Ozone Prediction Research Articles

Effect of chemistry‐aerosol‐climate coupling on predictions of future climate and future levels of tropospheric ozone and aerosols

We explore the extent to which chemistry‐aerosol‐climate coupling influences predictions of future ozone and aerosols as well as future climate using the Goddard Institute for Space Studies (GISS) general circulation model II' with on‐line simulation of tropospheric ozone‐NOx‐hydrocarbon chemistry and sulfate, nitrate, ammonium, black carbon, primary organic carbon, and secondary organic carbon aerosols. Based on IPCC scenario A2, year 2100 ozone, aerosols, and climate simulated with full chemistry‐aerosol‐climate coupling are compared with those simulated from a stepwise approach. In the stepwise method year 2100 ozone and aerosols are first simulated using present‐day climate and year 2100 emissions (denoted as simulation CHEM2100sw) and year 2100 climate is then predicted using offline monthly fields of O3 and aerosols from CHEM2100sw (denoted as simulation CLIM2100sw). The fully coupled chemistry‐aerosol‐climate simulation predicts a 15% lower global burden of O3 for year 2100 than the simulation CHEM2100sw which does not account for future changes in climate. Relative to CHEM2100sw, year 2100 column burdens of all aerosols in the fully coupled simulation exhibit reductions of 10–20 mg m−2 in DJF and up to 10 mg m−2 in JJA in mid to high latitudes in the Northern Hemisphere, reductions of up to 20 mg m−2 over the eastern United States, northeastern China, and Europe in DJF, and increases of 30–50 mg m−2 over populated and biomass burning areas in JJA. As a result, relative to year 2100 climate simulated from CLIM2100sw, full chemistry‐aerosol‐climate coupling leads to a stronger net global warming by greenhouse gases, tropospheric ozone and aerosols in year 2100, with a global and annual mean surface air temperature higher by 0.42 K. For simulation of year 2100 aerosols, we conclude that it is important to consider the positive feedback between future aerosol direct radiative forcing and future aerosol concentrations; increased aerosol concentrations lead to reductions in convection and precipitation (or wet deposition of aerosols), further increasing lower tropospheric aerosol concentrations.

Tropical tropopause layer

Observations of temperature, winds, and atmospheric trace gases suggest that the transition from troposphere to stratosphere occurs in a layer, rather than at a sharp “tropopause.” In the tropics, this layer is often called the “tropical tropopause layer” (TTL). We present an overview of observations in the TTL and discuss the radiative, dynamical, and chemical processes that lead to its time‐varying, three‐dimensional structure. We present a synthesis definition with a bottom at 150 hPa, 355 K, 14 km (pressure, potential temperature, and altitude) and a top at 70 hPa, 425 K, 18.5 km. Laterally, the TTL is bounded by the position of the subtropical jets. We highlight recent progress in understanding of the TTL but emphasize that a number of processes, notably deep, possibly overshooting convection, remain not well understood. The TTL acts in many ways as a “gate” to the stratosphere, and understanding all relevant processes is of great importance for reliable predictions of future stratospheric ozone and climate.

REGIONAL-SCALE MODELING OZONE AIR QUALITY OVER THE CONTINENTAL SOUTH EAST ASIA

Long range transport of ozone and its precursors can significantly impact the air quality in downwind regions. The problem of regional transport of ozone has been studied for more than three decades in Europe and U.S but not yet in Southeast Asia. This study investigated the regional scale distribution of tropospheric ozone over the Continental South East Asia Region (CSEA) of Thailand, Burma, Cambodia, Lao and Vietnam. The Models-3 Community Multi-scale Air Quality (CMAQ modeling system, driven by the NCAR/Penn State Fifth-Generation Mesoscale Model (MM5), is used for the purpose. The model domain covers the longitude range from 91'E to 111°E and the latitude range from 5°N to 25°N. Two most recent ozone episodes of March 24-26, 2004 and January 2-4, 2005 were selected which represent the typical meteorological conditions for high ozone concentrations periods of a year. The episode analysis was made based on available data from 10 and 4 monitoring stations located in Bangkok of Thailand and Ho Chi Minh City (HCMC) of Vietnam, respectively. The episodes were characterized with hourly ozone levels above the National Ambient Air Quality Standards of Thailand and Vietnam of 100 ppb at a number of the monitoring stations. The maximum ground level concentrations of ozone for March 2004 and January 2005 episodes reached 173 ppb and 157 ppb, respectively, in the urban plume of the Bangkok Metropolitan Region (BMR). The simulations were performed with 0.5o 0.5° emission input data which was prepared from the regional anthropogenic emission inventory used in the Transport and Chemical Evolution over the Pacific (TRACE-P), and the biogenic emissions obtained from the Global Emissions Inventory Activity (GEIA). The simulated overall picture of ground level ozone concentrations over CSEA domain shows that the concentrations were high at the downwind areas at a considerable distance from large urban areas such as BMR and HCMC. During March 2004 episode the ozone plume moved northeastward following the Southwesterly monsoon and the maximum width of the modeled plume with the ozone above 100 ppb was about 70 km from BMR. For HCMC the ozone plume moved northward and the concentration in the city plume was lower with the width of isopleth of 50ppb of around 40 km. During the Jan 2005 episode the ozone plume moved southwestward following the Northeasterly monsoon and the width of the modeled plume with the ozone concentration above 100 ppb in BMR was 50 km while for HCMC the width of the 40ppb isopleth was about 30 km. The model performance was evaluated on the available observed hourly ozone concentrations. The model system was shown to be able to reproduce the peak ozone levels that occurred during the episodes at these two large urban areas, and capture the day by day variations during the selected episodes. The performance statistics MNBE, NGE, and UPA for the simulated ozone concentrations are within U.S. EPA guidance criteria and are comparable to those reported previous for other regional ozone simulations. It is shown that the MM5/CMAQ system is the suitable modeling tools for ozone prediction over the CSEA.

Bias adjustment techniques for improving ozone air quality forecasts

In this study, we apply two bias adjustment techniques to help improve forecast accuracy by postprocessing air quality forecast model outputs. These techniques are applied to modeled ozone (O3) forecasts over the continental United States during the summer of 2005. The first technique, referred to as the hybrid forecast (HF), combines the most recent observed ozone with model‐predicted ozone tendency for the subsequent forecast time period. The second technique applied here is the Kalman filter (KF), which is a recursive, linear, and adaptive method that takes into account the temporal variation of forecast errors at a specific location. Two modifications to the Kalman filter are investigated. A key parameter in the KF approach is the error ratio, which determines the relative weighting of observed and forecast ozone values. This parameter is optimized to improve the prediction of ozone based on time series data at individual monitoring sites in the Aerometric Information Retrieval Now network. The optimal error ratios inherent in the KF algorithm implementation are found to vary across space; however, comparisons of the resultant KF‐adjusted forecasts using a single fixed value of this parameter with those using the optimal values determined for each individual site reveal similar results, suggesting that the uncertainty in the estimation of this parameter does not have a significant impact on the final bias‐adjusted predictions. The KF postprocessing is also combined with the Kolmogorov‐Zurbenko filter, which extracts the intraday variability in the observed ozone time series; the results indicate a significant improvement in the ozone forecasts at locations in the Pacific coast region, but not as much at locations across the rest of the continental U.S. domain. Both HF and KF bias adjustment techniques help significantly reduce the systematic errors in ozone forecasts. For most of the global performance metrics examined, the KF approach performed better than the HF method. For the given model applications, both methods are effective in reducing biases at low ozone mixing ratio levels, but not as well at the high mixing ratio levels. This is due in part to the fact that high ambient ozone levels occur much less frequently than low to moderate levels for which the current model exhibits a systematic high bias. Additionally, the 12 km model grid structure is often unable to adequately capture the magnitude of peak O3 levels. Thus, these extreme events at discrete monitor locations are more difficult to predict.

The impact of chemical lateral boundary conditions on CMAQ predictions of tropospheric ozone over the continental United States

A sensitivity study is performed to examine the impact of lateral boundary conditions (LBCs) on the NOAA-EPA operational Air Quality Forecast Guidance over continental USA. We examined six LBCS: the fixed profile LBC, three global LBCs, and two ozonesonde LBCs for summer 2006. The simulated results from these six runs are compared to IONS ozonesonde and surface ozone measurements from August 1 to 5, 2006. The choice of LBCs can affect the ozone prediction throughout the domain, and mainly influence the predictions in upper altitude or near inflow boundaries, such as the US west coast and the northern border. Statistical results shows that the use of global model predictions for LBCs could improve the correlation coefficients of surface ozone prediction over the US west coast, but could also increase the ozone mean bias in most regions of the domain depending on global models. In this study, the use of the MOZART (Model for Ozone And Related chemical Tracers) prediction for CMAQ (Community Multiscale Air Quality) LBC shows a better surface ozone prediction than that with fixed LBC, especially over the US west coast. The LBCs derived from ozonesonde measurements yielded better O3 correlations in the upper troposphere.

Contour diagram fuzzy model for maximum surface ozone prediction

A contour diagram approach is presented for the identification of surface ozone concentration feature based on a set of rules by considering the meteorological variables such as the solar radiation, wind speed, temperature, humidity and rainfall. A fuzzy rule system approach is used because of the imprecise, insufficient, ambiguous and uncertain data available. The contour diagrams help to identify qualitative ozone concentration variability rules which are more general than conventional statistical or time series analysis. In the methodology, ozone concentration contours are based on a fixed variable as ozone precursor, namely, NO x and as the third variable one of the meteorological factors. Such contour diagrams for ozone concentration variation are prepared for six months. It is possible to identify the maximum ozone concentration episodes from these diagrams and then to set up the valid rules in the form of IF-THEN logical statements. These rules are obtained from available daily ozone, NO x and meteorological data as a first approximate reasoning step. In this manner, without mathematical formulations, expert maximum ozone concentration systems are identified. The application of the contour diagram approach is performed for daily ozone concentration measurements on European side of Istanbul city. It is concluded that through approximate reasoning with fuzzy rules, the maximum ozone concentration episodes can be identified and predicted without any mathematical expression.

The Impact of Nudging in the Meteorological Model for Retrospective Air Quality Simulations. Part II: Evaluating Collocated Meteorological and Air Quality Observations

Abstract For air quality modeling, it is important that the meteorological fields that are derived from meteorological models reflect the best characterization of the atmosphere. It is well known that the accuracy and overall representation of the modeled meteorological fields can be improved for retrospective simulations by creating dynamic analyses in which Newtonian relaxation, or “nudging,” is used throughout the simulation period. This article, the second of two parts, provides additional insight into the value of using nudging-based data assimilation for dynamic analysis in the meteorological fields for air quality modeling. Meteorological simulations are generated by the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) using both the traditional dynamic analysis approach and forecasts for a summertime period. The resultant meteorological fields are then used for emissions processing and air quality simulations using the Community Multiscale Air Quality Modeling System (CMAQ). The predictions of surface and near-surface meteorological fields and ozone are compared with a small network of collocated meteorological and air quality observations. Comparisons of 2-m temperature, 10-m wind speed, and surface shortwave radiation show a significant degradation over time when nudging is not used, whereas the dynamic analyses maintain consistent statistical scores over time for those fields. Using nudging in MM5 to generate dynamic analyses, on average, leads to a CMAQ simulation of hourly ozone with smaller error. Domainwide error patterns in specific meteorological fields do not directly or systematically translate into error patterns in ozone prediction at these sites, regardless of whether nudging is used in MM5, but large broad-scale errors in shortwave radiation prediction by MM5 directly affect ozone prediction by CMAQ at specific sites.

Hourly ozone prediction for a 24-h horizon using neural networks

This study is an attempt to verify the presence of non-linear dynamics in the ozone time series by testing a “dynamic” model, evaluated versus a “static” one, in the context of predicting hourly ozone concentrations, one-day ahead. The “dynamic” model uses a recursive structure involving a cascade of 24 multilayer perceptrons (MLP) arranged so that each MLP feeds the next one. The “static” model is a classical single MLP with 24 outputs. For both models, the inputs consist of ozone and of exogenous variables: past 24-h values of meteorological parameters and of NO 2; concerning the ozone inputs, the “static” model uses only the 24 past measurements, while the “dynamic” one uses, also, the previously forecast ozone concentrations, as soon as they are predicted by the model. The outputs are, for both configurations, ozone concentrations for a 24-h horizon. The performance of the two models was evaluated for an urban and a rural site, in the greater Paris. Globally, the results indicate a rather good applicability of these models for a short-term prediction of ozone. We notice that the results of the recursive model were comparable with those obtained via the “static” one; thus, we can conclude that there is no evidence of non-linear dynamics in the ozone time series under study.

Evaluation of ozone total column measurements by the Ozone Monitoring Instrument using a data assimilation system

On 15 July 2004, the Ozone Monitoring Instrument (OMI) on board the EOS Aura mission was launched. One of OMI's priorities is to continue the record of high spatial resolution ozone total column measurements provided by the various Total Ozone Mapping Spectrometer (TOMS) instruments since 1978. To this end, it is essential to estimate the errors affecting OMI ozone total column measurements and to see whether the actual accuracy is consistent with estimated values before launch. In this paper, data assimilation techniques are used to create a large comparison data set composed of ozone analyses resulting from assimilation of standard meteorological observations and ozone retrievals (independent of OMI measurements) into a numerical weather prediction model. This data set provides excellent global coverage and temporal resolution, not limited by the spatial and temporal distribution of other satellite or ground based information. The accuracy of the analyses is evaluated against ozone total column retrievals from Brewer measurements, while the assimilated ozone data set is compared to ozone predictions made using the ECMWF model, to check for the presence of bias. The OMI ozone column measurements considered here are obtained with the TOMS‐V8 total ozone algorithm and denoted as OMTO3 columns. They are compared with simulated OMI ozone columns, i.e., the quantities that the TOMS‐V8 algorithm would retrieve in the case when the atmospheric ozone profile at a specific location and time is equal to the one prescribed by the analysis. In this way, the comparison is statistically robust even when data acquired during a relatively short temporal interval or over a relatively small geographical area only is considered. A discussion of relevant error sources (including systematic components), vertical resolution, and contributions from prior information is provided. Special attention is given to determining the importance of representativeness errors. Our results show a solar zenith angle (SZA) dependence of the bias between measured and simulated OMI columns. This is believed to be due to moderate nonlinearity of the observation forward model and its effects on our definition of simulated OMI columns at high SZA. In view of these findings the final results of the intercomparison methodology used in this paper are obtained from OMI ozone columns retrieved using the basic implementation of the TOMS‐V8 algorithm applied to measurements taken at SZA not exceeding 70°. Intercomparison results between measured and simulated OMI ozone columns at SZA less than 70° show a relative bias of −3.2 ± 3.1% and a root‐mean‐square error of 4.5 ± 1.5%. The resulting bias is consistent with available estimates of the bias of OMTO3 columns with respect to SBUV/2 between 60°S and 60°N, as well as with respect to global Dobson data and Brewer measurements between 30°N and 60°N.

Ground-level ozone prediction by support vector machine approach with a cost-sensitive classification scheme

For ground-level ozone (O 3) prediction, a predictive model, with reliable performance not only on non-polluted days but, more importantly, on polluted days, is favored by public authorities to issue alerts, so that concerned citizens and industrial organizations could take precautions to avoid exposure and reduce harmful emissions. However, the class imbalance problem, i.e., in some collected field data, number of O 3 polluted days are much smaller than that of non-polluted days, will deteriorate the model performance on minority class—O 3 polluted days. Despite support vector machine (SVM) obtaining promising results in air quality prediction, in this study, a cost-sensitive classification scheme is proposed for the standard support vector classification model (S-SVC) in order to investigate whether the class imbalance plagues S-SVC. The S-SVC with such scheme is named as CS-SVC. Experiments on imbalanced data sets collected from two air quality monitoring sites in Hong Kong show that 1) S-SVC is still sensitive to class imbalance problem; 2) compared with S-SVC, CS-SVC effectively avoids class imbalance problem with lower percentage of false negative on O 3 polluted days but with higher percentage of false positive on non-polluted days; 3) compared with both S-SVC and CS-SVC, support vector regression model (SVR), after converting its output to binary one, only has similar performance with S-SVC, which indicates class imbalance problem also impairs the regressor model. From point of protecting public health, CS-SVC, which less likely misses to forecast O 3 polluted days, is recommended here.

Development and comparative analysis of tropospheric ozone prediction models using linear and artificial intelligence-based models in Mexicali, Baja California (Mexico) and Calexico, California (US)

This study developed 12 prediction models using two types of data matrix (daily means and a selection of the mean for the first 6 h of the day). The Persistence parametric prediction technique was applied separately to these matrices, as well as semiparametric Ridge Regression and three non-parametric or artificial intelligence techniques: Support Vector Machine, Multilayer Perceptron and ELMAN networks. The target was the prediction of maximum tropospheric ozone concentrations for the next day in the Mexicali–Calexico border area. The main ozone precursors and meteorological parameters were used for the different models. The proposals were evaluated using specific performance measurements for the air quality models established in the Model Validation Kit and recommended by the US Environmental Protection Agency. Results with similar margins of error were obtained in various models developed in this study, and some of them have provided smaller margins of error than similar prediction models existing in the literature developed in other regions. For this reason, we consider it feasible to apply the prediction models developed and they could be useful for supporting decisions in the matter of ozone pollution in the region under study, as well as for use in daily forecasting in this area.

From diagnosis to prognosis for forecasting air pollution using neural networks: Air pollution monitoring in Bilbao

This work focuses on the prediction of hourly levels up to 8 h ahead for five pollutants (SO 2, CO, NO 2, NO and O 3) and six locations in the area of Bilbao (Spain). To that end, 216 models based on neural networks (NNs) were built. The database used to fit the NNs were historical records of the traffic, meteorological and air pollution networks existing in the area corresponding to year 2000. Then, the models were tested on data from the same networks but corresponding to year 2001. At a first stage, for each of the 216 cases, 100 models based on different types of neural networks were built using data corresponding to year 2000. The final identification of the best model was made under the criteria of simultaneously having at a 95% confidence level the best values of R 2, d 1, FA2 and RMSE when applied to data of year 2001. The number of hourly cases in which due to gaps in data predictions were possible range from 11% to 38% depending on the sensor. Depending on the pollutant, location and number of hours ahead the prediction is made, different types of models were selected. The use of these models based on NNs can provide Bilbao's air pollution network originally designed for diagnosis purposes, with short-term, real time forecasting capabilities. The performance of these models at the different sensors in the area range from a maximum value of R 2 = 0.88 for the prediction of NO 2 1 h ahead, to a minimum value of R 2 = 0.15 for the prediction of ozone 8 h ahead. These boundaries and the limitation in the number of cases that predictions are possible represent the maximum forecasting capability that Bilbao's network can provide in real-life operating conditions.

Combining principal component regression and artificial neural networks for more accurate predictions of ground-level ozone

This work encompasses ozone modeling in the lower atmosphere. Data on seven environmental pollutant concentrations (CH 4, NMHC, CO, CO 2, NO, NO 2, and SO 2) and five meteorological variables (wind speed, wind direction, air temperature, relative humidity, and solar radiation) were used to develop models to predict the concentration of ozone in Kuwait's lower atmosphere. The models were developed by using summer air quality and meteorological data from a typical urban site when ozone concentration levels were the highest. The site was selected to represent a typical residential area with high traffic influences. The combined method, which is based on using both multiple regression combined with principal component analysis (PCR) and artificial neural network (ANN) modeling, was used to predict ozone concentration levels in the lower atmosphere. This combined approach was used to improve the prediction accuracy of ozone. The predictions of the models were found to be consistent with observed values. The R 2 values were 0.965, 0.986, and 0.995 for PCR, ANN, and the combined model prediction, respectively. It was found that combining the predictions from the PCR and ANN models reduced the root mean square errors (RMSE) of ozone concentrations. It is clear that combining predictions generated by different methods could improve the accuracy and provide a prediction that is superior to a single model prediction.

Effects of using the CB05 vs. SAPRC99 vs. CB4 chemical mechanism on model predictions: Ozone and gas-phase photochemical precursor concentrations

A three-dimensional air quality model is used to examine the magnitude and spatial distribution of differences in predictions among three chemical mechanisms that are used for regulatory and research modeling. The Carbon Bond mechanism CB05 is compared to an earlier version, CB4, to assess how much changes due to an update might potentially affect previous model conclusions on ozone concentrations and behavior. SAPRC-99 is compared to identify differences that might be expected between two more recent mechanisms, namely CB05 and SAPRC-99. The predicted ozone concentrations are similar for most of the United States, but statistically significant differences occur over many urban areas and the central US. SAPRC-99 predicts higher concentrations than CB05 on average, and both predict higher ozone than CB4. The difference in ozone predictions depends on location, the VOC/NO x ratio and concentrations of precursor pollutants. We highlight where the largest differences occur, give some explanation for why they occur, and discuss the effect of differences on model applications.

Evaluation of the Community Multiscale Air Quality (CMAQ) model version 4.5: Sensitivities impacting model performance: Part I—Ozone

This study examines ozone (O 3) predictions from the Community Multiscale Air Quality (CMAQ) model version 4.5 and discusses potential factors influencing the model results. Daily maximum 8-h average O 3 levels are largely underpredicted when observed O 3 levels are above 85 ppb and overpredicted when they are below 35 ppb. Using a clustering approach, model performance was examined separately for several different synoptic regimes. Under the most common synoptic conditions of a typical summertime Bermuda High setup, the model showed good overall performance for O 3, while associations have been identified here between other, less frequent, synoptic regimes and the O 3 overprediction and underprediction biases. A sensitivity test between the CB-IV and CB05 chemical mechanisms showed that predictions of daily maximum 8-h average O 3 using CB05 were on average 7.3% higher than those using CB-IV. Boundary condition (BC) sensitivity tests show that the overprediction biases at low O 3 levels are more sensitive to the BC O 3 levels near the surface than BC concentrations aloft. These sensitivity tests also show the model performance for O 3 improved when using the global GEOS-CHEM BCs instead of default profiles. Simulations using the newest version of the CMAQ model (v4.6) showed a small improvement in O 3 predictions, particularly when vertical layers were not collapsed. Collectively, the results suggest that key synoptic weather patterns play a leading role in the prediction biases, and more detailed study of these episodes are needed to identify further modeling improvements.

Evolution of surface ozone in central Italy based on observations and statistical model

Hourly and daily variations of surface ozone have been analyzed in relation to radon and meteorological parameters to explore its controlling mechanisms. Measurements in central Italy cover the years 2004 and 2005, showing a relevant role of transport in the ozone concentration variability. An analysis based on back trajectories shows that the site is affected by air masses originating from the west to northeast sector in about 74% of the days, suggesting that L’Aquila could be considered a background site. The background hypothesis is also supported by the rather low values of the following ozone quantities: maximum of monthly averages (39 ppbv, July), annual median of hourly data (29 ppbv), and annual average of hourly maxima recorded daily (49 ppbv). Only six hourly data recorded ozone above 90 ppbv in 2 years but never above 100 ppbv. The regression model reproduces measured ozone with accuracy in 67% of hourly observations and 74% of daily mean data. Here the model includes information from the following meteorological parameters: temperature, relative humidity, horizontal wind speed/direction, sun radiation, and radon concentration. A tracer like radon that tracks the dynamical changes of the lower atmosphere has a significant role in the model ozone prediction improvement, especially for hourly observations and for the synoptic component. In the first case (hourly observation), inclusion of radon data improves the regression model performance by 5% (from 62 to 67%); in the last case (synoptic component), the model accuracy increases by 3% (from 78 to 81%).

Correcting photolysis rates on the basis of satellite observed clouds

Clouds can significantly affect photochemical activities in the boundary layer by altering radiation intensity, and therefore their correct specification in the air quality models is of outmost importance. In this study we introduce a technique for using the satellite observed clouds to correct photolysis rates in photochemical models. This technique was implemented in EPA's Community Multiscale Air Quality modeling system (CMAQ) and was tested over a 10 day period in August 2000 that coincided with the Texas Air Quality Study (TexAQS). The simulations were performed at 4 and 12 km grid size domains over Texas, extending east to Mississippi, for the period of 24 to 31 August 2000. The results clearly indicate that inaccurate cloud prediction in the model can significantly alter the predicted atmospheric chemical composition within the boundary layer and exaggerate or underpredict ozone concentration. Cloud impact is acute and more pronounced over the emission source regions and can lead to large errors in the model predictions of ozone and its by‐products. At some locations the errors in ozone concentration reached as high as 60 ppb which was mostly corrected by the use of our technique. Clouds also increased the lifetime of ozone precursors leading to their transport out of the source regions and causing further ozone production down‐wind. Longer lifetime for nitrogen oxides (NOx = NO + NO2) and its transport over regions high in biogenic hydrocarbon emissions (in the eastern part of the domain) led to increased ozone production that was missing in the control simulation. Over Houston‐Galveston Bay area, the presence of clouds altered the chemical composition of the atmosphere and reduced the net surface removal of reactive nitrogen compounds. Use of satellite observed clouds significantly improved model predictions in areas impacted by clouds. Errors arising from an inconsistency in the cloud fields can impact the performance of photochemical models used for case studies as well as for air quality forecasting. Air quality forecast models often use the model results from the previous forecast (or some adjusted form of it) for the initialization of the new forecast. Therefore such errors can propagate into the future forecasts, and the use of observed clouds in the preparation of initial concentrations for air quality forecasting could be beneficial.

Air quality impacts of distributed power generation in the South Coast Air Basin of California 2: Model uncertainty and sensitivity analysis

Uncertainty and sensitivity of ozone and PM 2.5 aerosol to variations in selected input parameters are investigated with a Monte Carlo methodology using a three-dimensional air quality model. The selection of input parameters is based on their potential to affect concentration levels of ozone and PM 2.5 predicted by the model and to reflect changes in emissions due to the implementation of distributed generation (DG) in the South Coast Air Basin (SoCAB) of California. Numerical simulations are performed with the CIT air quality model. Response of the CIT predictions to the variation of selected input parameters is investigated to separate the potential air quality impacts of DG from model uncertainty. This study provides a measure of the model errors for selected species concentrations. A spatial sensitivity analysis is used to investigate the effect of placing DG in specific regions of the SoCAB. In general, results show that confidence in the model results is greatest in locations where ozone and PM 2.5 concentrations are the highest. Changes no greater than 80% in the nominal values of selected input variables, cause changes of 18% in ozone mixing ratios and 25% for PM 2.5 aerosol concentrations. Sensitivity analysis reveals that nitrogen oxides ( NO x ) emissions and side boundary conditions of volatile organic compounds (VOC) are the major contributors to uncertainty and sensitivity of ozone predictions. An increase in NO x emissions leads to reductions in ozone mixing ratios at peak times and sites where the maximum values are located. PM 2.5 aerosol is most sensitive to changes in NH 3 and NO x emissions. Increasing these emissions leads to higher aerosol concentrations. Sensitivity analyses show that the impacts of DG implementation are highly dependent on both space and time. In particular, ozone concentrations are reduced during the nighttime nearby locations where DGs are installed. However, during the daytime ozone concentrations increase downwind from the sources. A major finding of this study is that the emissions of DG installed in coastal areas produce a significant impact on the production of ozone and PM 2.5 aerosol in the eastern regions of the SoCAB.

Impacts of meteorological uncertainties on ozone pollution predictability estimated through meteorological and photochemical ensemble forecasts

This study explores the sensitivity of ozone predictions from photochemical grid point simulations to small meteorological initial perturbations that are realistic in structure and evolution. Through both meteorological and photochemical ensemble forecasts with the Penn State/NCAR mesoscale model MM5 and the EPA Community Multiscale Air Quality (CMAQ) Model‐3, the 24‐hour ensemble mean of meteorological conditions and the ozone concentrations compared fairly well against the observations for a high‐ozone event that occurred on 30 August during the Texas Air Quality Study of 2000 (TexAQS2000). Moreover, it was also found that there were dramatic uncertainties in the ozone prediction in Houston and surrounding areas due to initial meteorological uncertainties for this event. The high uncertainties in the ozone prediction in Houston and surrounding areas due to small initial wind and temperature uncertainties clearly demonstrated the importance of accurate representation of meteorological conditions for the Houston ozone prediction and the need for probabilistic evaluation and forecasting for air pollution, especially those supported by regulating agencies.

Fuzzy system models combined with nonlinear regression for daily ground-level ozone predictions

This study focuses on applying a Takagi–Sugeno fuzzy system and a nonlinear regression (NLR) model for ozone predictions in six Kentucky metropolitan areas. The fuzzy “c-means” clustering technique coupled with an optimal output predefuzzification approach (least square method) was used to train the Takagi–Sugeno fuzzy system. The fuzzy system was tuned by specifying the number of rules and the fuzziness factor. The NLR models were based in part on a previously reported, trajectory-based hybrid NLR model that has been used for years for forecasting ground-level ozone in Louisville, KY. The NLR models were each composed of an interactive nonlinear term and several linear terms. Using a common meteorological parameter set as input variables, the NLR models and the Takagi–Sugeno fuzzy systems model exhibited equivalent forecasting performance on test data from 2004. For all 2004 ozone season forecasts for the six metropolitan areas, the mean absolute error was 8.1 ppb for the NLR model and 8.0 ppb for the Takagi–Sugeno fuzzy model. When a nonlinear term (which was part of the NLR model) was included in the fuzzy model, the combined NLR–fuzzy model had slightly better performance than the original NLR model. For all 2004 metropolitan area forecasts, the mean absolute error of the NLR–fuzzy model forecasts was 7.7 ppb. These small differences may be statistically significant, but for practical purposes the performance of the fuzzy models was equivalent to that of the NLR models.