Spatiotemporal Analysis of PM_2.5 Exposure in Taipei (Taiwan) by Integrating PM₁₀ and TSP Observations

Abstract

Many studies have shown a significant association between human exposure to Particulate Matter (PM) and population health effects (premature mortality, respiratory and cardiovascular diseases, emergency room visits, and systemic inflammation). Fine PM particles (PM_2.5) are believed to be more dangerous than PM₁₀ because fine particles are easier inhaled and accumulated deeply into human lungs. Taipei is the largest city in Taiwan, where a variety of industrial and traffic emissions are continuously generated across space and time. Thus, it is crucial for health agencies to improve their understanding of spatiotemporal PM_2.5 exposure of people living in Taipei city. The Bayesian Maximum Entropy (BME) theory of spatiotemporal statistics and science-based mapping provides valuable information about population exposure to PM_2.5 pollution in Taipei. PM-related data (PM₁₀, PM_2.5 and Total Suspended Particles, TSP) are collected by different central and local government institutes. BME analysis introduces concepts and techniques that have a number of important features (theoretical and computational): several kinds of site-specific data and core knowledge bases are integrated that are associated with different physical scales; a variety of uncertainty sources are taken into account; non-linear, in general, PM estimators are used at unobserved locations; non-Gaussian laws are automatically incorporated; and a complete characterization of the dynamic PM distribution is obtained in terms of the probability density function at each space-time point rather than a single PM value. These BME advantages have considerable consequences as far as health risk analysis is concerned. Detailed space-time PM_2.5 maps account for (i) composite space-time dependence structure of PM values, (ii) hard and soft datasets available about PM_2.5, PM₁₀ and TSP, and (iii) empirical evidence about the $\frac{\textrm{PM}_{2.5}}{\textrm{PM}_{10}}$ and $\frac{\textrm{PM}_{10}}{\textrm{TSP}}$ ratios. PM measures are investigated, including the fraction of fine particles that vary considerably across space-time. BME analysis properly identifies and quantifies factors that influence the spatiotemporal patterns of these measures, such as weather conditions and land use (e.g., the PM distributions in highly-developed commercial or industrial areas generally have higher fine particle fractions). Information generated by rigorous BME analysis and mapping across space-time constitutes valuable input to health management and decision-making under conditions of uncertainty.