A systematic framework for automated rip current detection from numerically modeled flow data

Abstract Rip currents are strong, offshore‐directed water jets that pose severe safety concerns to bathers at recreational beaches. While numerical modeling has been increasingly used in studying rip current dynamics and forecasting nearshore hydrodynamics, automatically detecting rip currents from model data remains challenging due to variable spatiotemporal characteristics and complex flow patterns associated with different types of rip currents. In this paper, a systematic framework for automated rip current detection was developed and validated across a range of bathymetries and wave conditions using data from the Boussinesq model FUNWAVE simulations. Specifically, the framework integrates physics‐based criteria, considering both rip current kinematics and structural characteristics, along with multi‐step identification and tracking procedures that account for spatiotemporal dynamics. Manual validation shows over 95% accuracy in locating rip currents and over 90% accuracy in detecting their occurrence times across different rip current types. Additionally, complex evolutionary behaviors, including the coexistence of multiple rips in close proximity and significant structural changes in individual rip currents, can be effectively detected. Furthermore, the method's potential for real‐world applications is demonstrated by its successful implementation under a realistic bathymetry and observed wave conditions at Habin Beach, Zhuhai, where rip currents lasting from a few minutes to hours at multiple locations were detected over a 5‐d period. Overall, the systematic and automated framework for rip current detection demonstrates promising outcomes for enhancing the analysis of numerically modeled data, with possible benefits from both scientific and hazard mitigation perspectives.