Exposing Risk: Developing Visualisation Tools for the Future
Richard Wall1, Stephen Edwards1, Kate Crowley2 and Brad Weir3
1. Aon Benfield UCL Hazard Centre, University College London, Gower St, London, WC1E 6BT, UK
2. Catholic Agency for Overseas Development, Romero House, 55 Westminster Bridge Rd, London, SE1 7JB, UK
3. Aon Benfield, 80 Raffles Place, #42-01 UOB Plaza 1, Singapore
richard.wall.09@ucl.ac.uk
1. Introduction
3. Anticipating and prioritising relief
Identifying and visualising risk is critical to improving disaster risk reduction. To address
this issue the Aon Benfield UCL Hazard Centre at University College London has
established a unique collaboration with the Catholic Agency for Overseas Development
(CAFOD) and Aon Benfield to apply risk analysis and management tools for reducing the
impacts of natural hazards. The project has used ImpactOnDemand®, an online software
platform developed by Aon Benfield, to assist humanitarian and development agencies
anticipate and respond to disaster scenarios. This pilot project has examined risk in
Cambodia, a country that is of interest to both the commercial sector and nongovernmental organisations (NGOs). The main hazard in the country is flooding, which
affects over half a million people and costs around $17 million dollars every year1.
The attributes provided within the census data have been used to create a social
vulnerability index for villages in Cambodia. This index gives a value of vulnerability for
each village that can be used to direct relief efforts. We have used participatory
vulnerability and capacity assessments (PVCAs), collected by NGOs, to select the various
attributes that the communities themselves believe define vulnerability. Each attribute has
then been ranked from low (1) to high (5) vulnerability (Table 1) and combined for each
village to give an overall value of vulnerability (Figure 4).
1http://www.preventionweb.net/
2. Datasets and methodology
Attribute
Rank
Total number of Population below Population over
females
15 years
65 years
1
2.5
Total number of
widows
Total number of
people with
disabilities
Total population with
no education (below
primary level)
3
5
3
2.5
Table 1. The attributes selected from the census data that define vulnerability. They have
then been ranked for use within the social vulnerability index.
To visualise risk, ImpactOnDemand®
requires a population dataset and a hazard
footprint. From previous visits to
Cambodia, we have made an assessment
of the available datasets to determine
which are the most appropriate for
visualising risk.
1. Population data are represented by
census data, available from the
Mekong River Commission (MRC) and
Open Development Cambodia (ODC).
Census data from 2008 are available
down to village scale and contain a
large number of attributes for each
data point (Figure 1).
Figure 1. Census data from 2008 provided
by the MRC and ODC.
Figure 4. The vulnerability of
each village in the census
dataset, determined using the
social vulnerability index.
4. Visualising risk
Having defined the hazard exposure and vulnerability of communities, risk has been
calculated using the equation Risk = Hazard x Vulnerability. Each village has been given a
value of risk, which can be plotted on maps (Figure 5A). To communicate these
visualisations to different stakeholders the values of risk have been divided into three
categories of high, medium and low (Figure 5B).
2. Flood
hazard
footprints
are
represented by the extent of flooding
events. The MRC provides estimates of
minor, medium and major events,
which are based on previous flood
events (Figure 2).
A)
B)
Figure 2. Minor, medium and major
flooding extents provided by the MRC.
The exposure of villages to flood hazard is determined by where they intersect the
different flooding extents. ImpactOnDemand® does this quickly and effectively (Figure 3),
with the results portrayed on maps or exported as a spreadsheet. The results can be
manipulated to show exposure for attributes above or below different thresholds. To
make this technique available to a wide range of users we have adapted the methodology
for use in both ArcGIS and freely available QuantumGIS.
Figure 5. A) The risk of individual villages to flooding in Cambodia, where risk is based on
recurrence time of the flooding events in figure 2. B) A simplified risk map, where risk is
portrayed in three categories.
5. Conclusions
Figure 3. A screenshot from ImpactOnDemand ® showing how it defines exposure
using both the population and flooding extent datasets.
The project has shown that the simple methodology developed can be used to create
accurate and effective visualisations, which can be modified to rapidly and effectively
present risk to different stakeholders. We are producing a set of guidelines to enable these
stakeholders to use our methodology with different software packages and with minimal
training. From an assessment of datasets available in Cambodia we are generating
recommendations for the systematic collection of data required to adequately capture and
portray risk.
For further information please visit https://www.ucl.ac.uk/abuhc/research/Risk_Visualisation/