A spatial data partition algorithm based on statistical cluster

Nowadays, high performance parallel computation is deemed as a good solution to the complicated processing of massive spatial data. It is a very important precondition to make the most of this technology that data be partitioned. In this paper, we talk about the general strategy of spatial data partition and summarize its principles are good space proximity, balanced data load, small data redundancy and short time consumed. After analyzing the current partition algorithms, we find that there are many partition problems, such as the space division and load unbalanced. In order to solve these problems, we presented a new partition algorithm based on the statistical cluster method, which has better spatial proximity and data load than traditional algorithms.