Search Results
Abstract
Statistical measures for evaluating the performance of urban air quality models have recently been strongly recommended by several investigators. Problems that were encountered in the use of recommended performance measures in an evaluation of three versions of an urban photochemical model are described. The example demonstrates the importance of designing an evaluation to take into account the way in which the model will be used in regulatory practice, and then choosing performance measures on the basis of that design. The evaluation illustrates some limitations and possible pitfalls in the use and interpretation of statistical measures of model performance. Drawing on this experience, a procedure for evaluation of air quality models for regulatory use is suggested.
Abstract
Statistical measures for evaluating the performance of urban air quality models have recently been strongly recommended by several investigators. Problems that were encountered in the use of recommended performance measures in an evaluation of three versions of an urban photochemical model are described. The example demonstrates the importance of designing an evaluation to take into account the way in which the model will be used in regulatory practice, and then choosing performance measures on the basis of that design. The evaluation illustrates some limitations and possible pitfalls in the use and interpretation of statistical measures of model performance. Drawing on this experience, a procedure for evaluation of air quality models for regulatory use is suggested.
Abstract
Regression models have been used with poor success to detect the effect of emission control programs in ambient concentration measurements of carbon monoxide. An advanced CO regression model is developed whose form is based on an understanding of the physical processes of dispersion. Its performance is shown to be superior to the more traditionally developed regression and time series models. The model separates the effects of emissions change from the effects of fluctuations in meteorological conditions. The separation appears to be most reliable for winter conditions. The model has sufficient precision to identify present trends in emissions ambient concentration data. This model should be useful for detecting changes in emission trends due to implementation of a control program on vehicular emissions such as an inspection and maintenance program.
Abstract
Regression models have been used with poor success to detect the effect of emission control programs in ambient concentration measurements of carbon monoxide. An advanced CO regression model is developed whose form is based on an understanding of the physical processes of dispersion. Its performance is shown to be superior to the more traditionally developed regression and time series models. The model separates the effects of emissions change from the effects of fluctuations in meteorological conditions. The separation appears to be most reliable for winter conditions. The model has sufficient precision to identify present trends in emissions ambient concentration data. This model should be useful for detecting changes in emission trends due to implementation of a control program on vehicular emissions such as an inspection and maintenance program.
