Smoothed Common Trend in Multivariate Time Series Air Pollution Data

Air quality improvement is a fundamental issue, given the possible effects of air pollution on human health. The paper focuses on detecting the trend component of an unobserved dynamic factor that we consider as a pollution indicator. This factor emerges from the multivariate monthly time series analysis of the levels of four air pollutants measured at four different monitoring sites in the Alpine Province of Trento over the last ten years. In particular, we suggest a procedure that can be used in order to assess whether any improvement in the pollution level has been observed during the period of observation. The suggested procedure is based on the double exponential smoothing which is simpler than state space time series analysis and gives comparable results.