A step towards risk-based flood forecasting - A pilot application in Dumfries, Scotland

Stuparu, D. and Bachmann, D. and Ververs, M. and Brown, E. and Boelee, L. and Tavendale, A. (2017) A step towards risk-based flood forecasting - A pilot application in Dumfries, Scotland. In: ICFM7, 5-7 September 2017, Leeds, UK.

Full text not available from this repository.


The Scottish Environment Protection Agency (SEPA), Deltares and HR Wallingford have recently collaborated to develop new tools and techniques to extend existing flood forecasting capabilities. The focus was on developing timely ‘impact’ information to support flood response activities. Within this research, the River Nith to Dumfries was used as a pilot catchment. SEPA operates a Delft-FEWS based flood forecasting and warning system which consists of multiple models that use numerical weather predictions to estimate future river levels and flows. Over time, SEPA’s flood forecasting system has continually evolved to incorporate the latest science and technology. This study focussed on the following aspects: 1. Live flood maps: forecasting flood inundation; 2. Live impact information: number of properties affected and damages; 3. Ensemble runs: incorporating meteorological uncertainty. These elements were connected to the existing forecasting system in a framework titled FEWS-Risk. Flood inundation mapping was performed using HR Wallingford’s Rapid Flood Spreading Model (RFSM-EDA). The RFSM-EDA is a 2D inundation model that is used for rapid flood inundation prediction in practical run times, appropriate for operational use. The RFSM-EDA was setup using Lidar digital elevation data on a 5 metre grid. For the Dumfries catchment the RFSM-EDA embedded in FEWS-Risk framework takes 5 minutes to run a 2 day simulation. Part of the FEWS-Risk framework, the potential damages were assessed by linking the flood depths with receptor impact information using Delft-FIAT (Flood Impact Assessment Tool). The possible benefits of including uncertainty in the forecasts using ensemble forecasting were also explored. To achieve this, 5 ensemble members were randomly generated based on Storm Frank (December 2015) data and used as boundary conditions for the RFSM-EDA model. The simulations were used to calculate the probability of exceedance of a given water level threshold. The results were presented and discussed at a workshop hosted at the Scottish National Centre for Resilience, with delegates attending from SEPA, Scottish Fire and Rescue Service, Police Scotland, Scottish Government and Dumfries & Galloway Council. Feedback indicated that the strength of such risk-based operational modelling is greatest for events and regions where impacts are unknown. Communication of risk is a challenge, but guidance with long lead times and high uncertainty can still be useful (e.g. to trigger preparation). Information regarding the onset of floods is especially useful for flood preparedness. Overall, probabilistic forecasting can give good guidance in advance of a flood event but needs to link closely with operational procedures to maximise its effectiveness. Results from this pilot study indicate that the FEWS-Risk framework may have great potential in helping to minimize flood impacts.

Item Type: Conference or Workshop Item (Paper)
Additional Information: 7th International Conference on Flood Management (ICFM7)
Uncontrolled Keywords: flood forecasting; impact; ensemble; probabilistic
Subjects: Floods > General
Divisions: Floods
Depositing User: Unnamed user with email i.services@hrwallingford.com
Date Deposited: 02 Apr 2020 09:53
Last Modified: 29 May 2020 07:40
URI: http://eprints.hrwallingford.com/id/eprint/1261

Actions (for site administrators only - login required)

View Item View Item