Drivers of Tropical Cyclone Induced Flooding

Abstract

<p>Tropical cyclone induced flooding has had limited research despite the regularity of occurrence, extensive damage, and fatalities associated with these events. In Australian coastal waters, cyclogenesis frequently occurs close to the coast, with little lead time for specific warnings announcement. The identification of predictors for tropical cyclone induced flood events would be of benefit to emergency services. The aims of this research are to address this gap by analysing the relationship between tropical cyclones and flood generation. Tropical cyclones are complex multifaceted systems. Therefore, this research was completed in three phases: 1. analysis of hydrology and climatic data to distinguish flood origin and importance of tropical cyclone flood generation at a catchment level; 2. identification of the differentiating attributes of tropical cyclones; and, 3. prioritisation of tropical cyclone attributes as predictors of flood severity, magnitude and extent. </p> <p>Flooding and the influence of cyclogenesis on the Barron River, located in the Wet Tropics, Far North Queensland, Australia, was investigated. This catchment experiences, on average, 1.4 tropical cyclones within a 400 kilometre radius per season. Between 1915 and 2014, 50 percent of all annual peak discharges greater than a two year average recurrence interval were associated with a tropical cyclone. This percentage of occurrence increased to greater than 80 percent in all sections of the catchment, for floods with an average recurrence interval of greater than 20 years.</p> <p>Nine tropical cyclone attributes were distinguished through examination of the historical record and dissection of radar images of East Coast, Australia, landfalling tropical cyclones. The identified tropical cyclone attributes (landfall location or proximity to the coast; direction of movement; category; duration within the catchment; size; rainfall volume; rainfall distribution; rainband asymmetry; and rainband rotation at landfall) were evaluated to determine their potential as tropical cyclone induced flood indicators. A MIKE 21 hydrodynamic model (DHI, 2011) was constructed for the catchment. Manipulation of the tropical cyclone inputs enabled the isolation of the variables. Statistical analysis prioritised the importance of the attributes as predictors of flood severity, magnitude and extent. Of the four predictors recognised, two primary attributes were identified to be of high importance to flood prediction (rainband asymmetry; and location of landfall or proximity to the coast). Secondary attributes were tropical cyclone category and size. The inclusion of the rainband asymmetry within tropical cyclone warnings or models represents a potential tool for flood prediction and management strategies.</p>