In this issue:

Director:
Professor Russell Blong

NHRC is kindly sponsored by:
Swiss Re Australia
QBE Insurance
Benfield Greig Australia
Guy Carpenter & Co

The Insurance Industry requires a firm understanding of the nature of building materials used in the residential housing stock in order to better estimate probable maximum losses caused by various natural hazard events.  In particular, brick veneer houses are less vulnerable to ground shaking and/or earthquake damage than houses constructed of double brick, so it is important to be able to provide relatively accurate estimates of the proportions of each building type across different areas in Sydney. A high ratio of brick veneer to brick houses could mean a significant difference in the potential loss payout by insurers.

The Earthquake PML Household Buildings - Sydney II study (Greig Fester 1997) notes that PML estimates using the brick veneer loss curves are considerably lower than those relying on the full brick curves, emphasising the value for insurers in distinguishing between these two forms of construction.

A preliminary study has attempted to quantify the distribution of different outer wall material types used in houses across the Sydney region.  The main aim was to provide a relatively accurate assessment of the spatial distribution of brick and brick veneer houses. A secondary aim was to collect data on the material used for roofs.

The 1971 Census provided the most recent complete count of separate brick and brick veneer houses.  These Census data formed the baseline to which was added building completion and approval data from 1971-72 to 1996-97.  Some supplementary data sources were used to compare the results.

A series of maps, tables and graphs have been created which illustrate the composition of the current stock of houses in Sydney.  In general, the data are based at the Local Government Area level which may prove an obstacle to the application of the results to the Insurance Council of Australia zones.

The distribution of different outer wall materials used for houses across Sydney generally follows the pattern of urban settlement and development.  The more established inner city, eastern suburbs, inner west and lower north shore areas are dominated by brick houses, while more recently developed areas in western and south western Sydney and the central coast are characterised by large numbers of brick veneer houses. The distribution of timber and fibro houses is also discussed.
 
The data used in this study show that, in the Sydney Statistical Division (all the LGAs shown in figure below), 37% of the stock of houses are made of brick veneer, 31% brick, 21% fibro, 10% timber and 1% other materials.

The changing use of different materials over the past 30 years has been influenced as much by cost factors and affordability as by changing consumer preferences. With the high interest rates and land costs experienced over this time, many developers and owner builders have opted for less expensive veneers and cladding.

This study provides some useful information for better estimating the composition of housing stocks.  However, there are a number of assumptions and data quality issues which require careful attention before increasing the scope of this study to include other capital cities of Australia.

For further information contact:
Frank Siciliano
Email:  fsicilia@ocs1.ocs.mq.edu.au
Phone: +61 2 9850 9473

   

The insurance industry, along with government relief, charity and emergency response services, plays an important role in determining the socio-economic impact of natural hazards.  Climate change, resulting from the anthropogenic enhanced greenhouse effect, may alter the severity, frequency or distribution of extreme weather related hazards such as bushfires, floods, tropical cyclones, storms and storm surge.  Property insurers and reinsurers are already financially vulnerable to changes in the frequency and severity of extreme weather events.  This vulnerability may be exacerbated by the potential impacts of climate change.

Details of the impacts of climate change involve a high degree of uncertainty, but they appear to present a high risk of increasing property losses as well as physical danger to future generations.  Appropriate responses to climate change at all levels is therefore a matter of risk assessment and risk management.  Information regarding the risk and cost of future natural disasters is also likely to be an important input to the policy development stage.  The insurance industry thus seems to be well equipped to be effectively engaged in the climate change debate and to be involved in developing local, national and international response actions.

The industry could respond to the threat of climate change either reactively (waiting to see what the impacts are) or pro-actively (adapting now).  Traditionally, insurers have coped with changes in risk or financial circumstance in four main ways:

  •  restricting cover to limit the risk;
  •  transferring the risk;
  •  raising premiums; and
  •  controlling claims through risk management.
The first three strategies are often implemented following a large event or series of events; that is, reactively.  Theoretically, all four could be applied as pro-active adaption strategies in response to the threat of climate change.  However, given the strong influence of competitive pressures, controlling claims through risk management appears to be the most effective way of reducing the impact of changes in risk that could be promulgated by climate change.  In particular, the following measures could be implemented as pro-active adaption strategies now:
  • using insurance-based incentives to encourage disaster mitigation measures by policyholders;
  • involvement in developing and endorsing appropriate building codes and zoning regulations in regions currently exposed to severe weather events; and
  • involvement in public education
Such strategies have the important benefit of reducing total losses to the community rather than redistributing losses among the stakeholders (as other insurance-based measures tend to do).

Given the well-established links between green-house gas emissions and climate change and the potential detrimental impact of climate change on the insurance industry, it may be prudent for the industry to also play a pro-active role in reducing any growth in greenhouse gas emissions (mitigation).  A number of possible strategies are indicated in Table 1.  In order to meet the expectations of shareholders and investors, such strategies must be "no regrets" strategies in terms of returns on investment:  that is, they must yield returns on investment that compare favourably with current insurance products and investment patterns. 

 
Based on "Adaption of the Insurance Industry to Climate and Consequent Implications", a report of research undertaken by Roy Leigh while employed at the Climatic Impacts Centre, Macquarie University.  The Research was supervised by Dr Roslyn Taplin and Dr George Walker and was funded by the Commonwealth Department of Environment, Sports and Territories.

Roy Leigh is currently employed by the Natural Hazards Research Centre and may be contacted on: Email: rleigh@laurel.ocs.mq.edu.au or telephone +61 2 9850 8118
 

 

The NHRC makes extensive use of GIS in determining the relationship between a hazard event and the damage caused by that event.  With the assistance of the Australian insurance industry the NHRC has compiled extensive databases detailing the type, extent and location of damage claims for damage to domestic buildings.

Some of the research to date is based on overseas experience and may not truly represent the Australian experience.  One of our research aims is to produce loss curves for building damage relevant to the Australian experience.  This requires intensive analysis of damage data, which would not be possible without GIS to provide the spatial framework for damage assessment.

The insurance industry is interested in estimating Probable Maximum Losses (PML) for given natural hazards. The NHRC has been involved in the development of PML models, particularly for losses that relate to domestic housing.  In order to produce loss curves, it is necessary to analyse the damage data from past events as well as information regarding properties that were not damaged by a hazard event.  The Sydney  March 1990 hailstorm is used here as an example of the value of GIS in event analysis.

On the 18th March 1990 Sydney experienced a severe hailstorm which caused $384 million (1997$) worth of damage from some 84,354 claims.  About 35% of the claims were for damage to domestic housing.  Data from this storm have formed the basis for the development of a hail PML model.  An initial analysis of damage undertaken by the NHRC  (NHQ Vol 2 Issue 2) revealed that claims for damage to buildings were spread across 130 postcodes (the Sydney area comprises of approximately 208 postcodes), with two postcodes holding approximately 20% of the claims. The largest hailstones (up to 8cm in diameter) were reported in and around the south-western suburbs of Liverpool and the nearby Bass Hill.  The strong winds accompanying the storm unroofed houses and brought down trees and power lines, with seventy suburbs blacked out.  In some areas it took two days to restore power.

The aim of the hailstorm research is to:  

  • determine the distribution of damage caused by this storm in terms of percentage of Total Sum Insured (TSI) and percentage of policies affected,
  • identify the domestic building and contents items most vulnerable to damage,
  • establish relationships between the item damage, the cost of the damage and the cause of the damage, and
  • develop relationships between losses and hailstone sizes in order to establish percent loss curves for the March 1990 hailstorm.
The development of the loss curves is presently underway, with much of the data analysis completed.  Data were obtained from several Australian insurers, who provided information on claims made in relation to this storm.  Some companies provided information on their complete portfolio so that percentages of claims to non-claims could also be calculated.  The analysis includes a total of 105,000 policies with a TSI of A$12 billion.

Our initial analysis used the ratio of the number of claims to policies per postcode. If only a small area of the postcode was affected, however severely, the ratio for that PC would be quite low.  Geocoding (locating the policy by street address using interpolation onto the appropriate street block) has enabled us to analyse the data on a much finer resolution than the postcode basis. Figure 1 shows the claims for the area of Sydney around the Bass Hill area, where the largest hailstone sizes were reported. The narrow but intense storm centre is now clearly visible.

An interesting feature of this figure is those areas with few claims that exist in areas with a high density of claims.  GIS was used to investigate those areas of low claim density.  A map was produced showing the housing density on a CD basis (the smallest spatial unit available).  When the claims and density maps were combined it was clear that low claims coincided with low housing densities, probably indicating industrial or parkland areas.

The ability to show various map layers is an invaluable tool in our analysis, and this is just one example.  Changes in basic GIS programs over the years have allowed us to improve our modelling techniques significantly. The numerical techniques used produce pages of numbers - interpretation of results becomes cumbersome without the aid of techniques such as GIS.  Most of our models rely on GIS to join spatial data, provide numerical model input and display the model results meaningfully.  As the NHRC undertakes modelling of other hazards, GIS will remain an integral part of the modelling process.

For further information contact:
Laraine Hunter
lhunter@laurel.ocs.mq.edu.au
Tel: +61 2 9850 9684

 


 

 

The Beaufort scale was devised in 1805 and became mandatory for log entries for all ships of the British Royal Navy in 1838, with the steps on the scale referring to a fully-rigged man-of-war. In 1874 it was modified for use on land. Numerical wind speeds were added early in the 20th century. The Modified Mercalli intensity scale (MM) has had a similar lengthy history - devised as a 10-point scale in 1902 by Mercalli, an Italian seismologist, and modified to a 12-point scale following a suggestion by Cancani. The Modified Mercalli intensity scale, the one used in many parts of the world, was produced in 1931 to fit construction conditions in California. The 1956 version, produced by Charles Richter, corrected a few anomalies. The MM scale is still alive and well, the latest New Zealand version appearing in 1996.

At NHRC, as part of a project that will utilise a new purpose-built damage scale focused on damage to buildings, we have reviewed the various scales that are already out there. To date we have come across more than two dozen different scales that express damage resulting from earthquakes, hurricanes, tornadoes, landslides, tsunami, hailstones, subsidence, and volcanic ash - and there are probably a lot more we haven't found yet!

The scales we have found use between 5 and 15 levels. Most scales have 6 to 8 levels indicating different degrees of damage. Even the Modified Mercalli scale, with 12 levels, has only 7 or 8 that relate to damage - this is probably enough, as the information available to assign a maximum damage level to an event or to an area is often sparse, of doubtful quality, conflicting, or even fanciful. Finer divisions seem unwarranted.

Some scales deal only with damage intensity; others refer to both the magnitude of the hazard and the damage intensity or damage potential. Scales such as the Saffir-Simpson hurricane damage potential scale begin with hazard information about wind speeds and surge heights to suggest potential damage. Others, such as the Fujita Tornado scale, begin with the damage and infer the associated wind speeds. That is, some scales work from Hazard -> Vulnerability, others from Vulnerability -> Hazard.


The above figure draws a comparison between the various levels on 6 earthquake intensity scales. The MM - MSK comparison is based on Munich Re [1988]. The MM - People's Republic of China (PRC) comparison is based on Krinitzsky [1993]. All other comparisons with MM are derived from Hopper [1984]. It is worth noting that comparisons drawn from other sources would have been different, emphasising that there is no unanimous agreement about the relativities between scales. The Damage % is based on the midpoints of the suite of MM - loss curves produced by Cochrane and Schaad [1992].

What qualities should a damage scale have? We think it should be simple to use, believable, robust with minor differences in diagnostic criteria making only small differences in the assigned intensity, and cross-cultural in its use. The criteria outlining the damage classes need to be specific, definable, and meaningful. A damage intensity should be assigned on the basis, where possible, of coherence in the information, the relative "values" of various sources of information, the proportion of unaffected buildings as well as those damaged, and without applying too much weight to single extreme observations.

How many of the damage scales currently being used have these qualities?  Few authors of scales other than those of the European Macroseismic Scale [1993] even discuss such issues. Whatever the answer, damage scales need to be understood and used as compromise solutions - no intensity scale can hope to resolve all disagreements between the diagnostic criteria for specific intensities. An intensity scale is a shorthand scheme for compressing lengthy descriptions - it is primarily descriptive rather than analytical. To paraphrase David Alexander, author of an intensity scale for urban landslide damage [1989]: The exercise of constructing a damage scale has contrasting aims - we require comprehensive, detailed and precise information yet, in the end, the scales have to be simple enough to use, frequently in the absence of solid information.

For more information about damage scales contact:
Russell Blong
Email: rblong@laurel.ocs.mq.edu.au
Fax: +61 2 9850 9394

 

nhq homepage
nhrc homepage