Introduction

The average annual cost of flooding in Australia is estimated to be $314 million (BTE 2001), making it the most costly of all natural hazards. Historically, insurance has covered about 10% of the total cost of flooding, but as residential flood insurance becomes more widely available, this proportion is likely to increase. Comprehensive, good quality flood risk data are therefore important and valuable.

To this end, Risk Frontiers has developed FloodAUS, a GIS-based model for estimating mainstream flood risk in urban areas on a per address basis. The main output for a catchment or flood prone area is a database of street addresses, each with a flood risk rating. FloodAUS risk ratings provide estimates of the Average Recurrence Interval (ARI) of inundation at ground level, 1 metre above ground and 2 metres above ground. Insurers can use this information to assess flood risk for their portfolios and to formulate portfolio optimisation strategies. The information is also potentially very useful for emergency response agencies and state and local government departments.

The first large-scale application of the FloodAUS model was completed in March 2002. The methodology was applied to 24 urban areas in eastern Australia and flood risk ratings estimated for approximately 1.18 million addresses in 174 postcodes. The study areas are shown on Figure 1 and listed in Table 1. This is the first time in Australia that flood risk in a significant number of areas has been assessed using essentially the same methodology.

Figure 1: Locations of the 24 FloodAUS study areas

This article reviews FloodAUS, examines the wide range of data types and formats encountered during the project and describes challenges faced (and surmounted) in managing and incorporating data into the FloodAUS framework.

FloodAUS

FloodAUS has been described in NHQ Vols. 5(3) and 6(3) and the methodology is summarised in Figure 2. The model uses the following information to estimate flood risk:

Digital terrain models (DTMs)
Flood surface elevation information
Proprietary street address databases

 

Combining the DTM for a flood prone area with a flood surface yields information about the extent and depth of flooding and forms the basis for estimating the risk rating. Figure 3 shows a 3-D representation of inundated areas for an ARI 100-year flood in an urban area in New South Wales. Inundated areas are shown in blue. Dark blue indicates deep water and light blue shallow water. Ground elevation ranges from green at lower levels through brown to red at the high points. Black lines indicate roads.


Figure 3:
3-D representation of an ARI 100-year flood

Although the broad methodology applied to each study in the recently completed project was the same, detailed methods required for each study varied considerably. Every study was to some extent idiosyncratic because of:

Geo-physical characteristics of each area
Variety in the type and format of available terrain and flood data
Existence or otherwise of levees
Quality of address data

Digital terrain models

Terrain data were generally procured from state government departments or local government authorities. Format and resolution varied considerably as indicated in Figure 4. The most common format was 2 metre resolution digital relief contours. High resolution data such as 0.5 metre and 1 metre contours were available for some study areas, but only low resolution 5 metre contours or 25 metre by 25 metre grid data for others. Only 3 out of 24 (12.5%) studies relied on less than the highest quality DTMs to represent ground elevation on the floodplain. Techniques were developed and applied to these areas to improve the DTM representation of ground levels.


Figure 4: Range and proportion of terrain input formats incorporated in FloodAUS studies to date

Creating, manipulating and incorporating terrain data into the FloodAUS model was labour and computer-power intensive. Often the purchased data required considerable processing before they could be used to create DTM “sheets”. Some study areas involved dozens of DTM sheets. Creating contiguous DTMs with proprietary software involved substantial amounts of computing time – over 24 hours for some study areas. Manipulating such large files within the model framework required skill and patience.

Flood surfaces

Flood surfaces were created in FloodAUS to approximate the actual height and shape of the water surface during a flood. Ideally FloodAUS relies on several flood surfaces, the most appropriate being the 20, 50 and 100-year ARI events and Probable Maximum Flood. Flood surface input data were sourced from the most recent available flood studies – usually hydrological and hydraulic modelling studies performed by reputable consulting engineers. The modelling results from these studies were presented in a myriad of different formats including:

Water surface elevation contours
Water surface elevation points
River cross sections
Maps of inundation extent
River profiles
Continuous surfaces

Each required different techniques for managing, translating and incorporating the data. Except for the continuous surfaces, the data were generally in hardcopy form.

Fifteen of the twenty-four FloodAUS studies completed in the project included flood surface data to the Probable Maximum Flood (PMF) or Extreme Flood Event (EFE) level. For 5 study areas the most severe flood for which data were available was the ARI 100-year event. 81% of addresses with risk ratings were derived from studies with flood information to the PMF/EFE level, while 10% of risk rated addresses were derived from studies with flood information to only the ARI 100-year level. Over 75% of the flood surface data were less than ten years old, with an average age of 7 years.

Street address data

Street address databases for 20 of the 24 studies were derived from the MapInfo products Streetworks and MapMarker. A street address database is simply a list of addresses in the format: number; street name; designation (Street, Road etc); postcode. Address databases for three study areas for which there were no Streetworks data were created from electronic cadastral maps provided by the relevant local government authority. The address database for one study was created by laboriously digitizing the information from a hardcopy cadastral map.

Levees

Flood mitigation effects of levees were incorporated in the FloodAUS studies during the flood surface creation and risk rating calculation steps. Levees existed in 11 of the 24 study areas. Overall, 7 ring levees and 10 open levees were incorporated into the studies. Ring levees fully enclose the protected area either with a continuous earthen bank or wall, or tie into high ground to complete the “ring”. Open levees often run along the riverbank parallel to the direction of flow or curve back on themselves to deflect flow. They do not necessarily join or tie into high ground. The defining difference is that open levees tend to flood slowly from behind the levee before the crest is overtopped, while the area within a ring levee is theoretically not flooded until the flood overtops the levee crest. The behaviour of levees and the manner in which they are modelled in FloodAUS depends on factors such as:

Type - open or ring
Size of area protected
Levee bank crest height
Location relative to direction of flow
Slope of flood
Rate of rise of flood

Techniques were developed to incorporate the effects of different types of levees in different situations.

Conclusion

In March 2002 Risk Frontiers completed a project that involved applying the FloodAUS model to flood prone urban areas in eastern Australia. The project covered:

24 urban areas
174 postcodes
1,177,854 addresses with flood risk ratings

Since the same methodology was applied to each area, the results for one area can be readily and legitimately compared with other areas.

An important operational outcome of the project was the realisation that no two studies are the same. At the outset we naively assumed the model could be applied to different areas in a “cookbook” fashion. The reality was that each study had different geo-physical scales and characteristics and the requisite input data came in a myriad of types and formats.

In particular, there was little consistency in the presentation of flood modelling results produced by different consulting engineers, and considerable variation in the resolution and format of the digital terrain data available within Australia. Almost every study revealed new issues or challenges that had to be resolved within the FloodAUS framework. The end result, however, is a more sophisticated and refined suite of tools available for application to other flood prone areas in Australia.

Reference

Bureau of Transport Economics 2001, Economic Cost of Natural Disasters in Australia, Report 103, BTE, Canberra.

For further information:

visit our FloodAUS website: http://www.riskfrontiers.com/floodAUS/floodausbrochure.htm

or contact Roy Leigh or Laraine Hunter on telephone: +61-2- 9850 9683/email: riskfrontiers@els.mq.edu.au


 


Email: riskfrontiers@els.mq.edu.au. Website: http://www.riskfrontiers.com/

View further Risk Frontiers newsletters
Return to Risk Frontiers homepage