Risk Frontiers recently completed a project that involved applying the FloodAUS model (a GIS-based methodology for estimating mainstream flood risk on a per-address basis) to flood-prone urban areas in eastern Australia. The model was applied to 24 urban areas and flood risk ratings estimated for approximately 1.2 million addresses in 174 postcodes. The flood risk ratings are the Average Recurrence Interval (ARI) of inundation at ground level, 1 metre above ground and 2 metres above ground. An important feature of the project is that, because the same methodology was applied throughout, the results for different study areas are comparable.
While essentially the same methodology was applied to each catchment, no two studies were the same. This is because of the different data formats available, the existence of mitigation structures such as levees and the geography of individual catchments. In particular, physical characteristics such as depth of flooding and difference in elevation between consecutive design floods (see Figure 1) influence the value and distribution of risk ratings and affect average annual damage calculations.
Figure 1: Design flood levels compared with ARI 20y level.
The main outputs from the project are 24 databases containing risk ratings for each address in each study area. Presenting the results in different formats allows broad pictures of flood risk to be drawn. Flood risk can be expressed with respect to particular flood levels (ARI 20-year, 100-year etc), at different spatial scales (catchment, postcode, census collection district) and in terms of absolute numbers or proportions of addresses (risk density). It becomes apparent that different definitions of flood risk lead to different rankings of flood-prone areas. For example, the catchment with the most addresses below the level of the ARI 20-year flood contains only the fourth highest number of addresses below the ARI 100-year flood level. Similarly, the postcode with the most addresses below the ARI 100-year flood contains only the ninth highest risk density with respect to that flood level.
The different measures of flood risk mentioned above provide useful overviews. However, there can be significant variations in risk within quite small areas (within postcodes or along a single street). The proportion of flooded properties for different flood intensities also varies spatially, and there is a difference between absolute risk and risk density. To effectively manage portfolio and company exposure to flood risk and set appropriate flood cover premiums, flood risk data should be expressed on a property-by-property basis. At worst, broad scale risk ratings should be derived from higher resolution data.
For more information contact
The Auckland Region, New Zealand, faces risks from a number of different volcanic sources and a range of volcanic hazards. These hazards have widely varying return periods and physical characteristics. We propose a methodology for ranking these volcanic hazards and eruption sources based on their potential risk, and apply this to the Auckland Region.
Auckland is New Zealand’s most populated and fastest growing region. The Region has a resident population of approximately 1,200,000, or over 30% of the total New Zealand population. Auckland is also home to over one third of New Zealand businesses, the largest cargo port, and the nation’s busiest airport. Hazards from a small local eruption or a widespread distal eruption would cause significant damage, economic losses and disruption to the Region.
Auckland City sits on a volcanic field comprising about 50 small monogenetic volcanoes. The eruption of Rangitoto occurred within the last 800 years. Recent work within the Auckland area has identified tephra layers, with a frequency of around 400 years, originating not only from the local volcanic field but also from at least five distal sources within the North Island.
Figure 1: North Island, New Zealand showing Auckland and the volcanic centres identified as posing a risk to the region.
Risk was calculated, for a variety of volcanic hazards from each identified eruption source, as the product of the likelihood of the hazard occurring given that an eruption has in someway impacted the region, the proportion of the region affected and the proportion of expected building loss.
Values for each variable were determined from previously published geological studies and observations from similar historic eruptions. However, characteristics of many hazards were largely unknown. For this reason, individual hazard variables were assigned to logarithmic-scaled categories based on their expected order of magnitude.
These values were then multiplied by the relative probability of the event occurring and individual values, calculated for each hazard and event, were added to determine relative risk rankings.
The results show the largest risk to the Auckland Region is from tephra fall. This exceeded the next most serious hazards – base surge and lava flow - by an order of magnitude. Poisonous gases and acid rain, earthquakes, scoria fall, tsunami and mudflow all have risk ratings yet another order of magnitude smaller. Risks from pyroclastic flows, lightning and flooding are insignificant and there is no risk to buildings from climate variations.
In terms of eruption sources, Mt Taranaki represents the largest risk to the region due to the high frequency of tephra layers within the Region. The next largest risk is from the Okataina volcanic centre followed by an Auckland volcanic field eruption centred on land, the Tongariro volcanic centre and Taupo volcano. The smallest risks are from Tuhua volcano and an Auckland volcanic field eruption centred in the ocean.
In terms of building loss, our ranking highlights tephra fall as the most serious hazard. Because of its small expected thickness, tephra is unlikely to cause structural damage to buildings. Damage would be restricted to paintwork, roof coatings, air conditioning units, television aerials and possibly building interiors. Insured losses associated with this, potentially over the entire region, would be huge.
This ranking also strongly suggests that volcanic hazard planning for the Auckland Region should include the more frequent, widespread hazards from distal eruptions as well as the more intense, but concentrated hazards from the local volcanic field.
For more information please
contact Chistina Magill
Effects of disclosure of flood-liability on residential property values are disputed. This article reviews a substantial body of international research that addresses the issue. Most studies come from the United States, with lesser numbers from Canada, New Zealand and Australia. Salient points to emerge from this analysis are as follows:
The advantages of disclosed floodplain maps for flood risk reduction (e.g. land use planning, warning systems, public education) are undeniable. The balance of evidence suggests that the grounds for refusing disclosure are weak. Gaining a measure of public acceptance for disclosure is key. This requires best-practice risk assessment and a well thought-out plan for risk communication.
Careful consideration needs to be given to the content and timing of disclosure. Risk needs to be communicated in precise, understandable, succinct language: ‘probability’ is likely to cause less confusion and fear than ‘recurrence interval’. Scheduling disclosure when flood awareness is already high or the property market is particularly strong would also minimise disruption.
Nonetheless, acceptance of disclosure is contingent upon responsible reporting. Perceptions do exert an influence on property values. The ‘gatekeepers’ who shape perceptions ought to be co-opted as partners in risk communication.
Further reading: This abstract is based on an article soon to be published in the Australian Journal of Emergency Management.
Acknowledgements: The project was an initiative of the Floodplain Managers Group of Victorian Catchment Management Authorities, and was managed on behalf of the Group by the East Gippsland CMA. The project was funded by the Floodplain Management Unit, Department of Natural Resources and Environment (Vic.), and Emergency Management Australia. Opinions expressed here are those of the author and do not necessarily reflect those of Emergency Management Australia.
For further information
please contact Stephen Yeo
Risk Frontiers is updating QuakeAUS to a fully probabilistic model combining Monte Carlo simulation and GIS technology. Its principal use will be to provide the insurance industry with improved estimates of Probable Maximum Losses for Australian cities. In brief, the model comprises a number of subroutines that deal with:
The model uses probability density functions to allow for uncertainty in key variables; these functions describe the allowable range and relative likelihood of occurrence within that range for each variable. Simulation refers to the repeated random sampling from these distributions, each time calculating an event loss. The final output is a ranking of simulated event losses and exceedance return periods for PML calculations. Statistical stability or convergence is ensured by simulating a sufficiently large number of low probability events having recurrence intervals much longer than the time scales of interest.
A number of issues arise from the modelling. The first is the need to identify any zones within the study area having distinctive seismicity. Figures 1 and 2 illustrate this for Adelaide showing the area to the north of the city having many earthquakes but mostly small ones; Adelaide itself has relatively few quakes but has experienced several quite large events including a Richter magnitude 5.6 responsible for the 1954 losses. Breaking the study area into zones of differing seismicity increases the risk to Adelaide and diminishes the risk further north.
More critical is the difference in response to ground shaking between buildings with double-brick (unreinforced masonry) walls and other construction types. The non-linearity of these so called vulnerability curves and the very local nature of soil amplification means that averaging losses at the coarse scale of postcodes or larger is fraught with error. Figure 3 indicates the sort of differences possible in the loss ratio (claims as a percentage of total sum insured) for two construction types and two different soil types. Clearly finer grained averaging is important and QuakeAUS can employ portfolio data at the street address level if this information is available.
Upgrades to the Sydney and Adelaide models for residential portfolios have been completed and work on Melbourne and Perth is in progress. Modelling of commercial portfolios and business interruption are expected to be completed in early 2003.
Figure 1: Location of earthquakes of magnitude >2.5 within 500 km of Adelaide. The inner circle of 300 km radius indicates the area over which losses are estimated.
Figure 2: As for Figure 2: As for Figure 1 but with events of magnitude >5.
Figure 3: Estimated loss ratios for a notional ground shaking of MMI = 7 showing sensitivity of the loss ratio to soil type and building construction type.
For further information
please contact John McAneney.