PSLE Science Reality Lab Vol No.123 | “A 100-Year Flood Happened Last Year” — Are We Safe for the Next 99 Years?

PSLE-SCI-REALITY-0123

Wait, What? A “100-Year Flood” Can Happen Two Years in a Row

A news report says a river has experienced a 100-year flood.

A pupil hears the phrase and says, “Then the next one should be about 100 years away.”

That is not what the phrase means. In hydrology, a “100-year flood” is commonly used as shorthand for a flood magnitude with about a 1% chance of being equalled or exceeded in any given year, under the statistical model and conditions used for the estimate. The clock does not reset to zero after the event. A rare flood can happen again the following year.

The Reality Lab job is not to turn Primary Science into advanced hydrology. It is to learn a powerful evidence habit: when a scientific phrase sounds like a schedule, check whether it is actually a probability.

Quick Answer

Translate “100-year” into the annual probability intended by the source.
Do not treat the number as a fixed waiting time between events.
Remember that rare events can cluster by chance.
Check what physical quantity defines the event: river flow, river level, rainfall amount or something else.
Check whether the probability estimate is based on old or current conditions and how much uncertainty surrounds it.
Do not assume a “100-year storm” automatically produces a “100-year flood”; catchment conditions and storm duration matter.

The Exact Learner Job This Page Owns

This page owns one real-world evidence-transfer job: interpreting recurrence-interval language in a scientific or news-style claim without converting a probability into a calendar promise.

It does not replace the canonical owners for weather, rivers, probability, graphs or model limits. It applies those skills to a phrase that appears in flood reports, engineering discussions, maps and news headlines.

Original Reality Lab Case: River M

This is an original composite teaching case with constructed data.

A fictional river has a flood level that scientists estimate has a 1% chance of being equalled or exceeded in any one year. A major flood reaches that level in 2025. A local social-media post says:

“The 100-year flood already happened. We should be safe until 2125.”

The post has changed a probability into a schedule. If the 1% annual probability model remains appropriate, the next year still has its own chance of a flood of that magnitude. The event last year does not spend the next 99 years of “flood allowance”.

Observed, Modelled and Claimed

Layer	Statement
Observed	A flood reached a stated river level in 2025.
Modelled estimate	A flood of that magnitude is associated with about a 1% annual exceedance probability under the statistical analysis.
Unsupported claim	“The next such flood cannot occur for 99 years.”
Why unsupported	Annual probability does not create a fixed interval between individual events.

What “Annual Exceedance Probability” Means

Hydrologists often use the idea of annual exceedance probability, or AEP. A 1% AEP event has a 1 in 100 chance, in the statistical model, of being equalled or exceeded in a given year.

The U.S. Geological Survey explains that the term “100-year flood” is often misunderstood for exactly this reason. It is a probability description, not a promise that the event arrives neatly once every century.

Why Two Rare Events Can Happen Close Together

Probability does not force events to spread themselves evenly through time. Imagine tossing a fair coin. Heads has a 50% probability on each toss, but you can still get heads twice in a row. The two results do not have to alternate perfectly.

A 1% annual event is much rarer than heads, but the same reasoning principle applies: rare events do not carry a rule saying, “Once I occur, I must stay away for a century.”

The Name Can Mislead You

“100-year flood” sounds like a clock. “1% annual chance flood” sounds like a probability. The second phrase often communicates the scientific meaning more clearly.

This gives a wider Reality Lab lesson: technical shorthand can be mathematically correct but still invite an everyday-language misunderstanding. A careful learner translates the label back into the scientific quantity it represents.

One Hundred Years Is an Average Recurrence Idea, Not a Countdown

If a statistical process stayed stable for a very long time, an event with a 1% annual probability would occur about once per hundred years on average across the long run. But individual events would not be spaced exactly 100 years apart. Some gaps could be much shorter. Others could be much longer.

That is the difference between an average pattern over many opportunities and a guarantee about one particular sequence.

The Event-Definition Check: Flood of What Magnitude?

Before accepting a “100-year” claim, ask what physical threshold is being discussed. It may refer to a river flow rate, a river level, a rainfall depth over a specified duration, or another measured quantity.

The event must be defined because different thresholds have different probabilities. “A flood happened” is not enough information to know which recurrence estimate applies.

A 100-Year Storm Is Not Automatically a 100-Year Flood

This distinction matters. A storm is a precipitation event. A flood is a response of a river or drainage system. The same rainfall can produce different flood outcomes depending on where it falls, how long it lasts, how wet the ground already is, the size and shape of the catchment, drainage, land cover and other conditions.

USGS explains that a storm with a particular recurrence interval does not automatically create a flood with the same recurrence interval. The causal chain contains more than rainfall amount alone.

The Data-Length Check

How can scientists estimate a 1% annual event when they may not have 10,000 years of observations? They use statistical models fitted to the available record and scientific information. This means the estimate has uncertainty.

A short record can make very rare-event estimates less certain. One station may have a longer record than another. Changes in land use, drainage or climate can also make historical data less representative of current conditions.

The learner does not need to calculate a flood-frequency distribution. The evidence habit is simpler: an estimated probability is not the same thing as a directly counted future schedule.

The Stationarity Check: Are Conditions Still the Same?

Recurrence estimates often rely on assumptions about the statistical behaviour of the system. If a river catchment is heavily urbanised, a reservoir changes flow, drainage systems change, or climate patterns shift, the relationship between past observations and future risk may change.

This does not make probability estimates useless. It means the model has conditions and must be updated when the physical system changes enough to matter.

The Representation Check: What Does a Flood Map Colour Mean?

A flood map may colour an area associated with a 1% annual chance flood. A reader can easily interpret that colour as “this place floods exactly once per century” or “everything outside the colour is safe”. Both are too strong.

The map is a modelled representation based on thresholds, elevation, hydraulic assumptions, data and uncertainty. It is evidence for risk assessment, not a wall separating certainty from impossibility.

What Would Strengthen a “100-Year Event” Claim?

The physical event and threshold are clearly defined.
The source states the corresponding annual exceedance probability.
The observation record and statistical method are documented.
Uncertainty or confidence bounds are reported where appropriate.
Changes in land use, drainage, climate or measurement practice are considered.
The communication avoids implying a fixed 100-year waiting period.

What Would Weaken It?

A headline says “once every 100 years” as though the timing is guaranteed.
The report does not define whether the event refers to rainfall, river flow or water level.
A very short observational record is presented as exact certainty about rare events.
A historical estimate is reused after major physical changes with no review.
The occurrence of one event is used to claim the next 99 years are safe.

Worked Case 1: Two “100-Year” Floods in Three Years

A town experiences two floods described as near the 1% annual level within three years. Does that automatically prove the scientific estimate was fraudulent? No. Rare events can cluster. However, repeated extreme events can also be a reason to re-examine the underlying data, model and whether conditions have changed. The correct response is investigation, not automatic dismissal or blind acceptance.

Worked Case 2: “We Had One Last Year, So We Are Safe”

This claim fails because it treats recurrence interval as a cooldown period. A probability estimate has to be interpreted anew for each year under the model; last year’s event does not eliminate this year’s possibility.

Worked Case 3: A “500-Year” Headline

A headline calls an event a “500-year flood”. Before repeating it, ask which agency or study assigned that recurrence interval, what threshold was exceeded, what data were used and how uncertain the estimate is. A dramatic label is not enough evidence by itself.

Worked Case 4: The Same Rain, Different Flood

Two catchments receive the same total rainfall. One produces severe flooding; the other does not. This does not contradict the rainfall measurement. Flood response depends on additional variables such as catchment size, soil saturation, drainage and where the rain fell. The event label must match the physical object being measured.

Tempting Reasoning That Fails

“A 100-year flood comes once every 100 years.” It is a probability shorthand, not a fixed schedule.
“If two occur close together, one must be mislabeled.” Clustering is possible, although the model may still deserve review.
“A 100-year storm guarantees a 100-year flood.” Rainfall and flood response are different scientific objects.
“Outside the 100-year flood area means zero risk.” Probability zones are not boundaries between possible and impossible.
“The recurrence interval is known exactly.” It is estimated from data and models with uncertainty.

How Far Can the Conclusion Travel?

A well-supported 1% annual exceedance estimate can help describe flood risk for a defined magnitude under stated assumptions. It cannot promise that the next event is 100 years away, guarantee that no larger event will occur, or remain permanently valid if the river system and climate conditions change.

PSLE-Style Transfer Case

A chart states that a certain river level has a 2% chance of being equalled or exceeded in any year. A pupil says, “That means it happens exactly once every 50 years.”

Question: Explain why this is too strong.

Reasoned answer: A 2% annual probability corresponds to an average recurrence idea of about 50 years, but individual events can happen closer together or farther apart. The probability does not fix the time between events.

Explained Practice

Practice A: A 1% annual event happened this year. Is next year’s chance automatically zero? No.

Practice B: A “100-year storm” occurred, but the river stayed below the “100-year flood” level. Is that scientifically possible? Yes. Storm magnitude and flood response are related but not identical.

Practice C: A flood estimate was calculated decades ago before major urban development. What question should you ask? Whether the probability model and physical assumptions have been updated for current conditions.

Delayed Independent Return: The R-A-R-E Check

R — Recurrence term: What label is being used?
A — Annual probability: What yearly probability does it correspond to?
R — Real event: What physical threshold is being equalled or exceeded?
E — Evidence limits: What data, model assumptions and uncertainties affect the estimate?

Parent and Tutor Teaching Guide

Use a bag containing 99 white counters and one blue counter. Explain that the blue counter represents a rare annual event in a simplified model. Replace the counter after each draw. A blue counter can appear twice in a row because each draw is another opportunity; there is no rule forcing 99 white draws after a blue one.

Then remove the bag and ask the learner to explain why the model is only an analogy. Real flood probabilities are estimated from changing physical systems, not literal identical draws. This second step protects against replacing one misconception with an oversimplified model.

Authoritative Sources

The official Singapore Science frame asks learners to interpret information, evaluate methods, weigh assumptions and uncertainty, and communicate reasoning. Recurrence language tests all four habits at once: the label sounds simple, but the scientific meaning lives in probability, event definition and model limits.

The Quiet Return

Rare does not mean scheduled.

When a scientific phrase sounds like a calendar, translate it back into the probability it was meant to describe.

Small Group Tutorials