In 1997 the NHRC set out to develop simple indices of natural perils risk for Australia. In 2000 we launched PerilAUS II, the software product that achieved this objective. While we have described previously some of the accomplishments of PerilAUS II at the postcode level (see NHQ Volume 6, Issues 2 + 3 - http://www.riskfrontiers.com/nhq/nhqfronttables.htm ), we have said little about the Relative Risk Rating summaries in PerilAUS II at the CRESTA Zone or ICA Risk Accumulation Zone level.

PerilAUS II is based on historic (20th century) and potential hazard data, weighted 30% and 70% respectively – that is, the risk ratings are weighted toward what could happen, rather than biased toward what has happened. The software provides Relative Risk Ratings (RRR) for nine natural perils – hailstorms, tropical cyclones, bushfires, earthquakes, floods, wind gusts, tornadoes, landslides and tsunamis. The Ratings can be compared peril-by-peril or zone-by-zone, allowing arithmetic calculation. The RRR are provided as simple files that can be downloaded to spreadsheets or other proprietary software. Additional tables indicate the standing of a postcode or ICA Zone relative to all other postcodes or ICA Zones. For example, the tropical cyclone rating for Trinity Beach (postcode 4879 – north of Cairns) is higher than the cyclone rating for 94.5% of the 2,573 postcodes included in PerilAUS II.

In fact, the rating for each postcode has been developed using a weighted linear combination method that integrates the risk in a postcode or ICA Zone. The nine perils have been weighted in the following descending order: tropical cyclones, floods, bushfires, wind gusts, hail, earthquakes, tornadoes, landslides, tsunamis – reflecting the total known damage to buildings in the 20th century (see NHQ 6(2)). The risk ratings have been developed, using GIS techniques, for each 2 km by 2 km cell across Australia – a total of more than 1.9 million cells. However, the ICA Zone Relative Risk Ratings have been combined from the postcode RRR rather than from simple combinations of the 2 km by 2 km cells.

The current ICA/CRESTA Zones were created in 1990. The boundaries, as shown on Figure 1, are a curious mixture with intricate patterns presumably reflecting local property boundaries in the north, and broad generalized sweeps in the southeast. Each CRESTA Zone contains specific postcodes, but the CRESTA zone boundaries shown in Figure 1 (next page) do not reflect accurately the contained postcode boundaries that are, in any case, revised several times per year by Australia Post.

Figure 1 shows the ICA Zones with the highest RRR occur scattered along the coasts of Queensland and New South Wales, reflecting the relative importance of tropical cyclones and floods in 20th century building damage, and one of the broad areas of major population concentration in Australia. Intermediate levels of RRR occur along coastal Western Australia, southeast Victoria and parts of coastal New South Wales. The lowest levels of risk, expressed at the CRESTA Zone level, occur throughout the continental interior and along the south coast, including most of South Australia, Victoria and Tasmania. The inclusion of Melbourne (ICA Zone 33) and Adelaide (Zone 27) in the lowest risk category will surprise many.

Figure 1: Relative Risk Ratings for 49 ICA/CRESTA Zones

Area-weighted RRR have their own problems when ICA Zones vary across almost three orders of magnitude in size. The smallest ICA Zone (Zone 17 - Darwin ) spans only 333 km2 whereas Zone 24 encompassing 2.36 million km2 includes much of Western Australia and most of South Australia. A typical tropical cyclone has a damage footprint many times the area of the smallest ICA Zone, but adversely affects only a fraction of the area of the largest Zone. At the other end of the scale, a typical tornado might be many times more damaging than a tropical cyclone but the damage footprint is usually less than a few square kilometers.

Such “problems” do not reduce the value of the RRR values or of PerilAUS II. More importantly, thinking through such issues encourages us to reflect on the nature of risk and how it might be most appropriately expressed for insurance purposes – for that matter, such issues persuade us to reconsider the advantages and shortcomings of the ICA/CRESTA Zones themselves. The Relative Risk Ratings in PerilAUS II provide the best summaries available of natural peril risk in Australia, and one of the very few attempts to provide quantified measures of integrated risk for any part of the earth’s surface.

The PerilAUS project includes NHRC, Bureau of Meteorology and Australian Geological Survey Organisation databases on the magnitude and frequency of natural hazards in Australia as well as maps and spreadsheets of the postcode- and ICA Zone-based Relative Risk Ratings. The PerilAUS project was funded by Macquarie University and the ICA Insurance Foundation. PerilAUS I and II can be purchased from NHRC.

For further information contact Russell Blong
or Keping Chen Tel. +61-2-9850 9683, email: NHRC@laurel.ocs.mq.edu.au.

The building damage suffered by residential properties during a flood can occur via a multitude of avenues. The force of the water can undermine building structures; pollution may be transported with the flood water causing contamination; inundation itself may give rise to items warping, swelling or cracking and the new environment contrived by the floodwaters is ideal for the growth of moulds. The focal point in this report is the actual tangible direct structural damage caused by flood waters and the cost involved in returning the property to its pre-flooded state, as certain materials used in buildings are more durable than others in flood conditions.

The flood

During a period of heavy rainfall throughout the semi arid tropics of north Queensland, the Cloncurry River began rising rapidly on the 2nd of March 1997. Rain fell at a rate of 25mm per hour and the river rose 6m attaining a maximum height of 10.1m that night. Floodwaters invaded the town of Cloncurry and inundated properties, much to the disbelief of residents. Flooding peaked at 2.6m overfloor in some dwellings before the water began to recede. Cloncurry had previously been flooded on a number of occasions (1883, 1954 and 1991) and the March 1997 flood can be estimated broadly as a once in 30-year event.

The aftermath

The floodwaters continued to advance and recede during the first week of March. Once the waters subsided, the full extent of damage was exposed and the magnitude of the inevitable cleanup was realized. The intense period of rainfall resulted in the inundation of 31 residential properties. Damage caused by the floodwaters was more severe in some areas than others and after assessment one property was judged beyond the cost of economical repair. The Disaster Impact Headquarters in Brisbane commissioned a survey that was conducted by a number of building inspectors on the 5th of March to acquire information in relation to building components and to assess the extent of building damage to the inundated properties.

The surveys

The surveys were used to record the physical details of each dwelling including the type of residence, its age, the construction and roofing materials, the building design and building components of the walls, ceiling, windows and flooring. The total cost of repair was split into various components including external walls, roofing system, internal walls/ceiling systems, floor systems, building services and ‘other’ (used to describe damage to kitchen/built-ins). The raw survey data were entered into an Access database and later exported to Excel for further analysis.

The damage

Figure 1 shows the contributions to the total of structural damage incurred by each property surveyed. Although the estimated repair cost differs for every property the various components that are damaged are consistent between dwellings. Despite a few extremes, most buildings cost between $7,000 and $17,000 to repair. It is evident that floor systems suffered approximately 40% of the total damage. Kitchen/built-ins and internal wall damage closely follow and it can be seen that each of these components form a substantial proportion of the total structural damage. Damage to the building services such as the general power outlets were reported by almost all properties, although they are only a minor contributing component in the total cost of repair.

Figure 1: Contributions to the total extent of structural damage incurred by each dwelling.

The majority of properties damaged were built on stumps with a smaller proportion on concrete slabs. The cost to repair houses built on timber and steel stumps averaged $13,620. However the cost to repair concrete stumps was much less at $11,300. The difference may be attributed to concrete stumps being less susceptible to water inundation. This is supported by the fact that damage caused to properties with concrete slabs averaged only $9,422. Floodwaters produce horizontal loads whereas stumps are designed to withstand vertical loads. Alternatively, the difference may be related to floor height and depth of inundation variations.

Once waters rise above floor height the first items to become inundated are the floor coverings. Most properties had both vinyl and carpet floor coverings. These were also the most expensive to repair at an average cost of $5,650 or $51 per m2. Carpet floor coverings sustained the least amount of damage ($2,850 or $49.50 per m2) though they were slightly more expensive to repair than vinyl floors (averaged $2,300 or $45 per m2). Further damage was generally reported after drying out as floor coverings may loose their dimensional shape and shrink.

Wall materials absorb water and damage often results due to warping, swelling or deterioration. Contaminants contained within the floodwaters are also responsible for staining. The average estimated cost to repair timber walls was $3,500, while the cost to repair walls constructed of fibro was $3,000. Plasterboard (gyprock) was the least expensive to repair and the average repair cost was $2,357. The average cost to repair the internal walls was $2,916.

Items constructed of timber were consistently found to be more expensive to repair and affected a large number of properties. Houses constructed of timber are more susceptible to flood damage and incurred a greater average repair cost of $13,207 while brickwork houses averaged $10,050. Although no damage was reported to foundations, it is interesting to note that the average cost to repair a dwelling on a timber foundation was $13,102 while repairs to dwellings on concrete foundations averaged $8,714. A similar trend can be seen for windows. Damage to aluminum louvre or sliding windows was estimated at an average of houses with $10,501 while the repair to dwellings with timber casement or hoppers was $14,715.

The reason for this trend in damage to timber items can be attributed to swelling or warping of timber upon water inundation. Soft timbers are more prone to flood damage than hardwood timbers. Damage to kitchen cupboards and built-ins were among the most frequently reported damage. Built-ins are usually constructed of chipboard (especially in newer properties) which is known to rapidly deteriorate once inundated.

The repair costs indicated above are roughtly estimated figures produced by building inspectors. The actual repair costs may significantly deviate from these. The independence and completeness of each survey is questionable due to similarities between them. Nonetheless it is possible that relationships are in fact meaningful. It is unknown whether this sample is representative of the whole population. Furthermore there is a chance that measures were taken to make dwellings more flood-resilient, particularly after the last Cloncurry flood (1991).

Damages compared The proportions of building damage at Cloncurry can be compared to similar studies conducted at Georges River and Nyngan by Water Studies Pty Ltd. The proportion of floor damage was substantially greater for the Cloncurry sample where almost 40% of damage occurred to the floor systems while only up to 4.1% at Georges River and Nyngan. These differences occur because floor coverings are included in the Cloncurry sample. There was a particularly strong emphasis (88.4%) on kitchen and built-in damage in Georges River with minimal damage occurring to other components. The distribution of damage at Nyngan is more diverse with a proportion of 50.5% damage occurring to the interior linings, 43.3% occurring to the built-ins, 3.3% to the foundations, 2.5% to the doors and windows followed by minimal damage to the floors (0.4%). The extent of damage at Cloncurry is significantly more evenly distributed. The floor systems (including floor coverings) suffered the most damage (38.2%) followed by 26% damage to kitchen cupboards and built-ins, 22.5% to the internal walls and 9.0% to the external walls.

The surveys of Georges River and Nyngan were both completed by Water Studies Pty Ltd, while the Cloncurry surveys were commissioned by the Disaster Impact Headquarters. This may partly explain some of the differences in damage proportions. Some differences may be attributed to different house styles. As an area is developed the dwellings constructed are generally of a similar price range and style. Timber sheeting and timber casement/hopper windows are common on older houses while plasterboard and aluminum sliding or louvre windows are used frequently in new houses.

Conclusion

Materials used in the construction of the house will have an impact on the cost of restoration to its pre-flooded state. Some materials are more resistant to the effects of water inundation than others. Timber is clearly susceptible to water damage and is also one of the most expensive to repair. Surprisingly more damage occurred to dwellings on stumps than to those on concrete slabs, while houses constructed from steel frame and brick tended to withstand water inundation more than other materials. Floor systems were a major contributor to total damage.

Building data on which this survey is based were supplied by John Rossiter and Bradley Clarkson at QBuild. We appreciate their assistance. Analysis of the data was undertaken by Monica Osuchowski, a Vacation Scholar in NHRC.

For more information contact Monica Osuchowski at mosuchow@postoffice.utas.edu.au
or Russell Blong at NHRC@laurel.ocs.mq.edu.au

Return to Natural Hazards Quarterly Homepage
Return to Natural Hazards Research Centre Homepage