Content uploaded by Krishna Nadimpalli
Author content
All content in this area was uploaded by Krishna Nadimpalli on Oct 13, 2015
Content may be subject to copyright.
National Exposure Information System (Nexis) For
Australia: Risk Assessment Opportunities
Nadimpalli, K., M. Edwards and D. Mullaly
Risk Research Group, Geoscience Australia, GPO Box -378, Canberra, ACT 2601
Email: krishna.nadimpalli@ga.gov.au
Keywords: Exposure, Risk Assessment, NEXIS, Natural Hazards
EXTENDED ABSTRACT
In August 2002 the Council of Australian
Governments (COAG) reviewed natural disaster
relief and mitigation arrangements for Australia
(COAG, 2003). In response to the
recommendation to “develop and implement a
five-year national program of systematic and
rigorous disaster risk assessments”, Geoscience
Australia (GA) is undertaking a series of national
risk assessments for a range of natural hazards.
Fundamental to any risk assessment is an
understanding of the exposure including the
number and type of buildings, businesses,
infrastructure and people exposed to the hazard
of interest. Presently there is no nationally
consistent exposure database in existence for risk
assessment purposes. It is important to
emphasise that understanding the risks
associated with various hazards requires more
detailed information than the population and
number of structures at a census district level.
The understanding of building type, construction
(roof and wall) type, building age, number of
storeys, business type and replacement value is
critical to understanding the potential impact on
Australian communities from various hazards.
The National Exposure Information System
(NEXIS) is aimed at providing nationally
consistent and best available exposure
information at the building level. It requires
detailed spatial analysis and integration of
available demographic, structural and statistical
data. Fundamentally, this system is developed
from several national spatial datasets as a generic
approach with several assumptions made to
derive meaningful information. NEXIS
underpins scenarios and risk assessments for
various hazards. Included are earthquakes,
cyclones, severe synoptic wind, tsunami, flood
and technogenic critical infrastructure failure. It
will be integrated with early warning and alert
systems to provide real time assessment of
damage or forecast the impact for any plausible
hazards. This system is intended to provide a
relative assessment of exposure from multiple
hazards and provide the geographic distribution
of exposure for regional planning. This will be at
an aggregated census district level now and at a
mesh block level in the future.
The system is scoped to capture the residential,
business (commercial and industrial), and
ancillary (educational, government, community,
religious, etc.) infrastructure. Currently the
NEXIS architecture is finalised and the system
provides residential exposure information. The
prototype for business exposure is in progress.
The system aims to capture ancillary buildings,
infrastructure and various critical infrastructure
sector exposures in future. More specific
building and socio-economic information will be
incorporated as new datasets or sources of
information become available.
The NEXIS will be able to provide the exposure
information for the impact analysis for a region.
This database will not support a site specific
assessment involving one or two buildings and
need more specific information about the
particular exposure to estimate the risk at micro
level. More detailed information suitable for
such analysis will be maintained in reference
databases.
1674
1. INTRODUCTION
Geoscience Australia (GA) began the
development of the National Exposure
Information System (NEXIS) in response to
COAG reform Commitment 2 – “establish a
nationally consistent system of data collection,
research and analysis to ensure a sound
knowledge base on natural disasters and disaster
mitigation” (COAG, 2003). It was also
recognised as a priority for the development of
better models and tools to allocate investment
across prevention, preparedness, response and
recovery (PPRR) and also to assess the impact of
emergencies on the community in the Emergency
Management Information Development Plan
(Harper, 2006). The NEXIS underpins various
activities of risk assessment modelling, critical
infrastructure failures, early warning systems and
several national priority initiatives. This system
will provide consistent and best available
information at a national scale (for example, the
number and type of buildings, businesses,
people, critical infrastructure, and institutions
such as schools and hospitals) to understand
hazard exposure, at all locations in Australia.
Fundamental to any risk assessment is an
understanding of the exposure including number
and type of buildings, infrastructure and people
exposed to the hazard of interest. There is no
such database or information in existence at a
national level to assess the risk consistently
across the nation. The risk assessment outputs
previously created for the Disaster Mitigation
Australia Package (DMAP) of the Department of
Transportation and Regional Services (DoTARs)
have provided exposure information at the level
of a census district. This information is critical as
the Office of Small Business is keen to
understand the economic impacts of natural
hazards for small businesses. It underpins
scenario (event) based risk modelling which
provides plausible scenarios for tactical planning
purposes like emergency training, planning,
response and capacity review. This information
is highly significant for the development of
probabilistic risk models which predict the
relative chance of different scenarios occurring
at a particular location. Probabilistic risk
definition is particularly helpful to prioritise risk
mitigation and preparedness.
The Earthquake Risk Assessment Model
(EQRM), wind risk assessment model, tsunami
inundation and risk assessment model and
Critical Infrastructure Protection Modelling and
Analysis (CIPMA) at GA are using NEXIS to
estimate the economic impacts of hazards.
2. METHODOLOGY
The purpose of NEXIS is to provide the best and
most available current information to multi-
hazard national risk assessment projects. The
NEXIS’ architecture comprises several
components and is depicted in Figure 1.
NEXIS-RefDb: A database will be designed to
accommodate all the external fundamental
reference datasets like Geo-coded National Address
Framework (G-NAF), Cadastre, Census, cost factors,
housing surveys, Australian business registry, etc.
These databases sourced from various
organisations like Public Sector Mapping
Agency (PSMA), Australian Bureau of Statistics
(ABS), Cityscope, local governments and
infrastructure sectors. It will be maintained in a
centralised corporate database environment with
regular updates.
NEXIS-App: Application software will be
developed to derive the required information and
will consist of several modules e.g. residential,
business, ancillary and infrastructure exposure. It
selects the best available datasets such as more
specific information from survey or local
councils and their conversion into NEXIS data
templates at building level.
NEXIS-Database: The derived information will
be stored and maintained in this database. The
database will be replicated for the Critical
Infrastructure Project.
NEXIS-Web: A web application will be
developed to aggregate the exposure information
at Local Government Area/Census District level
for the internet. This tool will be used to supply
exposure information to external agencies such
as the Australian Greenhouse Office (AGO),
State and Federal Emergency Services, and the
Bureau of Meteorology.
NEXIS-Act: This application will be developed
to provide information for various hazards like
earthquake, wind, tsunami etc. These activities
need slightly different information from the
exposure database.
Additionally, an error log report of data from
Quality Control (QC) activities will be provided
to Internal and External database custodians. The
report summarises the limitations of source data
1675
for internal users will summarise the limitations
to use heuristic judgements based NEXIS. This
process will help the external data providers to
rectify the errors in their datasets.
The development of nationally consistent
exposure information requires detailed spatial
analysis and integration of demographic,
structural and statistical data. Fundamental to
this process are several national and local
resolution datasets listed in Table 1.
Table 1. List of fundamental datasets referred to
derive the buildings information.
Accurate location of the buildings and the
information associated with them is critical for
risk assessments. The NEXIS will be maintained
and updated regularly as and when the
fundamental datasets like ABS census and G-
NAF are revised. It will also be revised to
enhance accuracy of attribute values as more
data becomes available about building age and
structure types.
Available fundamental datasets and some
reasonable assumptions are used to populate the
building information. A computerised script was
developed using Python and ArcGIS version 9.1
to automate the process and develop the
database. The process of deriving the
information is computationally intensive and
difficult to provide quickly. Therefore, the
NEXIS database for the entire nation is produced
and maintained to clip an area of interest. The
variables of the building information vary
significantly based on the usage (residential,
business (commercial and industrial), ancillary
and infrastructure). The development of this
generic information system is based on several
assumptions and approximations where specific
information is not available.
Figure 1. System Architecture for the NEXIS
2.1. NEXIS – Residential Buildings
The residential component is the most significant
for the account of the socio-economic impacts on
communities and emergency management. The
variables required to assess the risk for the
residential areas are primarily spatial, structural
and demographic.
The spatial location is highly critical for
assessing the risk from cyclones, tsunami,
flooding and bushfires. National datasets like
ABS Census provides demographic information
aggregated at Census District (CD) level. But the
risk from natural hazards is not uniformly
distributed across the CD or local government
Areas (LGA). In absence of comprehensive
residential exposure information, some generic
assumptions were made to derive the information
required for risk assessments. The required
variables and associated fundamental datasets
were summarised in Table 2.
G-NAF is a national address database with
latitude and longitude. It covers the entire nation
and is consistent across all locations. There are a
Dataset Resolu-
tion Data
Type Source
G-NAF Point PSMA
Cadastre Polygon PSMA/MapInfo
StreetPro Points PSMA/MapInfo
Census Census
District Polygon ABS
Census
Districts /
Mesh blocks
Polygon ABS
Housing
Survey States &
Capital
Cities
ABS
Business
Registry,
Others
Postcode Polygon ABS
Cityscope City
Buildings Point Cityscope
Cost Factors Cordell/Rawlinson
Cost Factors
1676
few known problems with this fundamental
dataset but it was considered as the best available
source to derive building location. The primary
assumption to derive the NEXIS is that every
residential address is linked to a residence and
every residence is within a building. Then further
assumptions are made to remove the buildings
where a building is not expected. For example a
residential building is unlikely to exist in a block
size less than 43 sq. metres or at an address in a
cadastre parcel of reserve or parkland.
Table 2. Summarised residential variables and
associated fundamental datasets.
The ABS has released a beta version of Mesh
Blocks with the final version to be released in
the future. ABS mesh blocks are another national
dataset which provide land usage categories such
as residential, rural, commercial and industrial,
education, hospital/medical, transport, parkland,
agricultural and water. NEXIS will capture the
addresses in residential, agriculture and rural
areas as residential addresses with some
omissions. The locations of these addresses are
considered as residential homes and associated
with a building.
The number of addresses having the same
latitude and longitude were counted to estimate
the number of residences in the building. The
cadastre database for each state is available and
is regularly updated. This spatial layer is
overlayed with residential buildings point layer
to get the block size for each building. Floor area
was estimated based on the location of the
building (city area, suburban or rural) and the
size of the block. The relationship of the block
size and floor area was established by searching
from a sample set of buildings across the nation.
The floor area for an apartment in a multi-storey
building is considered as 120 sq. metres.
Multiplying by the number of residences in each
building gives an estimate of the total floor area
of the building.
The type of the building and the number of
storeys were determined based on the number of
residences and the size of the cadastre parcel.
The building type attributes are aligned with the
ABS standard classification for dwelling types
Each address in the G-NAF database is
considered to be a residence. The number of
residences at one geographic location is counted
and matched with the block size. An extra rule
was also created to fine tune the building type
categorization for the buildings. If more than one
residence exists in the one cadastre parcel but not
at the same location, then those buildings are
considered as townhouses and categorised as
semi-detached buildings.
Most natural hazard risk assessments estimate
the damage to structures based on their wall and
roof type. GA’s risk models predominantly use a
structure type classification that has been
extended from the HAZUS classification system
(Robinson et al., 2005). This classification is
based upon the Australian Housing Survey
which provides information on the proportion of
structural types for each state and territory of
Australia (Australian Bureau of Statistics, 1999).
These proportions are provided separately for
each state or territory’s capital city and the rest
of the state/territory. The Australian Housing
Survey provides proportions of structural types
for separate houses, semi-detached structures and
apartments (Australian Bureau of Statistics,
1999). To date a single set of proportions for
each of these larger building types has been used
throughout Australia, however it is envisaged
that future work will refine these proportions to
account for variability across the country. The
random assignments of roof and wall types for
buildings are estimated.
Variables Datasets Derived Information
Latitude
Longitude
Address
GNAF Geo-coded Addresses
Block Size
Floor Area Cadastre Floor Area (sq. m) was
derived based on the
location and size of the
block
Usage ABS
Meshblocks Residential, agricultural,
commercial, industrial etc.
Building
Type Cadastre,
GNAF Separate house, semi-
detached house and
apartment
Roof Type
ABS Housing
survey Tiled, metal, fibro
Wall Type ABS Housing
survey Double brick, brick veneer,
fibro, timber, concrete
frame
Age Building
approvals Period of construction
Number of
Storeys GNAF,
Cadastre Based on number of
residences and size of the
block
Population ABS Census Census district average
population based on
building type.
Household
Income ABS Census Census district average
Building
Value Cost Factors Replacement value
Contents
Value Insurance
estimates Based on household
income and building value
1677
The population for each residential building
within the NEXIS database are populated from
the average population for different types of
dwellings extracted from the ABS Census at CD
level. The number of residences was multiplied
by the average apartment occupant numbers to
get the population of multi-storey buildings.
The household income underpins the socio-
economic impacts for families due to the
disasters. The average household income (ABS
household income, 2001) in each census district
is categorised into three income groups viz;
average ($500-$900 per week), quality ($900-
$1500 per week) and prestige (>$1500 per
week).
A Replacement Cost Estimate (RCE) provides
the cost of rebuilding a property in the event of it
being damaged beyond repair. The
reconstruction costs can be adjusted for inflation
in both capital city and regional centres by
applying available cost indices. Replacement
cost of structures for the residential component
of the NEXIS is estimated using the cost models
derived as part of the earthquake risk assessment
conducted for GA’s multi-hazard risk assessment
of Perth (Sinadinovski et al., 2005). Cordell
location cost factors (2001) were taken into
consideration to set the cost factor. The
replacement cost per square meter will depend
on the wall type, roof type and floor area. The
replacement value for individual structure is
estimated by multiplying the floor area and the
rate for reconstruction.
From several discussions with a number of
insurance organisations, it was apparent that the
estimation of the contents value in buildings is
difficult. The household income was found to be
the best variable to estimate the contents value in
general. Typically the contents value in the
average income group is 30 percent of the
buildings value, 40 percent for quality and 50
percent for prestige.
2.2. NEXIS – Business Buildings
Risk assessment for the business sector from
natural disasters is much more complex than
residential. The development of a business
component prototype is in progress. The risk
analysis for businesses is linked directly to
structural and inventory damage, business and
lifeline disruptions and overall macro economic
behaviour. The exposure information
requirements to assess the risk for business are
different from residential. To derive the required
information at a national scale is highly complex
and several assumptions for the generic approach
have been made where there is no fundamental
data available. The business information
comprises CBD commercial area, commercial
area in non-CBD and industrial buildings and
businesses.
Several datasets were referenced to derive the
information, which have different levels of
accuracy for those variables. Geoscience
Australia has surveyed several locations as a part
of the post disaster survey and collected building
information. It provides highly reliable
information for buildings but not the business
information. The Cityscope database includes
details on every property in the Central Business
Districts (CBD) of all major cities in Australia.
The focus of Cityscope is largely commercial but
also covers the residential properties in Sydney
and Melbourne. It is consistent across all the
cities and provides information about the
businesses, tenants, floor area etc. It lacks
information about building structure types. In the
process of making nationally consistent business
buildings exposure information, a generic
approach was developed where the relevant
information is missing and not covered by GA
Surveys and Cityscope. This generic approach
will fill the gaps of information available in the
GA Survey and Cityscope databases and also for
the areas not covered by these datasets.
The GA Survey database would be the first
database in our list to populate the data that is
required for commercial buildings. It will contain
some business and value estimation information.
If GA Surveys do not have this information, then
the Cityscope database is queried. The Cityscope
database contains some information about
structure details (number of storeys, floor area),
business and value estimation information. If
none of the specific reference databases are
available, a generic approach is currently being
developed to estimate the building details such as
structure details, business information and value
estimations.
The building footprints were mapped in the
Fyshwick and Hume industrial areas in ACT to
develop set rules for the generic approach to
estimate industrial exposure. The mapping of
industrial areas to derive information is also
highly complex.
1678
2.3. NEXIS – Ancillary Buildings
The exposure for ancillary buildings mostly
depends on the day and time. The buildings used
by the governments (federal, state and local),
emergency related (hospitals, emergency
services, police), educational (schools,
universities) and others (religious, museums,
stadiums) are covered in this category of NEXIS.
The required information related to ancillary
buildings is diverse and is required to capture
specific information like capacity of the facility.
It will further enhance the capability of temporal
changes in population dynamics.
2.4. NEXIS – Infrastructure
Infrastructure related to bridges, ports and
critical infrastructure of various sectors includes
transportation, energy, communication, banking
and finance and water will be captured from
existing specific databases. These have been
assembled as a central input to modelling the
system behaviour and critical infrastructure
modelling.
3. RESULTS AND DISCUSSION
The NEXIS residential component is being
utilised by a number of natural hazard risk
research projects within GA including: Cyclone
or severe wind risk assessment modelling; Flood
modelling and predicted impacts to residences in
Perth; Tsunami risk assessment modelling for
WA; and Post-incident analyses following
Cyclone Larry in Queensland.
A rigorous quality assessment is built in to the
NEXIS to identify the problems and gaps with
the system of supplying more meaningful and
realistic exposure information. The accuracy is
very low when the NEXIS database is derived at
buildings level. It is envisaged that the damage
estimations are accurate overall at an aggregated
census district level and can be used for
evaluating relative damages rather than absolute
figures. The adoption of G-NAF for the spatial
location has provided spatial accuracy. The
spatial accuracy of the buildings in the Gold
Coast city is depicted in Figure 2.
Figure2. Spatial locations of residential,
commercial and industrial buildings in Gold
Coast.
G-NAF is a nationally consistent geo-coded
addresses database but does not provide
consistent building information, such as number
of storeys. It is associated with inconsistent data
collection methods between states, several gaps
in the database (unknown) and lack of consistent
building information. G-NAF is being used for
spatial location of buildings and address. G-NAF
has sparse coverage for remote indigenous
communities, which are not incorporated. The
problems found through validation are being
reported to G-NAF custodians.
Due to lack of information available on
buildings, such as number of storeys, age and
construction type, assumptions need to be made.
A separate house is considered as single storey to
estimate the structure replacement value. NEXIS
is being used in risk assessment in GA and being
evaluated and modified. No information is
currently available that allows initial estimates of
impact at a building level. This information will
help local, state and federal governments identify
communities at risk and the possible impacts.
These studies can be expanded at the local level
if it is determined necessary. End users of
information and outputs generated using NEXIS
must be made aware of this and further research
will be conducted at local levels if accuracy is
required.
The number of buildings and the spatial location
is validated using high resolution images.
NEXIS is capturing more than ninety percent of
the buildings in the suburban residential areas.
The numbers are about sixty percent in rural
areas captured and spatially not located at the
building site. NEXIS provides average figures on
population demographics. The structure value
1679
and contents value aggregated at census district
are reasonably accurate in principle compared to
GA’s existing exposure catalogue (Robinson et
al., 2005). There is no consistent information
available to validate the accuracy quantitatively.
Even the insurance databases have several
constraints as to their use as reliable validation
benchmarks.
The risk assessment models for earthquake,
severe wind and tsunami inundation developed at
Geoscience Australia. NEXIS is being used to
estimate the risk from several hazards and also
critical infrastructure protection models. There is
a significant improvement in the assessing the
risk. The comparison of risk assessment using
exposure catalogue (aggregated to census
districts) and NEXIS (building level) for a set of
probabilistic return period wind speeds for
Sydney and the values are listed in Table 3.
Table 3. Percentage of losses for the Sydney at
each of six return periods.
Return Period Exposure
Catalogue NEXIS
50 Year 0.80 0.72
100 Year 1.11 1.01
200 Year 1.51 1.38
500 Year 2.01 1.85
1000 Year 2.31 2.12
2000 Year 3.01 2.77
4. CONCLUSION
The NEXIS is planned to provide information
from a generic approach for all natural hazard
risk assessment projects. It is aligned for critical
infrastructure protection modelling and analysis
and for future use in the upcoming alert systems
at GA. The development of strategic alliances
with external stakeholders is continuing to
capture more specific reference databases and to
provide aggregated exposure for areas of
interest. Providing high spatial accuracy of
building level information enhances the ability to
assess the risk, rather than the aggregated Census
District level which is currently being used.
NEXIS can assist in identifying possible impacts
to residential buildings and can be expanded to
assess the larger impact to local or state
economies. Communities can develop
preparation and evacuation plans by identifying
buildings at greater risk from a natural hazard
due to their location, building age or
construction type. NEXIS has been completed
for the residential component and is in the
process of developing the business component.
Further it will be extended to capture ancillary
buildings and infrastructure exposure
information. Age and number of storeys are very
critical in assessing the risk to the buildings. The
next version of NEXIS will evolve from a
generic approach to a specific reference exposure
database. In addition, NEXIS will improve
fundamental national datasets such as the ABS
Meshblocks and G-NAF by identifying and
reporting errors.
5. REFERENCES
ABS (2001), Functional Classification of
Buildings. Canberra: Australian Government;
Cat No. 1268.0.55.001.
ABS (2003), 2001 Census of Population and
Housing [CD-ROM]. Canberra: Australian
Government.
ABS Housing Survey (1999), Australian
Housing Survey - Housing Characteristics,
Costs and Conditions. Canberra: Australian
Government; Cat No. 4182.0.
COAG - Council of Australian Governments
(2003), Natural Disasters in Australia:
Reforming mitigation, relief and recovery
arrangement. Canberra: Australian
Government.
Harper, P., (2006). Emergency management
information development plan (EMIDP),
Australia. Information Paper: ABS Catalogue
No. 1385.0, Australian Bureau of Statistics.
Reed Construction Data (2003), Replacement
cost models for metropolitan Perth -
Consultancy report for Geoscience Australia.
Sydney: Reed Construction Data.
Robinson D., G. Fulford and T. Dhu, (2005),
EQRM: Geoscience Australia's Earthquake
Risk Model: Technical Manual: Version 3.0.
Canberra: Geoscience Australia; Geoscience
Australia Record 2005/01.
Sinadinovski, C. et al., (2005), Natural Hazard
Risk in Perth : Chapter 5 : Earthquake Risk,
Geoscience Australia Report, GeoCat No.
63527.
1680
