Research On Methods Of Parallel Overlay Analysis In The Simple Feature Model Based On MapReduce

Overlay analysis is one of the most important operation in the spatial analysis of GIS(Geographic Information System).With the arrival of the era of big data and spatial data,overlay analysis process between the massive spatial data was time-consuming,the demand of rapid response to GIS spatial analysis user determine the overlay analysis need to achieve parallelism to improve efficiency.The existing overlay analysis parallel algorithms are mostly based on high-performance multi-core computer or parallel computing environment,to a certain extent,it enhance the performance of overlay analysis,but there are still some defects such as difficulty to horizontal expansion,poorly tolerance,and complexed to application.With the development of distributed computing technology,MapReduce parallel computing framework reflect its enormous advantages in distributed parallel computing,it has been widely used in the Internet field.Based on analysis of MapReduce parallel computing framework and overlay analysis of feasibility,according to the needs of overlay analyzing simple feature model,this paper proposes MapReduce parallel computing framework and do a research about data non-normal partition of the grid index construction method and simple feature model overlay analysis of parallel algorithm,on the one hand,it solves the data partitioning problems under the ground of many elements in the distributed cluster,on the other hand,it improves overlay analysis efficiency.The main research contents and results of this paper include:(1)Construction of MapReduce based on simple elements stacked parallel analysis algorithm.By analyzing the features of simple elements model,overlay ana lysis features and MapReduce parallel computing framework,to decouple and parall elization analyze traditional stacked flow analysis algorithm of filtering and refining in two stages,raising idea of Parallel data and constructing Simple Feature overlay a nalysis model of parallel computing framework basing on MapReduce.(2)Construction of non balanced grid index based on data partitioning.Parallelism is based on Data partitioning,through the analysis of data segmentation factors and take into account the simple feature model of spatial proximity,static load balancing rule base,this paper research a Hilbert space filling curve data partitioning method.On this basis,combined with the construction mechanism of the non balanced grid index in the data slice,taking into account the principle of minimum time consumption,built non balanced grid index through parallel method based on MapReduce data partition.(3)Designing the parallel algorithm for overlay analysis based on MapReduce.On the basis of data division and the construction of spatial index,combined with the algorithm flow of filtration and refining in the overlay analysis,this paper constructed a parallel algorithm based on MapReduce to solve the problem on overlay analysis.And through the experiment,it can be verified that the algorithm can significantly improve the efficiency of overlay analysis.