A conversation
In the wake of our 2025 floods (Jan 28 – Feb 12), North Queensland Conservation Council posted a call to action on facebook. I agreed with it and then tried to say why the “1-in-100-year event” label has become so obviously misleading.
NQCC: Floods and heatwaves are the new normal.
North Queensland has been hit hard again, and these so-called “1-in-100-year events” are happening way too often. Climate change is making extreme weather more frequent and intense, yet our leaders still aren’t taking the urgent action we need. …Malcolm Tattersall: “1-in-100-years” is a (sort of) average and what it really means is that there is one chance in one hundred of such a flood occurring in any given year. But that measure is purely statistical and historical. As such, it relies on the future being like the past, and climate change has completely undermined its validity.
The calculation is really very simple – too simple, in fact, to include any of the changes (e.g. urbanisation, dam construction, climate change) which might make nonsense of the prediction.
That was enough for facebook but pursuing the science further seems worthwhile so here we go.
The science
As I said, our official flood probabilities are based on historical records. The highest flood level in 100 years of records is defined as the 1-in-100-year flood level. That’s useful enough if every decade or century is assumed to be like every other, aside from normal weather variability. That assumption is called “stationarity,” a word only a scientist could love.
More precisely, the highest flood in our 100 year record is the level which is likely to be exceeded only once in a hundred years. It is sometimes abbreviated to ‘ARI’, ‘Average Recurrence Interval’. Statistical techniques can be used to estimate 1-in-500 or 1-in-1000 year levels from only 100 years of records but their accuracy is debatable.
Alternative measures
Some authorities are switching from quoting ‘1-in-X-year floods’ to a percentage probability. Since the strict meaning of a ‘1-in-100-year flood’ is that there is one chance in 100 of that flood level (or higher) in any given year, two in quick succession are not at all impossible, just improbable – like rolling double-six with a pair of dice, twice in a row. Saying that ‘there is a 1% chance of such a flood in any given year’ is mathematically the same as calling it a ‘1-in-100-year flood’ but might be clearer to the general public. Any 1-in-50-year flood is therefore a 2% flood and 1-in-25 is 4%, etc.
Where flooding is due to a river rising in response to heavy rain, and particularly where dams and urbanisation have changed the environment, comparing rain events may be as informative as comparing the floods. The BOM did that in its report on Townsville’s 2019 floods (see the BOM statement quoted here on Green Path).
Stationarity is dead
Stationarity obviously fails if a dam is built on the river, but planners have ways to work around that. It also fails if every decade or century is not like every other, but planners ignored climate change until recently on the grounds that the science was still too uncertain and the effects seemed likely to be too small to matter much.
But alarm bells began to ring in the science community early this century. For instance, Milly (see references below) said, “Stationarity is dead” and asked “where do we go from here?” in 2008.
Scientists have made good progress since then but, like most aspects of climate science, the research depends on lots of data. It’s also complicated, needing to pick out the effects of a slow ongoing change from the effects of larger, faster, random variability. Recent studies (a few are listed below) have identified the impacts of climate change on flood recurrence but only for particular locations; and they find that the effects in some locations are wildly different from the effects in others.
Tentative conclusions
One general conclusion is that, given enough data, it is now possible to say that (e.g.) what used to be a 1-in-100-year flood is now a 1-in-50 flood, or that (e.g.) a 1-in-20 flood is now half a metre higher than it used to be. But even this kind of result is specific to a location.
Some of the results are quite scary. Here, for instance, is Rohan Eccles:
…The objectives of this study were to determine the impacts of climate change and sea level rise on streamflow and floodplain inundation in the subtropical Logan-Albert catchment, Australia. … The largest events (100-year ARI) tended towards an increase, whereas the smallest (5-year ARI) tended towards a decrease. Floodplain inundation from a 100-year ARI event increased in all simulations and inclusion of sea level rise resulted in increased floodplain inundation area nearly doubling by the end of the century…
The bottom line is that the science is so new and complex that it has barely trickled down to the planners or the councils which employ them, let alone the popular science press or public knowledge.
Back in the real world…
- We know that the changes are real.
- We know that the most recent past is the best basis for predicting the future.
- We know that updating flood maps regularly will help to make them more realistic. Slater, for instance, recommends such updates every five years but laments that practice lags far behind the ideal.
…and Townsville in particular
- Ross Dam (built in 1976) changed the river so much that we might as well forget flood heights from before that date. Conveniently, that still leaves us very nearly 50 years of records.
- In the same period, Townsville’s population has tripled, further reducing the relevance of old flood records.
- If council doesn’t keep up to date, we can do it ourselves, however roughly, by looking at what has happened locally in the last 25 or 50 years, i.e., in living memory.
- There happen to have been no comparable rainfall events between cyclone Althea in 1971 and the Night of Noah in 1998 (see BOM statement linked above), so we can simply look at how high the water reached in the 2025 floods and ask our neighbours or older relatives about 1998 and 2019.
- In terms of rainfall, 2019 broke Townsville existing records for both 7-day and 10-day totals. 2025 was a little smaller than 2019.
- By the definition of “Average Recurrence Intervals” over the years 1975-2025, then, 2019 was our 1-in-50-year rain event, with 2025 and 1998 as runners-up. Which of them created the worst floods depended somewhat on our exact location but 2019 seems to have been the biggest.
- Taken alone or as a group, those three events are clearly not 1-in-100-year floods, let alone the 1-in-500 or 1-in-1000 that some politicians claimed in 2019. They are, at most, 1-in-50 year floods. More realistically, perhaps 1-in-25.
- That gives us odds of 25 to 1 against another similar flood next year. How lucky are we feeling?
Technical reading
Milly, P.C.D. et al. Stationarity Is Dead: Whither Water Management? Science 01 Feb 2008: Vol. 319, Issue 5863, pp. 573-574. DOI: 10.1126/science.1151915 https://www.science.org/doi/10.1126/science.1151915
Slater, L., Villarini, G., Archfield, S., Faulkner, D., Lamb, R., Khouakhi, A., & Yin, J. (2021). Global changes in 20-year, 50-year, and 100-year river floods. Geophysical Research Letters, 48, e2020GL091824. https://doi.org/10.1029/2020GL091824
Rohan Eccles, Hong Zhang, David Hamilton, Ralph Trancoso, Jozef Syktus. Impacts of climate change on streamflow and floodplain inundation in a coastal subtropical catchment. Advances in Water Resources, Volume 147, 2021, 103825, ISSN 0309-1708, https://doi.org/10.1016/j.advwatres.2020.103825
Wasko C, Westra S, Nathan R, Orr HG, Villarini G, Villalobos Herrera R, Fowler HJ. 2021 Incorporating climate change in flood estimation guidance. Phil. Trans. R. Soc. A 379: 20190548. https://doi.org/10.1098/rsta.2019.0548
