Modelling rainfalls by dynamic copulas

Supervisors: Mélina Maillhot (Concordia University, Montreal) and Elena di Bernardino (CNAM, Paris).
Postdoctoral Fellow: Fátima Palacios Rodríguez (INRIA Sophia Antipolis- Méditerranée, Montpellier)

The main aim of the project is to provide adequate models for rainfalls time series by representing the time dependence related to flood events with Vine copulas. Copulas are easily understandable multivariate models allow for defining any type of dependence structure between random variables. Vine copulas are characterizes by trees of copulas. Firstly, we will concentrate our study on rainfall in Paris, since data and models (excluding copulas) are available and will allow a comparison of the methods. Secondly, in order to quantify the risk of floods in the considered geographic area, we aim to provide appropriate estimators of suitable multivariate risk measures using our proposed models. Finally, we would like to help to improve insurance’s regulation in Canada to ensure considerable prices of flood risk coverage.

Faculty Supervisor:

Melina Mailhot

Student:

Partner:

Conservatoire National des Arts et Métiers;Inria Sophia Antipolis - Méditerranée Research Centre

Discipline:

Mathematics

Sector:

Education

University:

Concordia University

Program: