Application of geostatistical approaches to predict the spatio-temporal distribution of summer ozone in Houston, Texas.

Mitigation of adverse effects of air pollution requires understanding underlying exposures, such as ambient ozone concentrations. Geostatistical approaches were employed to analyze temporal trends and estimate spatial patterns of summertime ozone concentrations for Houston, Texas, based on hourly ozone observations obtained from the Texas Commission on Environmental Quality. We systematically assess the accuracy of several spatial interpolation methods, comparing inverse distance weighting, simple kriging, ordinary kriging, and universal kriging methods utilizing the hourly ozone observations and meteorological measurements from monitoring sites. Model uncertainty was assessed by leave-one-out cross-validation. Kriging methods performed better, showing greater consistency in the generated surfaces, fewer interpolation errors, and lower biases. Universal kriging did not significantly improve the interpolation results compared to ordinary kriging, and thus ordinary kriging was determined to be the optimal method, striking a balance between accuracy and simplicity. The resulting spatial patterns indicate that the more industrialized areas east and northeast of Houston exhibit the highest summertime ozone concentrations. Estimated daily maximum 8 h ozone concentration fields generated will be used to inform research on population health risks from exposure to surface ozone in Houston.