|GEO-SAFE - Geospatial
based Environment for Optimisation Systems Addressing Fire Emergencies
May 2016 – April 2020
FSEG University of Greenwich.
Funded by: Marie
Skłodowska-Curie Research and Innovation Staff Exchange (RISE)
Project Budget Contribution from the EU:
In the EU and Australia, every year
thousands of square miles of forests and other regions burn due to
wildfires. These fires result in significant economic and ecological
losses, and often, human casualties. Both EU and Australian
governments are aware of how crucial it is to improve the management
and containment of wildfires. Scientists from different specialties,
both in EU and Australia, have already developed methods and models
in order to improve the management and decision process pertaining
to preparedness and response phases in case of bushfire.
Geospatial based Environment for Optimisation Systems Addressing
Fire Emergencies (GEO-SAFE) project, co-ordinated by FSEG of the
University of Greenwich, is a four year project (running from May
2016 to April 2020) with 17 partners from 7 countries (United
Kingdom, Spain, Italy, France, Switzerland, Netherlands and
Australia) with an estimated project cost of €1,386,000.00. The aim
of GEO-SAFE is to create a network enabling the two regions to
exchange knowledge, ideas and experience, thus boosting the progress
of wildfires knowledge and the related development of innovative
methods for dealing efficiently with such fires. Major outcomes of
this project are implementing solutions and tools in fire
suppression, life and goods protection, and implementation and
training. The three main fire management domains addressed in
GEO-SAFE are: ‘Fire Suppression and Fire propagation control’, ‘Life
and Property Protection’ and ‘Implementation and Training’. FSEG’s
main role in this project is in the domain of Life protection. The
three main breakthroughs to be achieved in this domain are:
Computing efficiently good
solutions for micro evacuation simulation software is vital for
applying them to large open spaces such as found in wild fire
Identification and quantification of the
human behaviour associated with wildfire evacuation scenarios.
Developing specific models dedicated to
wildfire constraints and merging both agent-based simulation and
OR approaches is a challenging direction of research for
handling large-scale evacuation.
|FSEG final reports
The GEO-SAFE project was completed in April 2020. The three reports
produced by FSEG are provided below:
Analysis report on specific requirements of large-scale evacuations in
case of wildfires
Authors: Gallego, D.M.,
Veeraswamy, A., and Galea, E.R.
deliverable provides an overview of the process of identifying the key
requirements for evacuation modelling tools as perceived by the
authorities involved in planning and real-time management of urban-scale
evacuation resulting from wildfires. The process involves identifying
organisations (13 in total) involved in the management of wildfire
evacuation in six countries (Australia, Ireland, Italy, Netherlands,
Spain, and the UK) and then approaching key staff in those organisations
(18 in total) that are involved in incident management with specially
designed interview questionnaires and on-line surveys. The methodology
used to prepare the interview questionnaires and surveys is described
followed by a description of the interview protocols and each section of
the questionnaire. This is followed by a description of the secondments
that were carried out to collect the data, which forms part of Geo-Safe
Task 2.4. The results are then presented with a detailed analysis of the
data collected from the interviews and on-line surveys. Finally, the
analysis of the interview/survey responses are distilled down into 22
key factors within 10 broad categories that identify the perceived needs
and desires of the emergency management end-user community in relation
to the use of urban-scale evacuation models for planning, real-time
applications and community training.
here to view the report.
Please cite this report as follows:
Gallego, D.M., Veeraswamy,
A., and Galea, E.R., Analysis report on specific requirements of
large-scale evacuations in case of wildfires, GEO-SAFE H2020 project
691161, final D2.4 report, 15 April 2019,
Report on wildfire large-scale evacuations - behavioural responses
Authors: Hulse, L.M., Veeraswamy, A, Vaiciulyte, S.,
and Galea, E.R.
Summary: This deliverable
provides an overview of the GEO-SAFE project’s Human Behaviour Study.
Survey data were collected from residents in the South of France,
Australia, and Central Italy, with the aim of identifying, quantifying,
and calibrating behaviours occurring during wildfire evacuations. The
evidence base generated by this study provides an understanding of
various behaviours including evacuation decisions, response actions, and
associated times. This data can be used to assist the development of an
agent-based evacuation model, urbanEXODUS. In turn, this tool can be
used by professionals involved in wildfire management to aid learning,
decision-making, and planning.
here to view the report.
Please cite this report as follows:
Veeraswamy, A, Vaiciulyte, S., and Galea, E.R., Report on wildfire
large-scale evacuations - behavioural responses, GEO-SAFE H2020 project
691161, final D2.5 report, 30 April 2020,
Title: Global wildfire large-scale evacuation model
Authors: Veeraswamy, A, Galea, E.R., Lawrence, P.J., Gallego,
D.M., and Blackshields, D.
deliverable describes the large-scale evacuation modelling work
performed as part of the GEO-SAFE project. The task addressed in this
deliverable is Task 2.6 in Work Package 2 and Work Group 6, Phase 3. A
key novelty of this work is the integration of the three models that can
be used to manage large scale evacuations due to wildfires and other
incidents that require evacuations. The pedestrian evacuation model,
urbanEXODUS, the vehicle evacuation model, SUMO and wildfire models such
as PHOENIX have been integrated as part of Task 2.6. The accuracy of
evacuation model predictions can be enhanced by accurate representation
of the walking speeds of people over long distances and on different
terrain types and slopes. Thus, walking speed data collected from four
different trials are being analysed statistically so the empirical data
can be included in the pedestrian evacuation model. The integrated
simulation environment was applied to two demonstration cases by
modelling hypothetical evacuation scenarios in two urban environments,
one in Spain and one in Australia. Finally, the import of Digital
Elevation Model (DEM) data into the urbanEXODUS evacuation model was
demonstrated enabling the incorporation of terrain slope into the
evacuation model. As pedestrian walking speeds are impacted by the slope
of the paths, the impact of terrain slope on evacuation was demonstrated
using a hypothetical evacuation in the Greenwich (UK) area.
here to view the report.
Please cite this report as follows:
Veeraswamy, A, Galea,
E.R., Lawrence, P.J., Gallego, D.M., and Blackshields, D., Global
wildfire large-scale evacuation model, GEO-SAFE H2020 project 691161,
final D2.6 report, 29 April 2020,
The project brings together the following research organisations and
practical wild fire response organisations undertaking research in the
following 7 workpackages:
|| M1 - M48
||CEREN, UTWENTE, PCF
||M1 - M47
||CEREN, CNRS/LAAS, UCM, PCF, UBP/LIMOS, UTC,
UoG-FSEG, RMIT, SDIS2B
|WP3-Robustness and Efficiency
||M1 - M47
||UPD, UBP/LIMOS, CEREN,UNIVAQ, UNIVPR, PCF, UTC
|WP4 - Implementation and Training Tools
||M1 - M47
||UoG-FSEG, UCM, PCF, EMN, UNIVAQ, UNIVPR, UoG-eCentre, CEREN
|WP5 - Innovative Solutions Think-Tank
||M1 - M48
|WP6 - Dissemination and Awareness
||M1 - M48
|WP7 - Project Management
||M1 - M48
The research methodology will be
based on the joint work of 7 multi-disciplinary research Work Groups
each of them meant to bring innovative solutions to wildfire management
|WG1 - Stochastic mapping workgroup
||UTWENTE and UOM
|WG2 - Global and alternative models
||UCM and RMIT
CNRS/LAAS, PCF, UBP/LIMOS, UTC
|WG3 - Robust processes and dynamic
||UPD, UBP/LIMOS, CEREN, UNIVPR, UPD, UBP/LIMOS
|WG4 - Efficient Responses
||GDR and RMIT
||UNIVAQ, UNIVPR, PCF,
UPD, UBP/LIMOS, UTC
|WG5 - Design of the implementation and
||EMN and RMIT
||UoG-FSEG, UCM, PCF, EMN,
UNIVAQ, UNIVPR, UoG-eCentre
|WG6 - Agent Based Evacuation Models
||UoG-FSEG, PCF, RMIT, SDIS 2B
|WG7 - Training tools
||UoG-eCentre, UoG-FSEG, PCF, CEREN
|FSEG’s Role in the Project
| Current urban scale evacuation models do not generally include
pedestrian evacuation, focusing more on vehicular evacuation or coarsely
model pedestrians utilising only vehicular routes. Disaster managers
require a tool that is capable of simulating pedestrian evacuation of an
on-going and evolving event. FSEG’s objectives in the GEO-SAFE project
are listed below:
Objective 1.0: Identification of Human Behaviour associated with wildfire evacuation:
Objective 2.0: Develop a Large Scale Wildfire Evacuation Model
- Objective 1.1: Develop a database of human response to wildfire, based on questionnaire studies involving those who have experienced evacuation from wildfires
and those who may need to evacuate from a wildfire situation.
- Objective 1.2: Identify, quantify and calibrate appropriate behaviours identified in the questionnaire study.
Objective 3.0: Embed evacuation simulation environment into a training environment
- Objective 2.1: Identify model requirements through interviews with fire authorities and disaster management organisations.
- Objective 2.2: Develop and implement these end user requirements within the urbanEXODUS evacuation simulation tool.
- Objective 2.3: Develop and implement behavioural responses unique to wildfires within the evacuation simulation tool based on data collected through the Human Behaviour study.
- Objective 2.4: Represent the effect of terrain (paved/unpaved paths) and gradients on the travel speed and behaviour of agents.
- Objective 2.5: Integrate the evacuation model with fire spread models.
- Objective 2.6: Represent the interaction of vehicles with pedestrians within the urban evacuation tool.
- Objective 2.7: Develop a large-scale wildfire evacuation model that can run at faster than real time and thus be utilised during an emergency evacuation incident.
- Objective 2.8: Calibrate the model to provide the right balance between speed and accuracy of the model to provide reliable results in a timely manner.
- Objective 3.1: Embed urbanEXODUS within the Pandora+ training environment to enable crisis managers to make informed evacuation decisions during the training sessions.
- Objective 3.2: Integrate the evacuation simulation results within the training environment so crisis managers can formulate appropriate evacuation procedures.
- Objective 3.3: Enable crisis managers to generate a new scenario on the fly and utilise the results provided at faster than real time to make informed decisions on the most appropriate evacuation
procedures to apply.
The work of FSEG will be undertaken in three of the seven work
Work Group 5: Design of the implementation and
This work group involves benchmarking
real/realistic instances (either instances inspired by a past event or
simulated ones), testing different processes on these instances and
comparing these solutions from different perspectives that include the
quality of the solution, the computational time and the flexibility of
the system. For such comparison, a key point is the choice of
standardized technologies making the different systems compatible and
FSEG’s role in this work group is to link the
urbanEXODUS simulation environment to the PHONEIX fire simulation
environment (Phase1), demonstrate the link (Phase2) and demonstrate the
integrated urbanEXODUS-Pandora+ prototype at
Service Departmental d’Incendie et de
Secours HAUTE CORSE, France (Phase 4).
Work Group 6: Agent Based Evacuation Models
- Phase 1:
Risk assessment solutions implementation:
(Objective 2.5 )
WG5 will strongly interact with WG1 to
develop and interface of the risk cartography with fire spread simulator
(in particular with PHONEIX used in Australia – Victoria) and integrate
the risk cartography directly in decision models concerning resource
allocations in initial attack and evacuation.
- Phase 2:
Fire fighting solutions:
Here WG5 will strongly interact with WG2 and WG6 to implement a
prototype for controlled burning solutions for fuel management with
Victoria AUS data and of a coherent procedure of evacuation in
- Phase 4: Simulation environment:
Last direction will be developed in
strong link with WG7. It is possible to embed the multi-agent simulation
environment into a training environment for first responders and
incident managers. Here a prototype for evacuation will be tested at
FSEG is the manager of this work group. This work group will explore the
application of large-scale multi-agent simulation techniques to
evacuation scenarios associated with wildfire evacuation situations. The
result will be large-scale evacuation tool that is capable of simulating
several interactions of different agents. This will build upon the
existing state-of-the-art evacuation simulation software, urbanEXODUS.
It is proposed that the EXODUS model can be employed to examine the
impact of the wildfire scenario (e.g. the nature, size, location of the
fire, fire spread rate, the population distribution and characteristics,
their expected response and the routes available) upon the time for the
target population to reach safety. This will involve a number of
research areas including:
- Phase 1: Explore the
requirements of large-scale evacuations resulting from wildfires
This will make use of interviews with fire
authorities and disaster management organisations that have experience
of managing large scale evacuations associated with wildfires.
- Phase 2: Develop and implement behavioural responses unique to
wildfires within the modelling environment (Objectives 1.1,
This will require the development of additional behavioural capabilities
and data sets to quantify and calibrate the behaviours.
Phase 3: Explore the use of alternative techniques in large-scale
evacuation scenario’s. (Objectives 2.2,
This will involve investigating the representation of terrain
effects, such as change in grade, in the different modelling
environments in order to capture the appropriate behaviour of the
agents. Large-scale evacuation scenarios will require an ability to
simulate the interaction of pedestrians with vehicles and must be able
to be executed in reasonable time.
Work Group 7:
This work group will be focussed around
research and innovation activities in the training of planners and
operational managers for crisis situations and in particular wildfire
situations. The lead partner in this work group, University of Greenwich
(Pandora team), will bring a novel and innovative augmented
reality-training environment developed from an EU FP7 project,
Pandora+, as background to
provide an appropriate technology for the management and delivery of a
developed training package encompassing the research outputs from the
FSEG’s role in this work group will involve
the integration of the training tools with the evacuation simulation
tool, urbanEXODUS (Phase 1).
- Phase 1: Adapting Pandora+ to other environments:
The potential exists to implement the Pandora+
environment training within the lifetime of the project, building in as
many of the outputs from the other working groups as possible, dependent
on the stability and robustness of those outcomes, within a framework of
the existing tools and expertise provided by the Australian partners.
The work group will also support similar activities with the end-user
European partners, and implementation of the developed training package
within the Pandora+ environment.
|Evacuation during wild fires must take
into consideration pets
|Evacuation by road during the Fort
McMurray wild fire
|Taken from a dashboard camera of an
evacuee evacuating Fort McMurray and heading south on 3rd May
|An area in Bracknell which was affected
by a real fire on May 02, 2011 simulated in urbanEXODUS. The
fire extent shown here is the simulated fire having a greater
extent than the actual fire.
|A snapshot of the Bracknell area
simulation at 30 minutes showing the progress of evacuation
while the simulated fire has just started.
|A snapshot of the simulation at 1 hour
where all people have evacuated and the fire is fast spreading
towards the evacuated area.
||Prof. Ed Galea
Fire Safety Engineering Group
University of Greenwich
Greenwich Maritime Campus
Old Royal Naval College
Queen Mary Building
Greenwich SE10 9LS
Tel: +44 (020) 8331 8730
|Consortium web page