Forecasting Ozone with Yesterday’s Meteorological Data

摘要

The study was undertaken, under the direction of the North Carolina Departmentrnof Environment and Natural Resources (NCDENR), to develop a predictive model forrntoday’s ozone in the Charlotte metropolitan area based upon yesterday’s ozone,rntemperature, and current meteorological variables. Accurately predicting ozone isrnimportant so that the public may be advised on how to deal with air pollution.rnForecasting is useful to conduct large-scale special studies to identify high pollution days.rnIt is also useful for employing supplementary control measures and voluntary airrnpollution reduction programs.rnIndividual models were developed for five ozone monitoring sites in Charlotte,rnNC using ozone and meteorological data supplied by NCDENR, along with backwardrnregression. Models were created for weekday and weekend data to account for changesrnin emission patterns. These predictive models accounted for up to 73% of the variabilityrnand contained up to 15 explanatory variables.rnThe models were able to predict with up to almost a 93% accuracy rate and willrnbe used by the NCDENR to help forecast ozone in Charlotte. Work is continuing onrntaking inter-correlation into account. By removing unnecessary variables the total costrnand effort of data collection could decrease, making future studies more efficient.