Aggregating the distribution tails in multi-model ensemble outputs

While accurate temperature and precipitation projections are fundamental for climate assessments, traditional multi-model ensembles often struggles to balance global accuracy with regional precision, particularly for extreme events. To address these biases, a distribution-based aggregation method known as α-pooling is employed, which combines and statistically weights different model outputs to produce a more robust result. Although α-pooling is typically applied to entire distributions, this project focuses on the tail of the distributions. By integrating models from extreme value theory,the method’s ability to reproduce the tail behavior of observations is evaluated on two climate variables: temperature and precipitation intensity.

This project is a collaboration with Emilia Siviero, postdoctoral researcher and Carlo Gatean, Professor, both at the Department of Environmental Sciences, Informatics and Statistics, Universita Ca’Foscari (Venice, Italy)