Abstract:
Ozone concentrations recorded by a network of rural monitoring stations surrounding Chicago have previously been modeled as a non-linear function of meteorology recorded in Chicago (NISS Technical Report #5). Because some of the stations are actually closer to Detroit or St. Louis, it may be appropriate to include meteorology from these cities in the model. Investigation showed, however, that model performance is not improved by these additional data. Another possibility is that the stations are not similarly affected by Chicago meteorology, and that the non-linear model may best be fit using the meteorology of Chicago, St. Louis, or Detroit for three geographically-determined subnetworks of the stations. When this was done, the model was shown to fit reasonably well for each of the subnetworks. The resulting meteorologically-adjusted trend estimates for Chicago, St. Louis, and Detroit are -2.0%/decade, -12.5%/decade, and +0.5%/decade, respectively. The differences in the trend parameters are only significant for the St.Louis/Detroit comparison. This implies that the meteorologicallyadjusted trends in ozone concentration vary between these two subnetworks but that no conclusion can be drawn from this analysis regarding the differences in trend between Chicago and St. Louis or between Chicago and Detroit.
tr17.pdf