Partitioning Strategy Research Of Forestland Border Data Based On The Parallel Computing

Forestland is the important resource in our country, it plays a very important role in making forest ecological system keep balance. Forestland border data is the reflection of forestland distribution situation, recording the spatial and temporal variations of forestland, and it provide the important data basis for the protection and planning of forest resources. In order to fully master the forestland distribution, the state has carried out a project which called national forestland "a map". Each forestland is marked with a label. It is the most comprehensive map for forestry. The data can be widely used in forest fire prevention, pest control, forestland planning, etc. The map makes forest management more precise. National forestland "a map" show the pattern of the national forest resources, has more than sixty-eight million forestland border sub-compartment data. In the face of such a large amount of forestland border data, the traditional methods have already can’t meet the requirements, and parallel computing is the effective means to address this issue of computationally intensive and data intensive. Partition of forestland border data is the precondition for parallel computing, but currently, the research of spatial data partitioning strategy is relatively few. The existing spatial data partitioning methods still have a variety of problems. They can’t meet the requirement of the massive forestland border data partition. It is also lack of quantitative research for data partitioning granularity. If data partitioning granularity is too large or small, it will affect the efficiency of the query. In this paper, aiming at these problems, we study on the partitioning strategy of forestland border data based on the parallel computing. Finally, we built a parallel query system for forestland border data, using the Liaoning Province forestland border data to make the parallel query experiment, verified the validity of the data partitioning granularity model, and obtained the optimal partitioning granularity with optimal query efficiency. In this paper, the research contents mainly as follows:(1) The parallel query analysis of forestland border data. Through analyzing the characteristics of the forestland border data, query application, parallel computing, and the application scenarios of parallel computing, built forestland border data parallel query model.(2) The forestland border data partitioning model. Through the analysis of the principle of data partition and parallel computing time consumption in the process of execution, built a model of data partitioning granularity and parallel execution time. The model described the relationship between data partitioning granularity and parallel execution.(3) Research for data partitioning method. By analyzing the shortcomings of the existing data partitioning methods, the article proposed a method based on dynamic grid and Hilbert space filling curve for massive forestland border data partition. With this, the spatial data can be divided quickly in accordance with the demand, and meet the static load balancing principles for data partition.(4) Experiment and performance analysis. For the parallel query experiment, the study built a forestland border data parallel query system, analyzed the parallel query test process, and determined the time of test records. The experiment is consist of attribute query and spatial query. The results show that the relationship between data partition and query time satisfy the data partitioning granularity model which was proposed in this paper, and got the best partitioning granularity for forestland border data.In this paper, the innovation points are as follows:(1) This article proposed data partitioning method for massive spatial data. The method integrated dynamic grid with Hilbert space filling curve, considered both the space data aggregation and the algorithm execution efficiency, and also can make sure that the volume of data blocks is consistent.(2) The quantitative research on the data partitioning granularity. By analyzing data partitioning principle, data partitioning granularity and parallel execution time, the article conduct the quantified research for the optimum partitioning granularity of forestland border data through the parallel query experiment.