Abstract: The paper discussed the key step of the process of generating Discrete Global Grid System, which is correspondence between planes grids and spherical grids. In the paper, an optimized target function was proposed to evaluating geometrical properties of discrete global grids. Optimizing by genetic algorithm (GA), direct spherical partition grids, which met optimization requirements under limited layers, were obtained. Applying control point data of the grids and combing theories of numerical projection, global hexagonal discrete grid system, with more even geometrical properties, was constructed. Experimental results proved that, compared with current Snyder Equal-Area polyhedral projection, the grid units obtained by our method were more even and had 2.5 to 3 times higher operation efficiency.

Key words: discrete global grid system, hexagon, numerical projection transformation, genetic algorithm