This work illustrates a model-based clustering method for analyzing PM10 measurements over time. In particular, we develop a Bayesian dynamic linear model coupled with a spatial product partition model for clustering monitoring stations that exhibit similar persistence and variability of the PM10 concentrations over time. The model integrates spatial information (the locations of the considered monitoring stations) into the clustering process in order to increase the probability that neighboring stations will be assigned to the same cluster. This methodology is applied to the time series of daily PM10 measurements recorded by 110 monitoring stations in Austria. Our analysis reveals three spatially cohesive clusters characterized by different levels of persistence and variability of the PM10 concentrations. These results may provide helpful insights for understanding air pollution dynamics and support policymakers in identifying intervention areas.
Aiello, L., Legramanti, S., Paci, L., A spatial product partition model for PM10 data, in Methodological and Applied Statistics and Demography III, (Bari, 17-20 June 2024), Springer, Cham 2024:<<ITALIAN STATISTICAL SOCIETY SERIES ON ADVANCES IN STATISTICS>>, 8-13. [10.1007/978-3-031-64431-3] [https://hdl.handle.net/10807/310417]
A spatial product partition model for PM10 data
Paci, Lucia
2024
Abstract
This work illustrates a model-based clustering method for analyzing PM10 measurements over time. In particular, we develop a Bayesian dynamic linear model coupled with a spatial product partition model for clustering monitoring stations that exhibit similar persistence and variability of the PM10 concentrations over time. The model integrates spatial information (the locations of the considered monitoring stations) into the clustering process in order to increase the probability that neighboring stations will be assigned to the same cluster. This methodology is applied to the time series of daily PM10 measurements recorded by 110 monitoring stations in Austria. Our analysis reveals three spatially cohesive clusters characterized by different levels of persistence and variability of the PM10 concentrations. These results may provide helpful insights for understanding air pollution dynamics and support policymakers in identifying intervention areas.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.