Mean Fractional Bias Research Articles

Assessment of Biomass Burning Emissions and Their Impacts on Urban and Regional PM2.5: A Georgia Case Study

Biomass burning is a major and growing contributor to particulate matter with an aerodynamic diameter less than 2.5 microm (PM2.5). Such impacts (especially individual impacts from each burning source) are quantified using the Community Multiscale Air Quality (CMAQ) Model, a chemical transport model (CTM). Given the sensitivity of CTM results to uncertain emission inputs, simulations were conducted using three biomass burning inventories. Shortcomings in the burning emissions were also evaluated by comparing simulations with observations and results from a receptor model. Model performance improved significantly with the updated emissions and speciation profiles based on recent measurements for biomass burning: mean fractional bias is reduced from 22% to 4% for elemental carbon and from 18% to 12% for organic matter; mean fractional error is reduced from 59% to 50% for elemental carbon and from 55% to 49% for organic matter. Quantified impacts of biomass burning on PM2.5 during January, March, May, and July 2002 are 3.0, 5.1, 0.8, and 0.3 microg m(-3) domainwide on average, with more than 80% of such impacts being from primary emissions. Impacts of prescribed burning dominate biomass burning impacts, contributing about 55% and 80% of PM2.5 in January and March, respectively, followed by land clearing and agriculture field burning. Significant impacts of wildfires in May and residential wood combustion in fireplaces and woodstoves in January are also found.

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 SF 6 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.

PM and light extinction model performance metrics, goals, and criteria for three-dimensional air quality models

In order to use an air quality modeling system with confidence, model performance must be evaluated against observations. While ozone modeling and evaluation is fairly developed, particulate matter (PM) modeling is still an evolving science. EPA has issued minimal guidance on PM and visibility model performance evaluation metrics, goals, and criteria. This paper addresses these issues by examining various bias and error metrics and proposes PM model performance goals (the level of accuracy that is considered to be close to the best a model can be expected to achieve) and criteria (the level of accuracy that is considered to be acceptable for modeling applications) that vary as a function of concentration and extinction. In this paper, it has been proposed that a model performance goal has been met when both the mean fractional error (MFE) and the mean fractional bias (MFB) are less than or equal to +50% and ±30%, respectively. Additionally, the model performance criteria has been met when both the MFE⩽+75% and MFB⩽±60%. Less abundant species would have less stringent performance goals and criteria. These recommendations are based upon an analysis of numerous PM and visibility modeling studies performed throughout the country.

A new pseudodeterministic multivariate receptor model for individual source apportionment using highly time‐resolved ambient concentration measurements

A new multivariate pseudodeterministic receptor model (PDRM), combining mass balance and Gaussian plume dispersion equations, was developed to exploit highly time‐resolved ambient measurements of SO2 and particulate pollutants influencing air quality at a site in Sydney, Florida, during the Tampa Bay Regional Aerosol Chemistry Experiment (BRACE) in May 2002. The PDRM explicitly exploits knowledge of the number and locations of major stationary sources, source and transport wind directions, stack gas emission parameters, and meteorological plume dispersion parameters during sample collections to constrain solutions for individual sources. Model outputs include average emission rates and time‐resolved ambient concentrations for each of the measured species and time‐resolved meteorological dispersion factors for each of the sources. The model was applied to ambient Federal Reference Method SO2 and 30‐min elemental measurements during an 8.5‐hour period when winds swept a 70° sector containing six large stationary sources. Agreement between predicted and observed ambient SO2 concentrations was extraordinarily good: The correlation coefficient (R2) was 0.97, their ratio was 1.00 ± 0.18, and predicted SO2 emission rates for each of four large utility sources lie within 8% of their average continuous emission monitor values. Mean fractional bias, normalized mean square error, and the fractions of the predictions within a factor of 2 of the observed values are −2.7, 0.9, and 94%, respectively. For elemental markers of coal‐fired (As and Se) and oil‐fired (Ni) power plant emissions the average ratio of predicted and observed concentrations was 1.02 ± 0.18 for As, 0.96 ± 0.17 for Se, and 0.99 ± 0.41 for Ni, indicating that the six sources located in the wind sector between approximately 200° and 260° well accounted for background‐corrected concentrations measured at the sampling site. Model results were relatively insensitive to the choice of upper bound used to constrain solutions.

ISC모델의 적용성 평가 - 소각장 주변지역의 단기농도예측

The short-term version of Industrial Source Complex Model(ISCST3) was evaluated for estimating short-term concentrations using criteria pollutant(SO<TEX>$_2$</TEX>, NO<TEX>$_2$</TEX>, CO, PM10) data from emission inventory of Young Tong area in Suwon for the year 2002. The contribution of pollutant concentration from point, line, area sources was found 21.8, 76.5 and 1.6%. Statistical parameters, such as correlation coefficient, index of agreement(IA), normalized mean square error(NMSE) and fractional bias(FB) were calculated for each pollutants. The model performance were found good for PM10(82%) and NO<TEX>$_2$</TEX>(69%), but poor for SO<TEX>$_2$</TEX>(34%) and CO(13%).

Performance of an industrial source complex model: Predicting long‐term concentrations in an urban area

AbstractThe short‐term and long‐term versions of Industrial Source Complex Models (ISCST3 and ISCLT3) are evaluated for estimating long‐term concentrations using sulfur dioxide data from emission inventory of Lucas County, Ohio for the year 1990. The ISCST3 and ISCLT3 models are used to predict monthly and quarterly averages of sulfur dioxide concentrations. Uncertainties associated with the model's predictions are estimated using the bootstrap resampling method. Confidence intervals on the fractional mean bias, normalized mean square error, geometric mean bias and geometric variance are calculated for each model and differences between models. Intercomparison of the ISCST3 and ISCLT3 models indicated that these models yielded relatively good performance in their prediction of monthly and quarterly average concentrations, with relative fractional biases of 0.26 to 0.55 and normalized mean square error values that are about 0.12 to 0.44. Both the ISCST3 and ISCLT3 models predicted concentrations that are lower than the observed concentrations. The concentrations predicted by the ISCST3 model were closer to the observed concentrations when compared with the concentrations obtained using the ISCLT3 model. The study suggests that the ISCST3 model is better for estimating long‐term concentrations of sulfur dioxide as compared to the ISCLT3 model.

Hybrid plume dispersion model (HPDM) improvements and testing at three field sites

Descriptions of several technical improvements to the Hybrid Plume Dispersion Model (HPDM) are given. The boundary-layer meteorological preprocessor now makes use of a surface moisture availability parameter and now accounts for mechanical mixing due to buildings in urban areas. A new dispersion algorithm has been added in order to simulate conditions better when a buoyant plume “lofts” against the capping inversion and spreads laterally before dispersing down to the ground. The modified model has been evaluated using 89 h of SF 6 tracer data at an urban power plant in Indianapolis, IN, and using a full year (8760 h) of SO 2 data at the Baldwin, IL, power plant and the Summit County, OH, industrial complex. The EPA's RAM and ISC models are also included in the evaluation exercises. Emphasis is on the second-highest concentrations, the correlation, the fractional mean bias and the fraction of predictions within a factor of two of observations for 1-, 3- and 24-h averaging times. The HPDM model exhibits significantly better performance (at the 95% confidence level) than the RAM or ISC model for most sites and performance measures. In most cases, the HPDM “second-highest” predictions are within 10–20% of the observations and the fractional mean bias values are less than 10%. Furthermore, the relative error of HPDM shows little trend with input variables such as wind speed and mixing depth.

A puff model simulation of tracer concentrations in the nocturnal drainage flow in a deep valley

During the 1984 ASCOT field study in Brush Creek, CO, two gaseous tracers were released into valley drainage flow at different heights at the same site on 5 nights. This paper describes simulations of surface concentration data from two tracer experiments, one elevated release and one surface release, which complement tose reported by Rao et al. (1989, J. appl. Met. 28, 609–616). The integrated Gaussian puff model adapted by them has been improved and used for these new simulations. The improved model, VALPUFF, affords a physically realistic simulation of the mean transport for elevated plumes under highly variable emission conditions. The model simulations are evaluated by comparing predicted hourly concentrations with the corresponding observed values at 51 surface samplers. For the elevated release case, the observed mean concentration is 28 pl ℓ −1, and the predicted mean is 31 pl ℓ −1. Almost 85% of predictions are within a factor of 5 of observations, and 50% within a factor of 2. For the ground release case, the observed the predicted means are 128 and 183 pl ℓ −1, respectively. About 53% of predictions are within a factor of 5 of observations, and 23% within a factor of 2. The effects of varying wind field inputs and turbulence parameterizations on model performance are tested using several variations of the VALPUFF model. Cumulative frequency distributions of the ratio of modeled and observed hourly concentrations, and other evaluation statistics, are presented for each model. Applying resampling methods, 95% confidence intervals for fractional mean bias and normalized mean square error are calculated for the predictions of each model and for differences between the predictions of various models. It is shown that the VALPUFF model that uses vertically-averaged winds (for elevated releases) and dispersion parameters based on on-site turbulence data gives the best overall performance.

Confidence limits for air quality model evaluations, as estimated by bootstrap and jackknife resampling methods

Air quality models are used to make decisions regarding the construction of industrial plants, the types of fuel that will be burnt and the types of pollution control devices that will be used. It is important to know the uncertainties that are associated with these model predictions. Standard analytical methods found in elementary statistics textbooks for estimating uncertainties are generally not applicable since the distributions of performance measures related to air quality concentrations are not easily transformed to a Gaussian shape. This paper suggests several possible resampling procedures that can be used to calculate uncertainties or confidence limits on air quality model performance. In these resampling methods, many new data sets are drawn from the original data set using an empirical set of rules. A few alternate forms of the socalled bootstrap and jackknife resampling procedures are tested using a concocted data set with a Gaussian parent distributions, with the result that the jackknife is the most efficient procedure to apply, although its confidence bounds are slightly overestimated. The resampling procedures are then applied to predictions by seven air quality models for the Carpinteria coastal dispersion experiment. Confidence intervals on the fractional mean bias and the normalized mean square error are calculated for each model and for differences between models. It is concluded that these uncertainties are sometimes so large for data sets consisting of about 20 elements that it cannot be stated with 95% confidence that the performance measure for the ‘best’ model is significantly different from that for another model.