Large floods in South East Queensland, Australia: Is it valid to assume they occur randomly?

ABSTRACT

Flood frequency analysis and existing engineering practice assumes that flood events are randomly distributed. However, as understanding of large-scale ocean-atmospheric processes and associated hydroclimatic variability (and change) increases, the assumption that flood events are random seems unlikely to be valid. This paper examines historical flood data from 10 major catchments in South East Queensland (SEQ), Australia with the aim of testing for non-random patterns in the timing of the largest flood events. The results show that most (~80%) large floods in the study catchments have occurred within sets of 5-year periods separated by 35 years of lower flood risk. This finding is consistent with previous studies that have reported 20- to 40-year cycles in rainfall, flooding and drought across most of eastern Australia. Based on these insights, an explanation for the nominal 40-year cycle in SEQ flooding is proposed, and implications for describing and managing flood risk are discussed.

KEYWORDS:

• Non-stationarity
• flood risk
• climate variability
• climate change
• flood cycles

Acknowledgements

Thank you to Robert French for his careful reading of earlier versions of this paper and his insightful comments and suggestions. Thanks also to Sedigheh (Sara) Askarimarnani from the Centre for Water, Climate and Land (CWCL) at the University of Newcastle for assisting with the production of Figure 1.