located in Raleigh, NC. The air quality measurements consisted of long path optical measurements of NO at distances of 7 and 17 m from the edge of the highway. Sonic anemometers were used to measure wind speed and turbulent velocities at 6 and 20 m from the highway. Traffic flow rates were monitored using traffic surveillance cameras. The dispersion model [Venkatram, A., 2004. On estimating emissions through horizontal fluxes. Atmospheric Environment 38, 2439–2446] explained over 60% of the variance of the observed path averaged NO concentrations, and over 90% of the observed concentrations were within a factor of two of the model estimates.Sensitivity tests conducted with the model indicated that the traffic flow rate made the largest contribution to the variance of the observed NO concentrations. The meteorological variable that had the largest impact on the near road NO concentrations was the standard deviation of the vertical velocity fluctuations, σ_w. Wind speed had a relatively minor effect on concentrations. Furthermore, as long as the wind direction was within ±45° from the normal to the road, wind direction had little impact on near road concentrations. The measurements did not allow us to draw conclusions on the impact of traffic-induced turbulence on dispersion. The analysis of air quality and meteorological observations resulted in plausible estimates of on-road emission factors for NO.     

The influence of aerosol dynamics on indoor exposure to airborne DEHP

A mass-balance model was extended to investigate the influence of aerosol particles on the accumulation of indoor airborne DEHP, which allows the consideration of a variable particle concentration. The calculated gas-phase di-2-ethylhexyl phthalate (DEHP) concentration is consistent with those measured within residences in both the United States and Europe. Model predictions suggest that there are differences of more than 10% of particle-phase DEHP concentrations between the variable-particle-concentration case and the constant one for over half (578 days) within the calculation time of 1000 days. Airborne DEHP consists primarily of a particle phase. The exposure data indicate that the influence of particle dynamics remains significant throughout the calculation period, and the size fraction of 0–0.5 μm contributes the most, at 39.1%, to the total exposure to particle-phase DEHP as a result of a strong “source” effect which brings particles into the indoor air and a weak “sink” effect which removes particles from the indoor air. The sensitivity analysis indicates that deposition exhibits the most apparent influence, and particle emission from cooking is a significant factor, as cooking is the main source of particles in the size fraction of 0–0.5 μm. The sensitivity analysis also shows that particle penetration has a less obvious influence on the exposure to airborne DEHP because air exchange rate caused penetration introduces and removes particles simultaneously, thus having a limited influence on the airborne DEHP; while resuspension exhibits the weakest influence because it contributes little to the small particles which are the main component of aerosol particles indoors. Strategies for enhancing deposition and reducing particle emissions from cooking and penetration may be helpful to reduce residents’ exposure to airborne SVOCs.     

Validation of road vehicle and traffic emission models – A review and meta-analysis

Road transport is often the main source of air pollution in urban areas, and there is an increasing need to estimate its contribution precisely so that pollution-reduction measures (e.g. emission standards, scrapage programs, traffic management, ITS) are designed and implemented appropriately. This paper presents a meta-analysis of 50 studies dealing with the validation of various types of traffic emission model, including ‘average speed’, ‘traffic situation’, ‘traffic variable’, ‘cycle variable’, and ‘modal’ models. The validation studies employ measurements in tunnels, ambient concentration measurements, remote sensing, laboratory tests, and mass-balance techniques. One major finding of the analysis is that several models are only partially validated or not validated at all. The mean prediction errors are generally within a factor of 1.3 of the observed values for CO₂, within a factor of 2 for HC and NO_x, and within a factor of 3 for CO and PM, although differences as high as a factor of 5 have been reported. A positive mean prediction error for NO_x (i.e. overestimation) was established for all model types and practically all validation techniques. In the case of HC, model predictions have been moving from underestimation to overestimation since the 1980s. The large prediction error for PM may be associated with different PM definitions between models and observations (e.g. size, measurement principle, exhaust/non-exhaust contribution).Statistical analyses show that the mean prediction error is generally not significantly different (p < 0.05) when the data are categorised according to model type or validation technique. Thus, there is no conclusive evidence that demonstrates that more complex models systematically perform better in terms of prediction error than less complex models. In fact, less complex models appear to perform better for PM. Moreover, the choice of validation technique does not systematically affect the result, with the exception of a CO underprediction when the validation is based on ambient concentration measurements and inverse modelling. The analysis identified two vital elements currently lacking in traffic emissions modelling: 1) guidance on the allowable error margins for different applications/scales, and 2) estimates of prediction errors. It is recommended that current and future emission models incorporate the capability to quantify prediction errors, and that clear guidelines are developed internationally with respect to expected accuracy.     

A fluctuating plume dispersion model for the prediction of odour-impact frequencies from continuous stationary sources

A fluctuating plume dispersion model has been developed to facilitate the prediction of odour-impact frequencies in the communities surrounding elevated point sources. The model was used to predict the frequencies of occurrence of odours of various magnitudes for 1 h periods. In addition, the model predicted the maximum odour level. The model was tested with an extensive set of data collected in the residential areas surrounding the paint shop of an automotive assembly plant. Most of the perceived odours in the vicinity of the 64, 46 m high stacks ranged between 2 and 7 odour units and generally persisted for less than 30 s. Ninety-eight different field determinations of odour impact frequencies within 1 km of the plant were conducted during the course of the study. To simplify evaluation, the frequencies of occurrence of different odour levels were summed to give the total frequency of occurrence of all readily detectable (>2 OU) odours. The model provided excellent simulation of the total frequencies of occurrence where the odour was frequent (i.e. readily detectable more than 30% of the time). At lower frequencies of occurrence the model prediction was poor. The stability class did not seem to affect the model's ability to predict field frequency values. However, the model provided excellent predictions of the maximum odour levels without being sensitive to either stability class or distance from the source. Ninety-five percent of the predicted maximum values were within a factor of two of the measured field maximum values.     

The potential of statistical state space models in urban ozone forecasting

State space models for tropospheric urban ozone prediction are introduced and compared with linear regression models. The linear and non-linear state space models make accurate short-term predictions of the ozone dynamics. The average prediction error one hour in advance is 7 μg/m³ and increases logarithmically with time until it reaches 26 μg/m³ after 30 days. For a given sequence of solar radiation inputs, predictions converge exponentially with a time scale of 8 hours, so that the model is insensitive to perturbations of more than 150 μg/m³ O₃. The slow increase of the prediction error in addition to the uniqueness of the prediction are encouraging for applications of state space models in forecasting ozone levels when coupled with a model that predicts total radiation. Since a radiation prediction model will be more accurate during cloud-free conditions, in addition to the fact that the state space models perform better during the summer months, state space models are suitable for applications in sunny environments.     

Partitioning of nitrate and ammonium between the gas and particulate phases during the 1997 IMADA-AVER study in Mexico City

The partitioning of nitrate and ammonium between the gas and particulate phases is studied combining available equilibrium models and measurements taken in Mexico City during the 1997 IMADA-AVER field campaign. Based on this analysis, there are no significant differences in model predictions, but some discrepancies exist between predictions and observations. The inclusion of crustal elements in the modeling framework improves agreement of model predictions for particulate nitrate against measurements by approximately 5%. Although some equilibrium aerosol models do not explicitly treat crustal elements, these species can be treated as equivalent concentrations of sodium. Atmospheric equilibrium models predict daily average PM_2.5 nitrate concentrations within 20% of the IMADA-AVER measurements at the MER site. Six-hour average PM_2.5 nitrate concentrations are predicted within 30–50% on average except for the afternoon sampling periods (12:00–18:00 h). Investigating the possible sources of these discrepancies, it appears that a dynamic instead of an equilibrium approach is more suitable in reproducing aerosol behavior during these afternoon periods. By applying the Multicomponent Aerosol Dynamic Model (MADM), model performance in predicting concentrations of particulate nitrate significantly improves during the afternoon periods.     

Turbulent Diffusion Behind Vehicles: Effect of Traffic Speed on Pollutant Concentrations

Recent theoretical and experimental investigations Indicate that turbulent diffusion behind moving vehicles Is Influenced by the speed of the vehicle. Vertical wake induced turbulent diffusion, explicitly treated in the numerical ROADWAY model, is proportional to the square of the wind speed relative to the moving vehicle. Hence, the model predictions of turbulent mixing and pollutant concentrations on and downwind of a roadway are dependent upon the traffic speed. It Is expected from theoretical considerations that the effect of vehicle speed on pollutant concentrations will be more significant during stable atmospheric conditions, because in neutral and unstable conditions the vehicle-wake turbulence is quickly masked by the ambient turbulence. In this study, experimental data are utilized to evaluate the theoretical predictions of the effects of traffic speed on the ambient pollutant concentrations. The effects of vehicle speed upon ambient concentrations are investigated through wind tunnel experiments and field studies that used dual tracers. Consistent with predictions of the ROADWAY model, data obtained near the Long Island Expressway indicate that the influence of traffic speed on the ambient pollutant concentrations Is not significant during unstable and neutral conditions. The Long Island experiment did not provide sufficient field data to assess the model predictions of the traffic speed effect during stable atmospheric conditions.     

Sensitivity of a molecular marker based positive matrix factorization model to the number of receptor observations

To investigate the impact of the number of observations on molecular marker-based positive matrix factorization (MM-PMF) source apportionment models, daily PM_2.5 samples were collected in East St. Louis, IL, from April 2002 through May 2003. The samples were analyzed for daily 24-h average concentrations of elemental and organic carbon, trace elements, and speciated particle-phase organic compounds. A total of 273 sets of observations were used in the model and consisted of all valid sets of observations from the year long data set minus one sixth of the measurements, which were collected every 6th day and were analyzed by different chemical analysis techniques. In addition to the base case of 273 samples, systematic subsets of the data set were analyzed by PMF. These subsets of data included 50% of the observations (135–138 days), 33% of the observations (90–92 days) and 20% of the observations (52–56 days). In addition, model runs were also examined that used 48-h, 72-h, 6-day, and weekly average concentrations as model inputs. All MM-PMF model runs were processed following the same procedures to explore the stability of the source attribution results. Consistent with previous MM-PMF results for East St. Louis, the main sources of organic aerosol were found to be mobile sources, secondary organic aerosols (SOAs), resuspended soil and biomass combustions, as well as an n-alkane dominated point source and other combustion sources. The MM-PMF model was reasonably stable when the number of observations in the input was reduced to ninety, or approximately 33% of observations present in the base case. In these cases, the key factors including resuspended soil, mobile and secondary factors, which accounted for more than 70% of the measured OC concentrations, were stable as defined by a relative standard deviation (RSD) of less than 30%. Similar results were obtained from the smaller data subsets, but resulted in larger uncertainties, with several of these factors yielding RSD of greater than 30%. The three factors with the largest OC contributions were more stable than the other minor factors, even when the number of observations was nominally 50 days. Secondary organic aerosol (SOA) was the most stable factor observed in the model runs. Since it is unclear if these results can be broadly applied to all MM-PMF models, additional studies of this nature are needed to assess the broader applicability of these conclusions. Until such studies are implemented, this paper provides a foundation to design future studies in sampling strategies for source apportionment using MM-PMF.     

Modelling concentrations of decamethylcyclopentasiloxane in two UK rivers using LF2000-WQX

Current regulatory environmental exposure assessments for decamethylcyclopentasiloxane (D₅), used in a range of personal care products, are based on a number of erroneous assumptions. Using an estimated D₅ flux to waste water of 11.6 mg cap⁻¹ d⁻¹, a 95.2% removal rate in Sewage Treatment Plants (STP) and a dilution factor of 10 results in modelled surface water concentrations that are up to an order of magnitude higher than concentrations observed downstream of STPs in two UK rivers. A GIS-based water quality model (LF2000-WQX) was used to predict concentrations of D₅ in two UK rivers. Assuming the STP removal rate is reasonable, a waste water flux of 2.4 mg cap⁻¹ d⁻¹ is needed in order to obtain a reasonable match between predicted and observed in-river concentrations. This flux is consistent with measured effluent concentrations. The results highlight major uncertainties in estimating chemical emission rates for volatile chemicals used in personal care products and suggest that measured concentrations in waste water are needed to refine exposure assessments.     

A Study of Wood Stove Particulate Emissions

Particulate emission factors for two wood stove models have been determined for two types of fuel and a range of operating conditions. The emission factors range from 1 g/kg (fuel) to 24 g/kg. A model is presented which represents the emission factor as a simple function of the ratio of fuel load to combustion rate, or the length of time between refueling. This model is felt to be appropriate for evaluating the impact of wood-based residential space heating on ambient air concentrations of particulate matter If certain assumptions can be made about stove operating conditions. An application of the emission factor model to a typical community suggests that the contribution of wood stoves to ambient particulate levels might reach 100 μg/m³ if the entire heating load were carried by wood.

Preliminary analyses of the particulate matter Indicate that benzene extractables range from 42% of the total particulate mass at short refuel times to 67% at longer refuel times. About 45% of the mass of benzene extractables appeared in the neutral fraction of acid base extractions. Polycyclic aromatic hydrocarbons are expected to be included in this neutral fraction.     

Contribution of transboundary air pollution to ionic concentrations in fog in the Kinki Region of Japan

To estimate the contribution of transboundary transported air pollutants from other Asian countries to Japan in ionic concentrations in fog water in March 2005, the Community Multiscale Air Quality (CMAQ) modeling system was utilized with meteorological fields produced by the 5th generation Mesoscale Model (MM5). For meteorological predictions, the model well reproduced the surface meteorological variables, particularly temperature and humidity, and generally captured fog occurrence. For chemical predictions, most of the model-predicted monthly mean concentrations were approximately within a factor of 2 of the observations, indicating that the model well simulated the long-range atmospheric transport from the Asian Continent to Japan. For SO₄^2?, NO₃^? and NH₄⁺, the contribution rates of the transboundary air pollution in the Kinki Region of Japan ranged from 69 to 82% for aerosols, from 47 to 87% for ionic concentrations in rain, and from 55 to 79% for ionic concentrations in fog. The study found that the transboundary air pollution also affected ionic concentrations in fog as well as aerosol concentrations and ionic concentrations in rain.     

Modelling atmospheric mercury transport and deposition across Europe and the UK

There are inadequate measurements of surface ambient concentrations of mercury species and their deposition rates for the UK deposition budget to be characterized. In order to estimate the overall mercury flux budget for the UK, a simple long-term 1D Lagrangian trajectory model was constructed that treats emissions (1998), atmospheric transformation and deposition across Europe. The model was used to simulate surface concentrations of mercury and deposition across Europe at a resolution of 50 km×50 km and across the UK at 20 km×20 km. The model appeared to perform adequately when compared with the few available measurements, reproducing mean concentrations of elemental gaseous mercury at particular locations and the magnitude of regional gradients. The model showed that 68% of the UK's mercury emissions are exported and 32% deposited within the UK. Of deposition to the UK, 25% originates from the Northern Hemisphere/global background, 41% from UK sources and 33% from other European countries. The total mercury deposition to the UK is in good agreement with other modelling, 9.9 tonne yr⁻¹ cf. 9.0 tonne yr⁻¹, for 1998. However, the attribution differs greatly from the results of other coarser-scale modelling, which allocates 55% of the deposition to the UK from UK sources, 4% from other European countries and 60% from the global background atmosphere. The model was found to be sensitive to the speciation of emissions and the dry deposition velocity of elemental gaseous mercury. The uncertainties and deficiencies are discussed in terms of model parameterization and input data, and measurement data with which models can be validated. There is an urgent requirement for measurements of removal terms, concentrations, and deposition with which models can be parameterized and validated.     

Testing the capability of the chemistry transport model LOTOS-EUROS to forecast PM10 levels in the Netherlands

Since particulate matter has a direct and adverse impact on public health, a good air quality forecast is important. Several European countries presently use statistical forecasting models, which have their limitations, especially for PM10. An alternative approach is to use a chemistry transport model. Here, the ability of the chemical transport model LOTOS-EUROS to forecast PM10 concentrations in the Netherlands was investigated. LOTOS-EUROS models several PM10 components individually. For sulphate, nitrate and ammonium aerosol the evaluation against observations shows that the modelled annual mean concentrations are within 20% of the measured concentration and that the temporal correlation is reasonably good (R > 0.6). For sea salt the model tended to overestimate the measured concentrations. For elemental carbon the correspondence with black smoke observations was reasonable. However, total PM10 is seriously underestimated, due to unmodelled components (secondary organic aerosols, mineral dust) and missing sources. Therefore, a simple bias correction for four seasons was derived based on the years 2004–2006. The model was compared with the Dutch operational statistical model PROPART and ground-level observations. With bias correction, LOTOS-EUROS performed better than PROPART regarding the timing of events. The major flaw of LOTOS-EUROS was that high values (>50 μg m^?3) were still underestimated. Another advantage of LOTOS-EUROS over the statistical model was the more detailed information in space and time, which facilitates communication of the forecast to the general public.     

Evaluating the size distribution characteristics and sources of atmospheric trace elements at two mountain sites: comparison of the clean and polluted regions in China

Size-resolved trace metal concentrations at two background sites were assessed during a 1-year observation campaign, with the measurements performed in parallel at two mountain sites, where Mt. Dinghu (DHS) located in the rural region of Pearl River Delta (PRD) and Mt. Gongga (GGS) located in the Tibetan Plateau region. In total, 15 selected trace elements (Mg, Al, K, V, Mn, Fe, Cu, Zn, As, Mo, Ag, Cd, Ba, Tl, and Pb) in aerosol samples were determined using inductively coupled plasma mass spectrometry (ICPMS). The major metals in these two mountain sites were Fe, K, Mg, and Ca with concentrations ranging between 241 and 1452 ng/m³, 428 and 1351 ng/m³, 334 and 875 ng/m³, and 376 and 870 ng/m³, respectively, while the trace metals with the lowest concentrations were Mo, Ag, Cd, and Tl with concentrations lower than 4 ng/m³ in DHS and 2 ng/m³ in GGS. The pronounced seasonal variability in the trace elements was observed in DHS, with lower concentrations in spring and summer and relatively high in winter and autumn, whereas seasonal variance of trace elements is hardly observed in Mt. Gongga. The size distribution pattern of crustal elements of Al, Mg, K, Ba, and Fe was quite similar in DHS and GGS, which were mainly found in coarse particles peaked at 4.7–5.8 μm. In addition, V, Mo, Ag, and Tl were also concentrated in coarse particles, although the high enrichment factor (EF?>?100) of which suggested anthropogenic origin, whereas trace metals of Cd, Mn, Zn, As, Cu, and Pb concentrated in fine mode particles. Specifically, these trace metals peak at approximately 1.5 μm in DHS, while those in GGS peaked at diameter smaller than 0.3 μm, indicating the responsible for long-range transport from the far urban and industrialized areas. Multivariate receptor model combined with the enrichment factor results demonstrated that the trace elemental components at these two background sites were largely contributed from the fossil fuel combustion (55.4% in DHS and 44.0% in GGS) and industrial emissions factors (20.1% vs. 26.5%), which are associated with long distance transport from the coastal area of Southeast China and the Northwestern India, respectively, as suggested by the backward air mass trajectory analysis. Local sources from soil dust contributed a minor variance for trace elements in DHS (9.7%) and GGS (13.8%), respectively.

     

Comparison of the performances of land use regression modelling and dispersion modelling in estimating small-scale variations in long-term air pollution concentrations in a Dutch urban area

The performance of a Land Use Regression (LUR) model and a dispersion model (URBIS – URBis Information System) was compared in a Dutch urban area. For the Rijnmond area, i.e. Rotterdam and surroundings, nitrogen dioxide (NO₂) concentrations for 2001 were estimated for nearly 70 000 centroids of a regular grid of 100 × 100 m.A LUR model based upon measurements carried out on 44 sites from the Dutch national monitoring network and upon Geographic Information System (GIS) predictor variables including traffic intensity, industry, population and residential land use was developed. Interpolation of regional background concentration measurements was used to obtain the regional background. The URBIS system was used to estimate NO₂ concentrations using dispersion modelling. URBIS includes the CAR model (Calculation of Air pollution from Road traffic) to calculate concentrations of air pollutants near urban roads and Gaussian plume models to calculate air pollution levels near motorways and industrial sources. Background concentrations were accounted for using 1 × 1 km maps derived from monitoring and model calculations.Moderate agreement was found between the URBIS and LUR in calculating NO₂ concentrations (R = 0.55). The predictions agreed well for the central part of the concentration distribution but differed substantially for the highest and lowest concentrations. The URBIS dispersion model performed better than the LUR model (R = 0.77 versus R = 0.47 respectively) in the comparison between measured and calculated concentrations on 18 validation sites. Differences can be understood because of the use of different regional background concentrations, inclusion of rather coarse land use category industry as a predictor variable in the LUR model and different treatment of conversion of NO to NO₂.Moderate agreement was found between a dispersion model and a land use regression model in calculating annual average NO₂ concentrations in an area with multiple sources. The dispersion model explained concentrations at validation sites better.     

A performance evaluation of the National Air Quality Forecast Capability for the summer of 2007

This paper provides a performance evaluation of the real-time, CONUS-scale National Air Quality Forecast Capability (NAQFC) that supported, in part, its transition into operational status. This evaluation focuses primarily on discrete forecasts for the maximum 8-h O₃ concentrations covering the 4-month period, June through September, 2007, using measurements obtained from EPA's AIRNow network. Results indicate that the 2007 NAQFC performed as well or better than previous configurations, despite the expansion of the forecast domain into the western half of the nation that is dominated by complex terrain. The mean, domain-wide, season-long correlation was 0.70. When examined over time, the domain-wide correlations exhibit a fairly consistent nature, with values exceeding 0.60 (0.70) over 90% (55%) of the days. The NAQFC systematically over-predicted the 8-h O₃ concentrations, continuing a trend established by earlier NAQFC configurations, though to a lesser degree. The summer-long mean forecast value of 53.2 ppb was 4.2 ppb higher than the observed value, resulting in a domain-wide Normalized Mean Bias (NMB) of 8.7%. Most of the over-prediction is associated with observed concentrations less than 50 ppb. In fact the model tends to under-predict when concentrations exceed 70 ppb. As with the bias, the error associated with the latest configuration was also lower. The summer-long Root Mean Square Error of 13.0 ppb (Normalized Mean Error (NME) = 20.4%) represented marked improvements over earlier forecasts. Examination of the spatial distribution of both the NMB and NME reveals that the NAQFC was generally within 25% for the NME and 25% for the NMB over a majority of the domain. Several areas of poorer performance, where the NMB and NME often exceed 25% and in some cases 50%, were noted. These areas include southern California, where the NAQFC tended to under-predict concentrations (especially on weekends) and the southeast Atlantic and Gulf coasts regions, where the model over-predicted. Subsequent analysis revealed that the incorrect temporal allocation of precursor emissions was likely the source of the under-prediction in southern California, while inaccurate simulation of PBL heights likely contributed to the over-prediction in the coastal regions.