Diverging projections for flood and rainfall frequency curves

Abstract

Flood damages are projected to increase with climate change due to intensification of extreme rainfalls, with water scarcity also projected to increase due to longer periods of drought leading to less runoff from flood events. In order to adapt to climate change, precise and robust projections of flood events are required of a magnitude (or rarity) relevant to engineering design and water resources planning. However, due to the complexity of catchment-specific processes which influence flood response, many studies disagree on the direction and magnitude of the projected change: some studies project increases in flood magnitude, even above projected increases in rainfall maxima, while other studies project decreases. Using a combination of continentally downscaled rainfall projections for Australia and rainfall-runoff models locally calibrated to rare floods using a novel objective function, we show that projected changes in flooding do not follow projected changes in extreme rainfall, with changes in flooding depending on region, runoff mechanism, and event severity. Decreases in frequent flooding up to the 1 in 5 annual exceedance probability are projected across the continent, with increases only projected for frequent flooding in some tropical regions, diverging from the projected increases in rainfall maxima across a majority of the continent. For rare events of annual exceedance probability 1 in 50, the flood magnitude is projected to increase across the northern and eastern coasts by the end of the century – commensurate with the median projected increase of 20% in rainfall maxima. However, decreases in both rainfall maxima and flooding are still projected for south-western Australia, even for the rarest events. The regional coherence of our results suggests the patterns in the projected changes may be applicable to other tropical, temperate, and arid regions around the world.