Why Can a ‘One-in-100-Year’ Flood Happen Twice in a Decade?

Geoffrey Chen

When Numbers Start Making Decisions · Season Five, “Things That Have Not Happened Yet” · Article 2

1. The risk clock does not reset after a flood

A town has just experienced what is described as a “one-in-100-year” flood. Residents naturally ask whether the next event is now about a century away. Should development be treated as relatively safe for the next few decades because the rare event has already been used up?

No. The phrase is a historical way of expressing an average recurrence interval, not a natural timetable. Australian flood management increasingly uses Annual Exceedance Probability, or AEP. A 1% AEP event is a flood magnitude that has a one per cent chance of being reached or exceeded in any year. An event this year does not automatically reduce next year’s probability. NSW Government: Flood Risk Management Manual

If annual risks are provisionally treated as equal and independent, the chance of at least one 1% AEP event in ten years is about 9.6%. The chance of at least two is about 0.43%: low, but not impossible. Over eighty years, the probability of at least one is about 55%, and the probability of at least two is about 19%.

The most dangerous feature of “one in 100 years” is that a long-run frequency sounds like a required interval between events. A design probability is heard as a guarantee of temporary safety.

2. The 1% line is not measured directly from the river

Flood probabilities combine rainfall and water-level records with terrain, catchment conditions, soils, river channels, drainage and hydrologic and hydraulic models. Those models translate rainfall into flows, depths, velocities and extents for events of different frequencies. Short records, urban development and changes to river works all affect the estimate.

A mapped flood line is not a natural mark waiting to be discovered on the ground. Continuous water elevations must be projected onto property boundaries. Model resolution, terrain surveys, assumptions about blockages and representations of buildings may move the line. Two adjacent houses can face similar physical risk while receiving different planning treatment because one falls just inside a mapped boundary.

AEP estimates also depend on a climate and land-use baseline. If extreme rainfall, sea level or catchment development changes, the level previously labelled 1% need not retain the same probability. The map’s date and scenario are therefore part of the number.

A common figure remains necessary. Without a design event, governments cannot compare levees, roads, drainage and building standards or allocate limited resources across risks. But 1% is a risk-management choice, not a scientific conclusion that land outside the line has no flood risk.

3. Why institutions rely on 1%

The 1% AEP event has long served as an important residential planning and flood-design benchmark in Australia. It creates a common point among safety, construction cost, land use and administrative practicality. NSW planning commonly adds freeboard—often around 0.5 metres—to the estimated 1% AEP flood level to create a Flood Planning Level. The added height accommodates waves, local blockage, model uncertainty and other effects not fully represented.

Freeboard is itself an admission that the modelled line is not an exact truth. A perfectly certain boundary would need no margin.

Nor is 1% the entire flood-planning system. The NSW manual uses the Probable Maximum Flood in defining the floodplain, and planning policy must consider events above the standard design event where evacuation, critical infrastructure or high hazard is involved. A hospital, aged-care facility or sole evacuation road cannot mechanically receive the same risk treatment as an ordinary dwelling.

Being outside the 1% area may mean that a routine floor-height rule does not apply. It does not mean there is no risk. Being inside does not mean a house will certainly flood; it means further assessment and specified design obligations are required.

4. How one modelled line changes land

Once a flood layer enters planning certificates, development controls, minimum floor levels, lending and insurance information, a probability begins to reshape reality. A property may need a raised floor, a preserved flow path, a restricted use or a specialist report. Prices and premiums can also respond.

Three decisions should be kept separate. A flood model describes relationships between events and terrain. A planning authority chooses a defined flood event and freeboard. Insurers, banks and buyers then use their own models for pricing or eligibility. The actors may share data, but they do not necessarily use the same threshold or exercise the same authority.

If a resident receives only the statement “your property is within the 1% extent”, they cannot know whether ground elevation, model version, drainage assumptions or a local feature produced the result. The final colour on a map acquires power while the chain that produced it becomes difficult to challenge.

5. Boundaries need stronger explanation near the edge

Suppose two neighbouring sites differ by two centimetres in modelled floor elevation. One falls inside and one outside. In practice, walls, driveways, drains and neighbouring works may create uncertainty larger than two centimetres. Yet planning consequences can jump at the line.

This is not sufficient reason to abolish a boundary. It is a reason to improve review at the margin. Owners should be able to obtain the flood study, model version, ground levels, design event, freeboard and planning provisions used. A qualified engineer should be able to submit better site-specific evidence through a reproducible technical process.

Review cannot mean that anyone who dislikes the risk can demand a new map. New evidence must satisfy professional standards. But complexity cannot justify refusing an explanation.

Map updates also produce questions of temporal fairness. An older house may have been lawfully built under earlier standards and later appear at greater risk. Imposing every cost immediately on the owner may be unmanageable; hiding the risk shifts loss to future buyers and emergency services. Information disclosure, staged retrofit, public mitigation and stricter rules for new development should be separated rather than allocated by one new line all at once.

6. Climate change makes recalculation a duty

Probability models are not permanent once completed. More rainfall records, new extreme events and altered catchments require frequency curves to be reviewed. A newer model may raise the level associated with 1%, or improved terrain data may reduce the mapped extent in a particular place.

Changes are often heard as proof that experts were previously wrong. Sometimes a defect was present. In other cases the evidence or the physical world changed. A responsible institution preserves versions, explains changed inputs and their effects, and provides transition arrangements for decisions that reasonably relied on the old map.

Correction does not mean searching for a perfect map that never changes. It means building a risk structure capable of learning. If a flood reveals a failed evacuation assumption, unmodelled blockage or missing data about vulnerable residents, the next study and plan must absorb that evidence. “The event exceeded design” should not become a ritual phrase that ends responsibility after every disaster.

7. The floodplain is also an action space we construct

River, terrain, buildings, model and plan do not first exist as complete independent objects and only later meet at a boundary. Development changes runoff; maps change buildings; buildings alter later flood behaviour. Flood risk forms within these mutual constraints.

The 1% AEP boundary is a forced structure. To act, government turns continuous risk into an area that can be mapped, approved and built upon. A forced structure need not be arbitrary; rigorous evidence can support it. The danger begins when the structure forgets its conditions, resists new evidence or silently converts risk outside the line to zero.

No engineer, councillor or resident knows the whole future of a catchment. Knowledge is distributed among instruments, models, local experience and emergency records. A responsible system preserves that evidence chain rather than publishing only the final polygon.

Conclusion: keep the 1% benchmark, abandon the fiction of “100 years of safety”

The 1% AEP event should remain a common residential planning benchmark and should be combined with explicit freeboard. A standard makes engineering and public investment comparable. Removing every threshold would merely move decisions into less transparent case-by-case discretion.

But the line cannot decide all critical facilities, evacuation safety or insurance questions. Planning must consider events beyond the design event. Near the boundary, model material and technical review should be available. Maps must state their version, climate assumptions and land-use assumptions and be reassessed as evidence changes.

After a 1% flood, the risk clock does not return to zero because it was never counting down. What should be reset is the belief that experiencing a rare event purchases a period of safety. Probability does not give the date of the next flood. It shows that a low annual chance can accumulate into a serious lifetime risk across the decades in which houses and roads remain in use.

Primary sources

Discover more from Geoffrey Chen

Subscribe to get the latest posts sent to your email.