| Ozone is a secondary pollutant whose concentration in ambient air depends on some precursors and some meteorological parameters. Part I is devoted to developing relationships between ozone concentration and these variables. Some of these variables affect ozone instantly while others show some time delay. The appropriate time delay can be determined by the maximum statistical correlation coefficient between ozone and these variables. Using these time delays, a multi-linear relationship can be established by using regression analysis and non-linear neural network to forecast ozone. Using 300 continuous data points for 3 different locations at Houston, Galveston and Beaumont areas for different times, the regression equations and neural network models were developed for prediction of ozone in Southeast Texas. The R2 values were found up to 0.88 for regression models and up to 0.79 for neural network models.; An experimental method was developed to determine the amount of HONO in ambient air. The total reactive nitrogen in air was extracted in the form of nitrite/nitrates and was analyzed by IC. The background nitrite/nitrate caused by the extraction of NOx was determined by extracting standard NOx in ambient concentration. The difference (NOy-NO x) represents the concentration of non-NOx reactive nitrogen, mainly HONG, PAN and HNO3. HONO and PAN was generated in the laboratory and extracted in the same device to calibrate the apparatus. The scrubbing efficiency of the scrubber was also determined. |
