Comparative Analysis of Wavelet-Based Anisotropy Index Calculation for Bathymetric Data

This study investigates the relationship between seabed structures and ship manoeuvrability by applying an anisotropic transformation to bathymetric data. The one of the selected study areas is the downstream section of the Mississippi River in Chalmette, Louisiana, a region characterized by meanders and a maximum depth of approximately 50 meters. The other study areas, which differ in their geomorphological and navigational characteristics, are the Upper New York Bay, Sacramento-San Joaquin River Delta and West Florida Escarpment. Bathymetric data, obtained from NOAA, were converted to fractional anisotropy maps using three wavelet kernels: Coiflet-1, Haar, and Daubechies-4. The anisotropy index was calculated per grid cell to capture directional dependencies in the seabed topography, which may influence ship movements. Statistical analysis, including descriptive statistics and non-parametric tests, was performed to compare the effectiveness of each wavelet kernel. The findings suggest that the Haar kernel is optimal for shallow areas, while the db4 kernel is most effective for detecting anisotropic patterns associated with heading deviations. This study demonstrates the importance of integrating seabed characteristics into predictive models for autonomous navigation, particularly in complex, shallow, and narrow waterways.