The Floods That Never Were

There are two distinct methods of determining peak flood discharges for a stream. Statistical analysis of recorded stream gauge data, and estimation of discharge from rainfall runoff routing. The former method is generally preferred as it is considered more realistic, but it requires an appropriately placed stream gauge and sufficient data to be statistically meaningful. Runoff routing allows estimation of peak discharges in catchments for which there is inadequate stream gauge data. However, the procedure requires numerous assumptions about the rainfall event and catchment. The underlying assumption is of this method is that the event probability of the rainfall and resulting flood are directly related. A recent flood study of Hutton Creek, located in the upper reaches of the Fitzroy River in central Queensland, used both methods to independently predict flood discharges. The resulting predictions showed significant (order of magnitude) difference. Further investigation identified almost no correlation between rainfall and streamflow. This assessment clearly demonstrated that the basic premise of the runoff routing method . that flood frequency is directly related to rainfall frequency . does not hold true in all cases. In the Hutton Creek catchment, and doubtless many others across Australia, the characteristics of the catchment and storm such as temporal and spatial distribution and rainfall losses can have far greater impact than the amount of rain that actually falls.