Abstract
Seismic methods constitute indispensable tools for characterizing subsurface structures in near-surface geophysical investigations. To address the inherent challenges associated with the joint inversion of Rayleigh wave dispersion curves and seismic refraction data, this study introduces an Enhanced Beluga Whale Optimization (EBWO) algorithm incorporating two principal advancements. A hybrid initialization scheme combining Tent chaotic mapping and reverse learning strategies is adopted to substantially enrich population diversity, thereby overcoming the limitations of purely random initialization employed in the original Beluga Whale Optimization (BWO). The performance of EBWO is systematically evaluated through a two-stage validation framework. First, benchmark tests on four complex multimodal functions demonstrate that EBWO achieves superior convergence accuracy and enhanced solution stability relative to the standard BWO. Subsequently, synthetic joint inversion experiments under both noise-free and noise-contaminated conditions further verify its robustness and reliability, with EBWO consistently outperforming both BWO and Particle Swarm Optimization (PSO). The applicability of the proposed approach is further demonstrated using field data sets from Türkiye, where EBWO yields lower data misfits, improved agreement with independent geophysical interpretations and available geological constraints, and greater inversion stability. By integrating surface wave dispersion and refraction datasets within a unified inversion framework, EBWO effectively exploits their complementary sensitivities, thereby alleviating the non-uniqueness inherent in single-method inversions. Overall, the results highlight the efficacy and robustness of EBWO for near-surface characterization and underscore its considerable potential for broader applications in multi-parameter geophysical inversion.