Master of Science (MS)
Geography and Anthropology
Ground level ozone (O3) is a pollutant of great public health concern. Spatial interpolation techniques provide powerful tools in estimating O3 exposure, but many fall short when predicting O3 on complex surfaces, especially given the high local variability typically associated with O3 data. Like most other locations, the Baton Rouge, Louisiana, O3 non-attainment zone (BRNZ) is plagued by a sparse density of O3 monitoring stations. This research explores land use regression (LUR) as an alternative spatial prediction method in and around the BRNZ. Multiple years of data are used to partially compensate for the small sample of spatial points. To better associate O3 measurements with the localized land cover, deviations-from-the-regional mean (devRM) are utilized rather than direct observations (DO). Land cover data used did not perform well in predicting the daily maximum O3 but performed moderately well for longer averaging periods. A model using the monthly mean O3 maxima averaged over a three-year period was able to explain 42.04% of the variance in devRM data. Predicted devRM using this model accounts for 4.55% of the variance in DO, the regional mean accounts for 88.65% of the variance, and when summed, the regional mean and modeled devRM account for 93.50% of variance in DO O3 data. These results are useful for future refinement of LUR models and will be useful to environmental planners and epidemiologists as they evaluate and mitigate the effects of O3 in Louisiana.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Thomas, Mallory Nance, "Tropospheric Ozone Prediction with Land Cover Regression in Baton Rouge, Louisiana" (2016). LSU Master's Theses. 794.