Johan Lindström, 
Centre for Mathematical Sciences, 
Lund

Title: Spatio-Temporal Modelling of local ambient air pollution.

Abstract:
The Multi-Ethnic Study of Atherosclerosis (MESA) Air is a major cohort
study focused on the association between cardiovascular disease and
chronic exposure to ambient air pollution. In order for MESA Air to
estimate health effects we must first provide good predictions of ambient
air pollution at subject locations. Here I will describe the modelling and
prediction of NOx in Los Angeles and Baltimore.

The main focus of the study is on chronic exposure (i.e. long term
averages at each location). However, the long study period combined with
the prohibitive cost of long term monitoring at many locations has lead to
a very complex dataset; containing a mixture or long time series at a few
locations, and spatially rich data at a few time points. The complex
dataset has necessitated the development of a novel spatio-temporal model
for air pollution.

This talk will describe the available MESA Air NOx data in Los Angeles and
Baltimore and the statistical methods developed to predict air pollution
at subject locations. Finally I will also discuss our efforts to combine
observations with the output from deterministic air pollution models in
attempt to increase prediction accuracy.