A data-driven exploratory study of extreme events based on joint probability analysis
Lead Research Organisation:
NERC CEH (Up to 30.11.2019)
Department Name: Boorman
Abstract
The aim of this project is to examine records of past moderate and extreme flood events, and of associated environmental variables related to the causes of flooding, and to undertake an exploratory data analysis of these. This would avoid relying immediately on the assumptions built into existing methodology and provide an independent check on these. It would also allow the construction of a statistical methodology tailored both to the observed properties of the datasets and to the estimation of the relevant properties of extreme events that need to be extracted from the data. Joint probability analysis would be one of the main statistical approaches being used. As well as providing useful insight into the occurrence of flooding, this has the potential to lead to more statistically efficient estimation of floods. Further insights into flooding problems will be sought by directly considering the seasonality of flood events in all the analyses. The datasets available can realistically be expected to provide good estimates of floods with return periods of 10-20 years, but the statistical models used can be employed to extrapolate to return periods of 50, 100 or even 1000 years. For such extrapolation the uncertainties inherent in this estimation are likely to be large and an important aim of the project will be to provide a useful assessment of this uncertainty. Hourly datasets already held at CEH will be used for the project.
Organisations
Publications
Kjeldsen T R
(2010)
A joint probability approach to flood frequency estimation using Monte Carlo simulation
in Proc. BHS Third International Symposium Managing Consequences of a Changing Global Environment
Svensson C
(2013)
Flood frequency estimation using a joint probability approach within a Monte Carlo framework
in Hydrological Sciences Journal
Description | Event-based flood estimation methods in the UK have hitherto relied on expert judgement in assigning pre-determined values of the input variables to a rainfall-runoff model. The output flood is expected to be of a particular rarity, but this may be biased by the initial choices. In contrast, this work presents a novel joint probability approach that combines values from the full range of each input variable's distribution. For the first time, dependencies from one event to the next, and between different variables within a single event, are accounted for. Uncertainties in flood estimates are reduced compared with traditional statistical methods. |
Exploitation Route | The methodology can be generalised to ungauged basins, provided that relationships between catchment descriptors and model parameters can be found. This means that the methodology could be applied anywhere, regardless of whether or not river flow observations are available for the target site. |
Sectors | Construction,Energy,Environment |
Description | The findings from the project have contributed to the scientific discussion on joint probability methods for flood frequency estimation. In the longer term they may lead to changes in design guidelines. |
First Year Of Impact | 2013 |
Sector | Environment |
Impact Types | Societal |