Research On Organization And Management Of Spatial Temporal Data Of Climate Model Based On Tensor

With the accumulation of the geographical space observation data , the enhancement of the monitoring ability of in earth environment, meteorology and society, and the development of geographic information system and computer technology, a huge amounts of spatio-temporal data has formed. The climate model is a mathematical model of the climate system, with the help of calculating the complex evolution process of the climate system. The mass space-time field data generated with climate model, has a massive, multi-dimension, complexity and analysis oriented features. The management and analysis of these data has become an important bottleneck of existing information systems, and has become a hot issue of academic research. Tensor is the most natural form of high dimensional data , which can maintain the original data structure, and can be used for multi-variable, multi-object unified coordinate system. So it is a good way to solve the organization and management problems of multi-dimensional space-time field data ,which producted by the climate model.Based on the theory of tensor and tensor decomposition of mathematics,constructing the data expression and compression model of multi-dimensional spatio-temporal, defining the data structure and file storage scheme based on the data model, and constructing strategy of data retrieval and multi-scale analysis method. On the basis of the earth's climate mode data,using vmmlib and Eigen ascomputing engine and VTK (Visualization Toolkit) visualization engine, constructing tensor climate model of temporal spatial field data management and visualization system,which realize the compression storage of spatio-temporal field data, dynamic analysis and perspective of data management and visualization expression frame. The main work in this paper includes the following aspects:(1)Based on tensor of spatio-temporal data model and storage structure.According to the organization management and requirement analysis of climate model data, constructing spatio-temporal field data model , based on tensor constructing the original tensor tensor space, and based on tensor Tucker decomposition and hierarchical decomposition, constructing data mode and file storage scheme . The structure of spatial temporal field tensor data is defined, and the file storage of climate mode data is realized.(2) Spatio-temporal data retrieval and analysis based on tensor data model. On the foundation of tensor Tucker and hierarchical decomposition of spatio-temporal data mode, by retrievaling and sampling spatio-temporal field data decomposition results to retrieval and analysis of multi-scale. And based on the tensor hierarchical decomposition the spatio-temporal field data compression and dynamic appended has realized.(3) Constructing the tensor spatio-temporal field data management and visualization system which realized spatio-temporal field data compression and storage, dimension perspective, dynamic analysis and multidimensional data visualization expression, and verification and analysis of cases with the climate model of temporal spatial field data.Based on the mathematical theory of tensor , building the spatio-temporal data model and data storage structure , which can realize spatio-temporal field data compression, spatial retrieval, multi-scale analysis and dynamic additional, and it has certain reference meaning to solve the explosive growth of geoscience spatio-temporal field data and information from massive data mining.