Centre for Earth Systems Engineering Research

Flood Incident Management

Flood Incident Management

Agent-based modelling for flood incident management.

Project Leads

Introduction

Recent flood events in the UK have highlighted the need for improved flood incident management (FIM). 

The aims are to:

  1. Obtain a greater depth of understanding of the dynamics of FIM process and systems.
  2. Enable implementation of a risk-based approach to FIM.

To do this we have coupled an agent based model with a hydrodynamic model. We've demonstrated its capability in Towyn in North Wales. 

Original development of the model was funded by the Modelling and Risk them of the DEFRA/Environment Agency joint R&D programme (SC060063). Miao Wang’s studentship is funded through an EPSRC Doctoral Training Account.

For more information, visit the project website

A publication based on this work has been published in the journal Natural Hazards:
Dawson, R. J., Peppe, R. and Wang, M. (2011) An agent based model for risk-based flood incident management, Natural Hazards, 59(1):167-189.

An agent based model of flood evacuation

Agent-based model of flood incident management

The model has been coded in NetLogo, a freeware agent based modelling platform.

Key features

Information 

The model uses nationally available information. This includes OS MasterMap and AddressPoint. It identifies roads and buildings, digital elevation, flood defence properties from the National Flood and Coastal Defence Database.

Model

A network model is constructed from MasterMap data. Agents use it when moving through the network. the model:

  • describes the road system
  • preserves topological connectivity of the road network
  • automates identification of traffic rules (eg speed limit for different road types, travel direction etc.)

A raster-based hydrodynamic model has been coded in the NetLogo environment.

Agents

Agents have been categorised by behaviour patterns (eg single mothers, retired, young professional. This captures the natural variation of location and traffic activity throughout a typical day. Generic daily movement routines have been identified and described in probabilistic terms. This is for each group from National Travel Survey data and Census.

Agents go about their daily business according to the generic routines unless they successfully receive and respond to an evacuation warning. In this case they either:

  • head to the evacuation shelter
  • come across some flood water forcing them to re-route their journey
  • get ‘drowned’

Results

The model can explore the impact, in terms of number of people exposed to floodwater. This can be done according to the time of day, different storm surges and defence failures. 

The benefits of issuing flood warnings can be tested. The benefits of different evacuation shelters for saving lives and managing traffic congestion can be explored.

Ongoing developments include:

  • representation of additional agents (eg blue light services and engineers)
  • improved individual vulnerability functions
  • additional flood management measures (eg individual responses, deployment of temporary defences)

The expected number of agents exposed to dangerous flood flows (or risk to people) was calculated using the method of Dawson and Hall (2005). This was originally developed to evaluate economic risks. 

The results showed the risk to people was four times larger where flood warning was not provided.  

Interestingly, the flood defences that contributed most towards risk to people were not the most important for economic risk.

References

Dawson, R.J. and Hall, J.W. (2006) Adaptive importance sampling for risk analysis of complex infrastructure systems, Proc. R. Soc. A, 462(2075): 3343–3362.