Daily to yearly variations in rip current activity over kilometer scales
Rip currents are seaward directed jets of water that originate nearshore and frequently occur along many U.S. beaches. Rip currents are well known to be the number-one public safety risk at the beach, yet there are research voids, particularly in regard to rip current forecasting. This dissertation seeks to describe the factors that influence the daily to yearly variations in rip current activity and provide the statistical basis for a probabilistic rip current forecast model. First, an open-source toolbox to process and analyze directional wave spectra from Acoustic Doppler Current Profilers (ADCPs) is presented. The toolbox, Doppler Profiler Waves Processing toolbox (DPWP), proves to be a flexible alternative to the instruments' proprietary software and provides comparable performance. DPWP processes all ADCP data used in this dissertation. Second, an analysis of historical rip current rescue data collected by Kill Devil Hills (KDH) Ocean Rescue on the Outer Banks of North Carolina from 2001 to 2009 is described. This analysis suggests that rip currents are most likely when there are large significant wave heights, a shore-normal wave direction and at low tidal elevations. The presence of two swells increased the likelihood of rescues when there were large differences between the mean directions of each swell. Alongshore location is important, as the southern half of KDH tends to be more favorable to hazardous rip occurrence than northern KDH. Third, daily variations in observed rip intensity are related to wave field and surf zone bathymetry features. Rip intensity was found to increase substantially when the daily averaged significant wave height exceeded about 0.7 m, and then increase gradually as the significant wave height approached 2 m. Rip intensity was also found to be greatest at locations where there were substantial surf zone bars that varied in depth (∼ 0.5 m) over 50 m alongshore. Lastly, a probabilistic rip current forecast model is created using rip current observations and a logistic regression formulation. Given a set of input predictor variables, the probabilistic model predicts the likelihood of hazardous rip current occurrence (0 to 1). Using rip current rescues to indicate hazardous rip current occurrence the probabilistic model has a Brier Score of 0.15 (0 is perfect prediction) compared to a minimum Brier Score of 0.45 for the present National Weather Service (NWS) Weather Forecast Office model. The change in score represents a 67% improvement in prediction for the probabilistic model compared to the NWS model.