Hierarchical Model to Separate Background and Local Sources in PM2.5 Concentrations at Pittsburgh

[5]) show that exposure to particulate matter(or PM, which is a complex mixture of extremely small particles and liquid droplets in the air) have various short-term and long-term health effects, e.g.: 1. associations between daily average ambient PM concentrations and corresponding cardiopulmonary mortality, morbidity, and functional impairments; 2. significant association between PM10(particulate matter with aerodynamic diameters less than 10µm in diameter) and mortality, especially for short averaging period such as a day; 3. increased mortality rates (decreased life expectancy) in cities with higher average ambient PM2.5(particulate matter with aerodynamic diameters less than 2.5µm in diameter) and sulfate concentrations; 4. association between decrease of PM concentration and increase in mean life expectancy. PM mainly causes cardiac and respiratory health effects to human body through several pathways[1], and PM concentrations are now widely used as an indicator of air pollution extent. Our study is particularly focused on PM2.5. Data about PM2.5 as well as other pollutants have been collected in Pittsbugh, PA and vicinity in the past a few years. PM2.5 concentration can be affected by several factors. For example, the particles can come from industries, automobiles or photo chemical process in the air. Environmental/meteorological variables, like mixing height, and wind direction , can also affect PM2.5 concentrations. The mixing height is the height of the inversion layer. It basically means the highest height within which the atmosphere is mixing and thus the pollutants can reach. It was recognized([10]) for a long time that the PM2.5 has two main type of sources: one is the PM2.5 formed somewhere far from Pittsburgh(mainly from states west to Pennsylvania); the other is from all the local emission sources, e.g., steel factories, coke works, power plants,etc. The difference between these two types is: the former is well mixed during long distance transport, and therefore is dispersed relatively uniformly spatially in the whole Pittsburgh area (although it still fluctuates with other factors like the meteorological variables); the later is emitted from local sources, and therefore has some non-uniform spatial distribution since it doesn't get a chance to disperse (uniformly). A natural question is: which type contributes more to the total concentration at Pittsburgh? Or more precisely, how can we separate the levels of those two type of sources? Practically 1 the answer to this question sheds light on methods to mitigate PM2.5 and could hopefully make people live longer and healthier. For example, report([4]) shows that a …