This study is a continuation of Stephens' preliminary study. The same RAMA data set was used. The data was averaged into three categories: Weekday, Saturday, and Sunday. Next, to establish statistical significance between weekday and weekend ozone concentrations, a two sample t-test was performed in the statistical program R. The same t-test was performed for weekday and weekend CO concentrations, and weekday and weekend NOx concentrations. The t-test results were compared and studied for possible connections. After the weekend effects were quantified, the next analysis involved calculating the VOC/NOx ratio. Both VOC and NOx were involved in ozone formation, so changes in this ratio can lead to changes in ozone concentrations. Ozone concentrations were plotted against the CO/NOx to find possible correlations.
However, CO (representing VOC) and NOx were not the only pollutants that could affect ozone concentrations. It was suspected that CO/NOx ratios, PM10 and SO2 were precursors that could affect ozone chemistry by scattering light needed for photolysis in ozone production. Hence, bivariate correlations were performed in the statistical program R between each precursor (CO, NOx, SO2, PM10) and ozone concentration in all of the five locations.
Finally, to properly identify the limiting reactant in ozone production, the ozone sensitivity coefficient was calculated. The RAMA data was used for this calculation, but only during the month of March from 1986-2006. Ozone, CO and NOx averages were computed for March at station MER using three consecutive values from the daily morning maxima. These values were used to find the sensitivity coefficient “r” which indicates whether the ozone levels were NOx sensitive or HC sensitive. The weekend effect analyses showed that there was a definite normal weekend effect for NOx and CO, but there was no consistent weekend effect for ozone. At times, ozone concentrations increased when NOx concentrations increased, but at other times, the opposite occurred. These conflicting results made it difficult to pinpoint the limiting reactant of ozone formation. The CO/NOx ratio was then calculated to clear up the confusion. There was a positive and almost linear relationship between ozone concentration and the CO/NOx ratio. This positive, linear relationship was not apparent in the early years (1986-1991), but gradually became evident by the later years (2000-2003). This means that as CO (VOC) increased, ozone concentrations increased as well. The bivariate correlation calculations showed that for CO/NOx ratio vs. ozone concentration, SW, NW, and SE regions had strong positive correlations. The Central region showed weak positive correlations, while the NE region showed weak negative correlations. So overall, ozone concentrations and the CO/NOx ratio have a positive relationship. There was no correlation between PM10 and ozone concentrations, and SO2 and ozone concentrations.
The sensitivity coefficient was calculated for the month of March over the years of 1986-2006 in station MER. They showed that ozone concentrations are mainly VOC sensitive. In weekday to Saturday, there were four years with NOx sensitivity. In Saturday to Sunday, there were six years with NOx sensitivity. This finding agrees with Sherry Stephens' preliminary conclusion that from Weekday to Saturday, ozone concentrations were VOC sensitive, and from Saturday to Sunday, ozone was less VOC sensitive. There were a lot more instances of VOC sensitivity over the twenty year period than NOx sensitivity. Hence, it is reasonable to conclude that ozone concentrations were VOC sensitive in the MCMA.