Date on Master's Thesis/Doctoral Dissertation


Document Type

Master's Thesis

Degree Name



Mechanical Engineering

Committee Chair

Cobourn, W. Geoffrey

Author's Keywords

Applied sciences; Pure sciences


Atmospheric ozone--Forecasting; Atmospheric ozone--Kentucky


Ground-level ozone forecast models were developed for the following middle and small metropolitan areas in Kentucky: Ashland, Bowling Green, Owensboro, and Paducah. These models were nonlinear regression models, based on models previously developed for Louisville and Lexington. For each of the four cities, the mean absolute errors (MAE) for the model estimates, based on the 1998-2002 model-fitted data sets, were less than 7.7 ppb; the MAE/O3 were less than 12.7%. The models could explain at least 66% of the variance of the daily peak ozone. On average, the errors of the model were within +/- 15.0 ppb on 88% of days, and were within +/- 10.0 ppb on 73% of days. Using an alarm threshold 80 ppb, the detection rates for National Ambient Air Quality Standard (NAAQS) Exceedences ranged from 0.48 to 0.67 for the four cities. The corresponding false alarm rates ranged from 0.29 to 0.44. The results of this study demonstrate that the ozone forecast models for each of the four cities can be expected to be useful tools for making next-day forecasts of local ground-level O3 in those areas. Similar models, updated using 2003 data, will be used during the 2004 O3 season for providing daily automated forecasts for these metropolitan areas.