PREDEX - Predictability of Extreme Weather Events

Lead Research Organisation: University of Exeter
Department Name: Engineering Computer Science and Maths


This collaborative interdisciplinary project will deliver and apply new mathematical complexity methods for quantifying the predictability of extreme events in complex dynamical systems. Our new tools will specifically help three European Weather Services improve their capability to forecast extreme weather events (e.g. catastrophic windstorms) - a rapidly emerging real world challenge expected due to anthropogenic climate change. The questions we address are: 1) What are the deterministic time limits of predictability for extreme weather events?2) How can we use emergent dynamical patterns to enhance predictability of extreme weather events?3) What is the role of spatial scale interactions in the physical processes leading to extreme weather? Our approach is based on properties of the atmospheric system which are dynamical (evolve in time, with relevant timescales ranging from hours to years) and emergent (arise from the nonlinear interaction of physical processes having different spatiotemporal scales). The project is strongly grounded on mathematical and physical sciences and is designed around collaboration with meteorology and weather forecasting operational services. The project investigators provide a unique combination of world-leading expertise from different disciplines: atmospheric and climate sciences, the mathematical theory of dynamical systems, the statistical theory of extreme events. Knowledge exchange will be built on strong existing working relationships with three European National Weather Services: the Met Office, the Met \'Eireann and the KNMI. Members of these Services will be involved in steering the project towards real-world applications and monitoring the achievements.

Planned Impact

This project will develop new mathematical tools urgently needed for the better prediction of the more severe weather events. There is growing interest in the probabilistic forecasting of hazardous meteorological events. Climate change is expected to lead to an increase in the frequency and severity of extreme weather events, including catastrophic windstorms (see e.g. the 4th assessment report of the IPCC, 2007). Hence, improved forecasting of such extreme events is a pressing challenge for National Weather Services. This project will deliver proof-of-concept methodologies for that purpose. Involvement of the three National Weather Services will be crucial to focus the project on techniques which can then be further developed and implemented in operational setting, such as the MOGREPS ensemble prediction system of the Met Office. The techniques developed in this project will be applicable to more general nonlinear deterministic systems, including models from economy. Therefore, the potential domain of exploitability is broad, including all those cases in which deterministic models are used to produce forecasts of extreme events. The main beneficiaries of this project are the three National Weather Services involved in this project. They will benefit from this research by obtaining improved methodologies to be used for producing probabilistic forecasts of severe weather events. The benefits consist therefore of: * shaping and enhancing the effectiveness of the public service provided by these Weather Services; * contributing toward environmental sustainability, protection and impact, and * enhancing the efficiency, performance and sustainability of businesses/organisations including public services through better prediction of hazardous events such as windstorms; Other benefits include training skilled researchers for non-academic professions, such as jobs in National Weather Services, but also in commercial companies operating in weather risk (e.g. providing risk mitigation and management procedures for insurance and reinsurance companies) and prediction (e.g. delivering tailored weather forecasts for the offshore oil industry and the wind energy sector).


10 25 50
Description For weather forecasting, predicting the occurrence and frequency of extreme events is of significant interest. For weather models of low and intermediate complexity, the PREDEX project developed new theoretical approaches to predict extremes and statistics of recurrence. In this set up an extreme event is characterized by a climatic observable on the system (such as wind speed, or energy) exceeding a critical value. In the model set up, the weather system evolved according to a deterministic rule, such as differential equation or discrete iterative map, but exhibiting chaotic behaviour. A significant outcome of the project was the realisation that the statistics of extremes could be understood in terms of the geometry of the underlying chaotic attractor: the (fractal) subset of phase space for which the system converges to as time evolves.

As part of the PREDEX project, a theoretical formula was established between the fractal dimension of the underlying chaotic attractor and the tail index of the probability distributions that govern the extremes. Such a formula may be of practical interest, especially to those working in weather/climate who wish to quantify extreme risk.

For certain weather models, and for certain specified (extreme) events, progress within PREDEX was made on determining the set of initial conditions that are most likely to lead to an extreme event within a given time frame. The underlying problem here is to determine whether extreme events are more (or less?) predicable relative to other specified weather events. A finding is that the predictability of extremes depends formally on the observable and weather model under consideration.
Exploitation Route Potential uses include: providing modelling tools to assist those who work in the prediction of extreme events. The focus on PREDEX has been within low complexity weather models, but the tools developed would equally assist those who work on the prediction of extremes in economics and finance. In other contexts, the findings of PREDEX may assist policy makers in land management, security and environment (such as in flood defense).
This research is of interest to those who want to quantify the statistics of extremes using mathematical modelling. The underlying questions this research explores include: How frequently do extreme events occur? If an extreme event occurs, how long do we have to wait until it occurs again? Are extreme events easy (or hard) to predict? If the evolution of the process (in time) is given as a mathematical model, then the PREDEX project gives methods to measure the typical statistics of extremes as time evolves.
Sectors Environment,Financial Services, and Management Consultancy

Description The aim of the research is to predict extreme events. So far the impact has been academic inpact through publications, public engagement lectures and international conference/workshop presentations. A potential aim is to develop early warning systems in the case of predicting weather extremes. It is anticipated that such research will be carried out in collaboration with the Met Office services in the first instance.
First Year Of Impact 2012
Sector Education,Environment
Impact Types Societal