A spatial data partitioning and merging method for parallel vector spatial analysis

Based on the principle of the proximity of spatial elements and the equilibrium of spatial data´s size, this paper presents a data partitioning and merging method based on spatial filling curve and collection of spatial features. In the data reducing section, this method takes the principle of dynamic tree merging and reduces the times of data serialization and deserialization. The experiment shows that such methods can cut down the time of every process´ computing and merging, improve the load balancing degree, and make a great improvement to the efficiency of parallel algorithm and expandability.