Agenda

26 May 2026 11:00

Modelling Spatial Dependence of Urban PM₁₀ Concentrations Using a Copula-Based Approach

Aula DELTA 2C - Edificio DELTA | Campus Scientifico

Speaker:
Vincenzo Gioia, Università di Trieste

Link Zoom             Meeting ID: 851 5326 8624             Passcode: SanMarco2

Abstract:
Understanding the spatio-temporal dynamics of particulate matter (PM₁₀) is essential for environmental monitoring and public health assessment. The talk presents an evolving research programme on the joint modelling of daily PM₁₀ concentrations. Building on earlier work where a multivariate Gaussian additive model with dynamic covariance matrix was used to explore the role of meteorological drivers on local variability and pairwise dependence, we extend the modelling framework by adopting a two-stage Bayesian procedure based on the Inference Functions for Margins approach. In the first stage, flexible marginal models, incorporating seasonality, meteorological covariates, and temporal dependence, are estimated independently for each station via Hamiltonian Monte Carlo in Stan. The resulting probability integral transform residuals are then used in the second stage to estimate the spatial dependence structure through parametric copula models. Gaussian and Student-t copulas are compared with a family of Archimedean copula models, allowing for upper tail dependence and non-Gaussian spatial dependence structures. The approach is illustrated using daily PM₁₀ measurements from three monitoring stations in Friuli Venezia Giulia covering the period 2015–2025. The proposed framework proves effective at capturing spatial asymmetries and estimating the probability of joint extreme events across stations.

Bio sketch:
Vincenzo Gioia is a post-doctoral researcher in Statistics at the Department of Economics, Business, Mathematics and Statistics "Bruno de Finetti", University of Trieste. His main research interests are in the area of semi-parametric statistical modelling and computational statistics.

Language

The event will be held in English

Organized by

Gruppo Statistica (Prosdocimi)

Search in the agenda