Drift Trajectory Prediction for Multiple-Persons-in-Water in Offshore Waters: Case Study of Field Experiments in the Xisha Sea of China

With the increasing frequency of maritime activities, large-scale man overboard incidents raise higher demands on maritime search and rescue (SAR) decision-making. Most existing drift models are designed for single-person-overboard situations and have limited ability to model multiple-persons-in-water (MPIW) scenarios. To address this gap, this study proposes a drift trajectory prediction method for MPIW based on full-scale field experiments in the Xisha Sea, South China Sea. In December 2023, six drift experiments were carried out, providing 57 h of tracking data under typical conditions with wind speeds from 0.17 to 7.77 m/s and surface current speeds from 0.06 to 0.96 m/s. Two basic MPIW scenarios were considered, side-by-side connection and random connection, and four MPIW drift models were built for upright 3-person (UP_3), upright 5-person (UP_5), upright–facedown–upright (U-F-U) and facedown 2-person (FD_2). The corresponding wind-induced drift coefficients were estimated. The stochastic variability of the crosswind leeway (CWL), including sign-change frequency and the probability of positive CWL, was systematically analyzed. For unconstrained regressions, the downwind leeway slope coefficients range from −2.96% to −12.61%, while CWL slope coefficients range from 1.01% to 2.78%, depending on group configuration. Monte Carlo simulations were then used to compare different model groups. In typical test cases, the proposed MPIW models reduce the normalized cumulative error for 11 h trajectory prediction from 0.18–0.23 to 0.08–0.17, indicating a clear improvement in the accuracy of search area delineation for group drowning scenarios. The results provide a useful reference for MPIW drift prediction and SAR decision-making in similar offshore and deep-water environments.