the 22nd Pacific−Asian Marginal Seas meeting (PAMS)

Bathymetric information is fundamental to navigation safety, coastal infrastructure design, marine environmental management, and disaster mitigation. In the coastal waters of the Korean Peninsula, national hydrographic surveys have been conducted continuously since 1957. However, survey observations are spatially concentrated near major ports and shipping routes, while other regions and periods exhibit survey gaps, resulting in uneven spatiotemporal coverage. Despite the accumulation of long-term hydrographic survey data, studies that systematically analyze survey coverage across Korean coastal waters domain remain limited. This study analyzes long-term hydrographic survey coverage across the Korean coastal waters using survey data collected since 1957 and proposes a grid-based spatiotemporal interpolation framework to reconstruct bathymetry from multi-decadal survey records. Hydrographic survey datasets (1957–2025) archived by the national hydrographic authority were collected, standardized, and reorganized into long-term gridded bathymetric datasets.
The methodology consists of five main steps: (1) data preprocessing, including outlier removal and coordinate system standardization; (2) selection of representative grid cells in dredged and non-dredged areas to examine artificial and natural depth variations; (3) analysis of survey density and spatial distribution patterns using clustering algorithms (k-means and mean-shift); (4) interpolation of bathymetry using a spatiotemporal inverse distance weighting (ST-IDW) method that incorporates both spatial proximity and temporal dependency; and (5) evaluation of prediction accuracy and data quality improvement through year-wise leave-one-out cross-validation and bias correction.
Validation results indicate that the mean interpolation error decreased from 0.62 m to 0.45 m in the dredged area and from 0.18 m to 0.12 m in the non-dredged area after bias correction.

bathymetry,hydrographic survey,spatiotemporal interpolation,ST-IDW,clustering analysis