Using historical data to improve long-return period flooding estimation

Institution: Centre for Ecology & Hydrology, Wallingford

Lead Supervisor: Lisa Stewart

Project Description

Estimating flood magnitudes for long return periods, such as the “once-in-a-century” flood is of critical importance to building appropriate flood defences which will withstand the most extreme events, events which may not have been observed within the period of systematic river catchment recordings. Existing methods used in the Flood Estimation Handbook (1999) and subsequent reports extrapolate from existing data to estimate less frequent, more extreme flood events than those observed. By obtaining and including even a small amount of historical data on observed extreme flood events, estimates can be improved and uncertainty decreased. The project will involve the collection of historical data from a number of sources from both online and physical archives (council records, newspaper archives, anecdotal sources), and then applying existing statistical methods to improve the existing models to observe more accurate flood estimation for the catchments in question. If time permits, this will be supplemented with the use of Approximate Bayesian Computation methods to see if simple simulation methods can improve on existing Likelihood based methods.

The student will receive supervision from Dr Adam Griffin on the statistical and computer-based background and Lisa Stewart on the hydrological aspects within the Hydrological Modelling and Risks group. Training will be provided in data management, statistical methodologies regarding analysis and visualisation of censored datasets of unknown quality, and basic hydrological processes. Supervision will be given at the sites finding data from historical archives off-site, and office and computer space will be arranged for the student within the department.

This work follows on from the report “Making better use of local data in flood frequency estimation” published by the Environment Agency, and will hopefully build towards a larger body of historical data for estimating such long return period floods. There are no known intellectual property rights concerns arising from this work.

Indicative timescale for project (subject to change)

• Week 1: Hydrological, statistical and programming background
• Weeks 2: data acquisition on site through internet-based archives
• Weeks 3-4: data acquisition at archive sites
• Week 5: initial data analysis
• Week 6-7: work with group to incorporate into existing framework
• Week 8: preparation of report to summarise findings.

Powered by WordPress. Designed by WooThemes