Abstract: Gravity reference maps are essential foundational data for the navigation and positioning of underwater vehicles. As an important paradigm for multi-resolution spatial data management, hexagonal grids offer significant advantages due to their highest spatial coverage efficiency and angular resolution, as well as the uniformity of all grid cells. A representation paradigm for hexagonal reference maps in a certain area of the South China Sea is presented in this paper based on the H3 geospatial indexing system. Considering the diverse sources of marine gravity measurement data, a dispersion analysis of shipborne gravity measurement points is conducted, and several evaluation parameters related to measurement point density and spatial distribution are derived. Two sliding windows at scales of 1° and 0.25° are used for calculation and analysis. Experimental results indicate that, given the volume of multi-source marine gravity data, indicators such as density, average nearest neighbor distance, expected distance, nearest neighbor index (NNI), and coefficient of variation (CV) should be comprehensively considered to adaptively analyze the dispersion of measurement values in different regions. By performing multiple rounds of evaluation under varying spatial scales, the most suitable resolution of hexagonal grids for local sea areas can be refined, thereby achieving finer representation of gravity reference maps in local areas with adaptive spatial scale variability.