Advances in quantifying the compound flood hazard in South Florida

Extensive development on low-lying, karst coastal catchments leaves South Florida vulnerable to flooding through a combination of rainfall, storm surge and high groundwater levels. Flooding drivers are typically produced by the same large-scale weather conditions and exhibit evidence of statistical dependence. In existing level of service assessments of the flood control structures which discharge canal water to tide, the South Florida Water Management District assumes peak rainfall intensity and the peak water level on the ocean-side of these structures (O-sWL) are fully dependent i.e., that the 100-year rainfall and the 100-year O-sWL always occur simultaneously. Through the novel application of several multivariate statistical models, able to capture the dependence between the drivers, we demonstrate the conservative nature of this assumption. According to regional sea level rise scenarios, however, this implicit factor of safety may be rapidly eroded. Without modelling the full hyetograph and “intensity” of an O-sWL event it is not possible to determine the impact of a flood event or even whether the rainfall and surge events coincide in a way that flooding occurs or is enhanced. We therefore generate time varying events that account for the time lag between the peaks by exploiting the intrinsic structure of hyetographs associated with peak rainfall events and the relationship between O-sWL intensity and the other rainfall/surge characteristics. The synthetic series can act as boundary conditions for unsteady flow models to estimate the physical interaction of the flooding drivers more robustly. The time invariant models were also extended to incorporate groundwater levels.