ceglogo  unilogo

An Agent-Based Model for Flood Incident Management 

The applet requires Java 1.4.1 or higher. It will not run on Windows 95 or Mac OS 8 or 9. Mac users must have OS X 10.2.6 or higher and use a browser that supports Java 1.4. (Safari works, IE does not. Mac OS X comes with Safari. Open Safari and set it as your default web browser under Safari/Preferences/General.) On other operating systems, you may obtain the latest Java plugin from Sun's Java site.

This model is powered by NetLogo (built in Version 4.0.4) who can be contacted at: feedback@ccl.northwestern.edu.

If you can’t get the interactive demo to work, then you can view a wmv format video here. One trick to get it to work can be to clear the Java cache. In Windows enter the Java Console (from the control panel or right click on the icon near the clock), click on the ‘Settings’ button from the ‘Temporary Internet Files’ panel and the click ‘Delete Files’. You can manually delete the specific java files if you click on ‘View’ instead of ‘Settings’.


This model demonstrates a simulation approach to flood event management by coupling an agent-based model of individual behaviour with a raster cell flood model. These simulations can provide valuable evidence to flood risk managers and local authorities to support disaster management through identification of evacuation routes liable to congestion, areas most at risk of flooding, potential for loss of life, flood damages and the effectiveness of flood incident management actions.

A paper outlining the first stage of the model's development and application can be downloaded from the Journal of Natural Hazards website (for those without access, an earlier draft can be found here).


To operate the model:

(i) Choose the number of vehicles, sea level and starting time (and scenario name if appropriate) for the simulation.

(ii) Click ‘Setup’ to load in the data.

(iii) To start the simulation click ‘Step’.

(iv) To start a flood, click on ‘Destroy Defence’ this will randomly choose a flood defence section and remove it.

(v) To issue a warning and start an evacuation of individuals driving click ‘Evacuate’.

(vi) You can monitor the number of vehicles and buildings flooded and vehicles that have been diverted or isolated by floodwaters.


This web-enabled demonstration is quite basic. The flood model is based upon the LISFLOOD-FP raster cell model (Bates and de Roo, 2000) although for the web enabled version we have used an even simpler implementation for speed. The floodplain inundation model operates on a 50m grid. For the purposes of this study the river is not modelled. A recent development has be the incorporation of a full hydrodynamic simulation using a quadtree grid (Liang et al., 2007) for more accurate simulation of the floodwave.

Agents move over the transport network, and not directly on the raster grid, to preserve the topological connectivity of the road network within the floodplain.

Agents are assigned a “Home”. From here they undertake their daily grind – which may involve visiting schools and/or shops etc. throughout the day. Properties types are given in the Address Point database.

When water enters the floodplain agents avoid roads they know or turnaround and reroute to avoid roads they see to be flooded. If the depth is greater than 20cm then they are assumed to be ‘drowned’. If the depth is 0-20cm then agents travel more slowly through the water.

If the evacuation order is given, a proportion of the agents (user set) of each type receive the warning and act within a given time and head towards the evacuation shelter.


Roads that are flooded under different flood scenarios.

Roads that when flooded lead to isolation of large parts of the urban area.

Roads that become congested during an evacuation.

Different flood extents and impacts in the floodplain from the defence breach and storm surge scenarios.


This demonstration model predominantly focuses upon interactions between vehicles and the flood. Further development will include representation or consideration of:

(i) flood warning systems and lead time,

(ii) emergency responders,

(iii) use of warning signs, loud halers and public address systems,

(iv) deployment of temporary defences and engineers,

(v) location of temporary defence storage depots and evacuation shelters,

(vi) traffic management rules under evacuation conditions,

(vii) individual awareness to the meaning of flood warnings,

(viii) the role of organisations and communication mechanisms,

(ix) individual responses and actions, and,

(x) individual vulnerability and more sophisticated mortality and morbidity functions (Jonkman and Kelman, 2005; Jonkman and Vrijling, 2008).


This model was originally developed at Newcastle University by Roger Peppe and Richard Dawson. The research was funded by the Environment Agency (Reliability in Flood Incident Management Planning – SC060063) as part of the Joint DEFRA/EA Flood and Coastal Erosion Risk Management R&D Programme. The overall programme was led by Halcrow but also in collaboration with JBA Consulting, Middlesex University and Bristol University.

The model remains under development, with recent advances involving the incorporation of additional agents (e.g. blue light services) and improved modelling of the floodwave (using the Liang et al. (2007) quadtree approach)

If you are interested you can explore the code for this web enabled version here

Please contact Richard Dawson (richard.dawson@newcastle.ac.uk) for further details.


Bates, P.D. and De Roo, A.P.J. (2000), A simple raster-based model for flood inundation simulation, Journal of Hydrology, 236: 54-77.

Dawson, R. J., Peppe, R. G. and Wang, M. (2011) An agent based model for risk-based flood incident management, Natural Hazards , (doi: 10.1007/s11069-011-9745-4).

Jonkman S.N., Kelman I. (2005) An analysis of causes and circumstances of flood disaster deaths, Disasters, Vol. 29 No. 1 pp. 75-97.

Jonkman, S. N. and Vrijling, J. K. (2008), Loss of life due to floods, J. Flood Risk Management, 1: 43-56.

Liang, Q., Zang, J., Borthwick, A.G.L, Taylor, P.H. (2007) Shallow flow simulation on dynamically adaptive cut cell quadtree grids, International Journal for Numerical Methods in Fluids , 53(12): 1777-1799.