Research On Efficient Grid Coding Method And Application For Large-scale Spatiotemporal Data Organization

The continuous development and maturity of sensor equipment and information technology have greatly enriched spatiotemporal data resources.Facing massive,multi-source heterogeneous,multi-scale and dynamic spatiotemporal data,the traditional spatiotemporal data organization method is difficult to meet the current application requirements.The spatiotemporal data organization method based on the grid model has the characteristics of high index efficiency,suitable for parallel operation,etc.,which may provide a solution for efficient organization of scale spatiotemporal data.However,existing grid coding methods for spatiotemporal data organization have poor applicability to multi-scale data,and the large-scale data indexing and query capabilities they support are insufficient.In view of the above problems,this paper mainly studies efficient spatial/spatiotemporal grid coding methods,and takes remote sensing images and spatiotemporal trajectory data as examples to realize the efficient organization of massive spatiotemporal data to support efficient spatiotemporal data analysis and application.The main research content and results of the paper are as follows:1.Aiming at the problem of poor clustering of existing space filling curves in scale-dimension,a cross-level space-filling curve(CSC)is proposed.CSCs have high clustering of space and scale dimension,CSC-based codes have characteristics of recursion and high computational efficiency.Experiments show that the clustering of CSCs are improved by more than 20% compared with the existing multi-scale space filling curves,and as the total level of the grid increases,the proportion of clustering increases continuously,but there is a convergence trend.The query efficiency of subunit-codes is improved by more than 17 times on average,and is not easily affected by changes in the level of query codes.2.Based on the quadtree grid division method of latitude and longitude space,the two-dimensional CSC and its coding method,an efficient earth surface spatial grid coding and calculation method was designed and provided for the organization of remote sensing image data.Based on the octree grid division method of the space-time area,the three-dimensional CSC and its coding method,an efficient spatiotemporal grid coding and calculation method was designed and provides for the organization of spatiotemporal trajectory data.The above spatial/ spatiotemporal grid coding and calculation methods are mainly completed by computer bit operations,which can make the calculation and data query efficiency high;and due to using CSCs,the spatial/spatiotemporal grid coding method is more applicable to multi-scale data.3.An improved polygon meshing method is proposed to improve the accuracy of polygonmeshing filling results under the condition of limited number of meshes.Compared with the existing methods,the experiment shows that the improved method has a significant improvement in accuracy.The paper proposed and implemented spatial indexing and querying methods of remote sensing image data to study the data organization method of space-time separation.Experiments show that: by setting the number of associated grids,index efficiency,query efficiency and accuracy can be better balanced;in the comparison experiment with Oracle Spatial and Geohash,the query efficiency of the proposed method is increased by more than 10 and 2.5 times on average,the average accuracy of its query can reach 94.6%,and there is no missing inspection.4.The paper researched and implemented the gridding algorithm of spatiotemporal trajectory and query area to realize the effective correlation between spatiotemporal data and grid,then proposed the spatiotemporal index and query method of trajectory points / lines to study the data organization method of spatiotemporal integration.Compared with the spatiotemporal Geohash-based data organization method,the results show that: the proposed method has increased by about 16% and 2.8 times on average,in the query efficiency of massive trajectory points and trajectory lines,respectively.