Wide range lower atmosphere duct parameter inversion from automatic identification system signals using hybrid strategy artificial lemming algorithm

Lower atmospheric ducts significantly alter the propagation of Very High Frequency (VHF) and higher-frequency electromagnetic waves in the marine boundary layer, critically impacting radar and communication systems. Effective duct monitoring is essential for optimizing these systems. Emerging, more economical remote sensing approaches—such as satellite-based remote sensing and the monitoring of ubiquitous shipborne Automatic Identification System (AIS) signals—provide promising alternatives for large-scale, cost-effective data acquisition compared to the traditional monitoring methods (e.g., radiosondes, lidars). To tackle the challenges posed by the inversion of lower-atmospheric duct with large vertical extent and high parameter dimensionality, robust and accurate parameter inversion techniques are urgently required. To address this, this study proposes the Hybrid Strategy Artificial Lemming Algorithm (HSALA), an intelligent optimization framework for prior-information-free duct inversion. Comparative inversion simulations of HSALA, the standard Artificial Lemming Algorithm (ALA), Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and Harris Hawk Optimization (HHO) demonstrate HSALA’s superior accuracy and stability under ideal, noise-free conditions. It achieves a mean inversion accuracy exceeding 98% and reduces the root mean square error (RMSE) by over 80% on average in noise-free conditions across 30 trials. Further validation using field-collected AIS data yields a mean parameter inversion accuracy of approximately 81.4%, confirming the method’s practical applicability while highlighting the performance gap introduced by real-world complexities such as signal noise, model bias and atmospheric horizontal inhomogeneity. This method provides a promising and effective solution for operational duct monitoring using AIS signals, bridging a significant gap toward real-time, large-range inversion. The insights from the field validation underscore the value of this approach for engineering practice and outline a clear path for future refinement.