Research And Implementation Of Parallel Strategy For Basic Spatial Analysis Algorithm Based On GRASS

Basic spatial analysis is one of the core functions of GIS,while the general information management system does not have this function.Spatial analysis is an important part of the relationship between spatial data processing and spatial data application model.By observing,analyzing and processing the original data characteristics,more knowledge is obtained,which provides a basis for the analysis of spatial behavior.With the rapid development of GIS application,the basic spatial analysis has gradually become an important method to deal with various geographical objects in the field of geographical research.However,with the gradual deepening of GIS application research,the researchers have become more and more accurate in the massive geographic data processing and real-time data processing.Their requirements on the functions of basic spatial analysis in terms of computational efficiency and performance are also increasing.Due to the basic spatial analysis algorithm is a kind of very important GIS algorithm,it is necessary for us to do further research on the algorithm structure.While,doing the introduction of parallel computing and doing parallel transformation on the basic spatial analysis algorithm,to deal with the time-consuming problem of handling large scale geospatial data.The main research contents and innovative achievements of this paper are as follows:(1)In the Linux cluster environment,this article uses the source code installation method to install GRASS GIS.Firstly,it uses Yum to analyze the external installation packages that GIS GRASS need to rely on.It makes the installation and debugging of the open source software GRASS GIS.Then,it introduces the MPI parallel scheduling interface to achieve parallel platform transformation of the GRASS GIS serial algorithm.(2)To do research on parallelization of calculating algorithm for intersection relation of spatial topology.This paper introduces the characteristics of the serial algorithm v.select firstly,which is one of the GRASS GIS spatial topology intersection relationship algorithms.It then analyzes the structure and main function of the serial algorithm,and determines the feasible data parallel partitioning scheme.It verifies the feasibility of the classic data parallel partitioning strategy based on geometric objects.Next,it proposes the new data parallel partitioning strategy based on arc segment.And,it does experiments to compare and analyze the acceleration ratio curve between its serial algorithm and parallel algorithm.(3)To do research on parallelization for neighbor object search algorithm.This paper studies the basic characteristics of the neighbor object search algorithm v.distance from GRASS GIS.Then,this paper determines the function and structure of the important functions from the serial algorithm,and explores the data parallel partitioning strategy suitable for this serial algorithm.By doing experiments,it can verify that the classic data parallel partitioning strategy based on geometric objects is suitable for this algorithm,and propose a data partitioning strategy based on region.At last,it makes comparisons between the serial algorithm and its parallel algorithm,and analyzes their characteristics.In the Linux environment,this paper realizes to make parallel transformation on these two kinds of serial basic space analysis algorithm,including algorithms for calculating intersection relation of spatial topology and neighbor object search algorithms,with MPI on the GRASS GIS environment.The experiment shows that it can get a better parallel speedup,which is enough to meet the practical application.