A0909
Title: Modeling weekly temperature-mortality dynamics in Italy (2011/2024) with EVT-based analysis
Authors: Francesca Serra - La Sapienza (Italy) [presenting]
Abstract: The aim is to investigate the relationship between extreme temperatures and weekly elderly mortality in Italy from 2011 to 2024 using a two-step approach. First, regional indicators of extreme heat (T90) and cold (T10) are developed by comparing weekly temperature values with historical thresholds from a 1985-2000 reference period. Weekly mortality rates are computed for five elderly age groups using official data. Initial findings suggest non-linear associations: Cold extremes (T10) are more strongly linked to excess mortality in northern and central regions, while heat extremes (T90) are more relevant in the South. To remove trends and seasonality, ARIMA and ARIMA-GARCH models are applied to both temperature and mortality series, and standardized residuals are extracted. In the second step, extreme value theory is used to analyze the joint tail behavior of residual temperature and mortality. A Peaks-over-Threshold method combined with bivariate copulas models the dependence in the extremes. Results reveal more frequent and intense co-extremes between high temperature anomalies and mortality, especially among the oldest groups in southern regions. Cold-related co-extremes appear weaker and more regionally dispersed. These findings highlight the value of modeling joint extremes to better understand climate-related health risks and inform adaptation strategies for vulnerable populations.
