Page 51 - Contributed Paper Session (CPS) - Volume 8

CPS2179 Giuliana Passamani et al.
aggregation processes, or as in Plaia, Di Salvo, Ruggeri and Agrò (2013), where
they suggest an index based first on a spatial aggregation and then on a
pollutants synthesis.
Our proposal is to provide a synthetic measure of air pollution in a given
town/place by means of a stochastic dynamic-factor model where we combine
the daily measurements of air pollutants with the meteorological conditions,
taking into account also the pollution measure observed the day before. That
is, we suggest a multipollutant synthesis based on human-caused and natural
sources emissions levels and their relationship with meteorological factors and
with the lagged pollution level. Giving that meteorological conditions can
change within a distance of some miles, the daily measures for each pollutant
cannot simply be aggregated over different monitoring sites, but, first of all,
we must analyse and aggregate different pollutants at the same monitoring
site for which we have also the corresponding observations on meteorological
variables. Once the synthetic air pollution measure has been calculated in a
given place, an aggregated AQI could be suggested for an entire area of
interest. After these premises we state that the focus in the present paper is
the proposal of a statistical methodology aiming at evaluating the different
ambient air qualities in the Province of Trento that covers an almost entirely
mountain area located in the northern part of Italy. The data set is made up of
the time series observations relative to three pollutants, particulate matter 10
(PM10), nitrogen dioxide (NO2) and ozone (O3): these are now generally
recognized as the three main pollutants that most significantly affect human
1
health . The pollutants are observed at different monitoring sites located in
traffic and non-traffic areas. Time series observations of some meteorological
variables are also available for the same monitoring sites. The empirical
analysis is based on a two years' period, 2014-2015.
The plan of the paper is the following. In Section 2 we discuss the
methodological approach and the dynamic-factor model adopted for
analysing the available daily time series dataset. In Section 3 we describe the
results in terms of the estimated pollution indicator for each site and in Section
4 we conclude the paper and outline possible lines for further research.

2. Methodology
The principle which is at the basis of a dynamic-factor model is that few
unobservable dynamic factors drive the co-movements observed in a higher
dimensional vector of endogenous time series variables which can also be
affected by exogenous variables, as well as by a vector of mean-zero
idiosyncratic disturbances. For the purpose of our approach in which we
assume the existence of a single dynamic factor underlying the observed

1 European Environment Agency: https://www.eea.europa.eu/themes/air/intro
40 | I S I W S C 2 0 1 9