Global and full-parametric optimization of multi-mode Rayleigh wave dispersion curves for estimating glacial ice structure

HB Ai and HX Li and XH Song and T Li and KJ Su and H Chen and Y Fu and ZX Mei, CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION, 68, 2320-2347 (2025).

DOI: 10.6038/cjg2024R0349

Ice thickness determination and freezing extent identification in subglacial sediments are of significant importance in glaciological studies. Various non-expensive and non-invasive geophysical methods, including Multi-channel Analysis of Surface Waves (MASW), are utilized to achieve these goals. Barnacles Mating Optimizer (BMO), quantitatively simulating the unique mating and reproduction characteristics of barnacles, has been proven to be an effective global optimizer in solving various optimization problems. Nevertheless, the effectiveness of BMO can be significantly affected by the selection of an algorithm- based parameter (the genital length pl), the methods of evolving barnacles, and the processing techniques when barnacles exceed the model space. We therefore introduce the Modified Barnacles Mating Optimizer (MBMO), combining a variable pl(var) technique, a novel evolving strategy for next generation barnacles, and an out-of-bounds correction method to achieve a full-parametric elucidation of multi-mode Rayleigh Wave Dispersion Curves (RWDCs), namely, to derive the layer thicknesses (h), densities (rho), the shear wave velocities (V-S), and the primary wave velocities (V-P) or V-P:V-S ratios of different layers from a joint inversion of picked fundamental and higher modes RWDCs, and to better distinguish different materials within the glacial structure. Synthetic RWDCs with different properties have been inverted the most successfully utilizing the MBMO algorithm in terms of the smallest data-fitting misfit and the most accurate model parameters with fewer uncertainties compared to those of the BMO optimizer and the Particle Swarm Optimization (PSO) algorithm. 10 well-established benchmark functions verified the superiority of MBMO in obtaining satisfactory optimization performance on a general scale. An additional inversion test with noise contamination and limited dispersion data validates the effectiveness of the proposed three modifications further. In the field data case, an active-source data set recorded by vertical geophones placed at Midtdalsbreen, an outlet of Norway's Hardangerj & oslash;kulen ice cap, was utilized to investigate the practicability of MBMO. The obtained results were in good agreement with previous inversions of RWDCs and available GPR interpretations. Furthermore, three detailed discussions illustrated that the interpretation process of MBMO was not sensitive to the selection of layers considering the range of 5-8; the accuracy of MBMO was undoubtedly corrupted by the consideration of a larger model space; yet additional depth information derived from co-located GPR data along the seismic survey line can be directly integrated into MBMO to generate satisfactory results again.

Return to Publications page