Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

When disaster strikes, its vital to understand how people will be affected – so that humanitarian aid can be effectively targeted. Researchers in the Department of Statistics have used EPSRC IAA funding to develop new and effective disaster risk models using open access datasets and machine learning techniques.

Collapsed buildings after an earthquake © Angelo Giordano from Pixabay

Predicting the impact of natural and human-induced hazards on populations is crucial to disaster preparedness and response.

Most disaster risk models (DRMs) focus on economic loss rather than impact on people. Moreover, predictive models are often the preserve of the insurance or commercial sector, have not been subjected to scientific peer review, and are primarily designed to calculate loss of insured assets.

In recent years, however, the availability of open-access datasets such as satellite and drone-based earth observation data, and the development of new artificial intelligence models with which to analyse them, have created potential for better, more accurate disaster risk forecasting.

The Oxford Disaster Damage Real-time Information Network (ODDRIN), established by a team at the Department of Statistics working in collaboration with the humanitarian agency Internal Displacement Monitoring Centre, aimed to develop a more effective disaster risk model to predict population displacement.

The team developed a ‘backend’ statistical tool to analyse open access datasets and a ‘front-end’ system which interactively visualises the data and predictions. The approach has been used to analyse data from 180 earthquakes worldwide, as well as cyclone Harold in the Pacific, proving ten times more accurate than leading disaster risk models.

The team now hope to develop the capacity of the model to predict floods and tropical cyclones and to add data on mortality and building damage – which will ensure ODDRIN predictions are comparable to that of leading DRM codes.

The ultimate ambition for ODDRIN is for it to be integrated into real-time disaster risk information systems such as the European Commission’s Global Disaster Alert and Coordination System and the United States Geological Survey; and for the model to inspire others to use more advanced statistical models and methodologies to develop open-source predictive tools which can be used in any context, including in low-income countries.

Prof David Steinsaltz says: “ODDRIN is a great example of what IAA funding can help achieve. In less than two years we were able to take a theoretical idea to a product that compares favourably with established methods and is increasingly attracting the interest of experts in the DRM community as a potentially superior paradigm for predicting the impact of earthquakes on exposed population.”

Dr Hamish Patten says: “The IAA-funded project allowed me to work closely with humanitarian experts at the Internal Displacement Monitoring Centre, building relationships and understanding which will hopefully inform future research discussions and knowledge exchange. With the increase in public awareness of the impact of hazards such as earthquakes, especially hazards related to climate change, we hope our model can help ensure that future humanitarian responses are well-targeted and effective.”

The project, Estimation of Displaced Populations Post-Disaster: a Question of Data and Demography, was funded by the University of Oxford’s EPSRC Impact Acceleration Account (2020-2021)

Project team:

Dr Hamish Patten works with the Risk Information Exchange Development Team, UNDRR

David Steinsaltz is Associate Professor of Statistics in the Department of Statistics