Modeling And Implementation Of Land Cover Spatiotemporal Data Representation

For half a century,various earth observation instruments have accumulated a large number of geospatial distribution time series data.At present,the work of national geographic state monitoring is in full swing,making the periodic collection of China's land cover spatial distribution time-series data possible.And how to manage these time-series data to form the spatial-temporal data of the process of land cover change(LCC)becomes one of the problems to be solved urgently in current researches.According to the previous study of spatio-temporal data storage expression model(STDSEM),they can be seen that it is very difficult to establish a universal STDSEM.To reduce this difficulty,it is necessary to establish the STDSEM according to the needs of specific application research.In this paper,the present situation and the existing problems are studied,and the following theoretical and modeling studies are carried out to study the spatial-temporal data storage of LCC.(1)Taking the time and space view of GIS as the starting point,this paper systematically studies the current spatio-temporal expression methods of GIS.In addition,the merit and demerit of each method is analyzed.And the conclusion is that the implementation of universal STDSEM is very difficult.Then,the existing methods and their shortcomings of temporal-spatial expression in the study of LCC are expounded.Based on the actual demand of the further quantify of the process of LCC,and linking process-and pattern-based approaches,possible solutions are proposed in this paper.(2)According to actual changes of LCC,a method of spatio-temporal process expression of land cover based on process object is proposed.This method is based on the spatio-temporal events in the hybrid data model,which further organizes the spatio-temporal change events in the lifecycle of a patch spatio-temporal object into microscopic spatio-temporal processes,and the mutual transformation between patches into local spatio-temporal processes.Then local processes become parts of the change of macro landscape pattern.This method associates the change process of the land cover with the spatial distribution,and establishes the association between the process-and pattern-based methods.(3)Aiming at the requirement of spatio-temporal expression of LCC,a method to determine the type of change on land cover and to establish spatio-temporal inheritance relationship are proposed.Firstly,land cover patches are abstracted into polygon geographical objects,and then the change types of the patch itself and between patches are summarized according to spatial and topological relations change.Then,types of patch change and types of mutual transformation between patches are determined.Finally,the method of spatio-temporal inheritance relationship establishment is determined according to these types of change.(4)In order to manage and store spatio-temporal process data of LCC,a hybrid spatial-temporal data model based on object-oriented and snapshot is proposed.Firstly,the spatial data of the geographic entity is stored and managed by the snapshot model,in which the spatial distribution data of all the geographic entities are stored in a raster snapshot,and the spatial data of each geographic entity are stored in vector snapshots.Then,we use the object-oriented model to associate the snapshots with the space-time inheritance relationship,and at the same time to correlate the change events and other attribute information of the geographic entities to form a hybrid spatio-temporal data model with the spatio-temporal object as the constituent unit.The model can help the LCC study to further quantify the change of spatial characteristics from microscopic patches to macroscopic distribution.(5)In order to validate the effectiveness of the model and method proposed in this paper,a prototype system of land cover spatio-temporal data management and expression is established.Based on Moderate Resolution Imaging Spectroradiometer satellite remote sensing time series data,the spatio-temporal database of Heilongjiang Province from 2002 to 2012 was established.The ability of spatio-temporal data management and expression model was analyzed by spatio-temporal query,statistical analysis and patches' variation pattern mining.The original contributions of this paper are embodied in three aspects:a)A expression method of spatio-temporal process based on process object is proposed,which organizes the spatio-temporal change events of geographic entities into a whole that is related to the process of macroscopic distribution of all geographic entities.b)According to the temporal and spatial characteristics of LCC,the determining conditions of the LCC type and land cover patch spatio-temporal inheritance relationship are proposed.c)A hybrid spatio-temporal data management model based on object and snapshot is proposed to solve the problem of temporal and spatial data storage of land cover.The problem of the relationship between macroscopic distribution variation and microscopic patch change is solved by this model.Based on the main theory of 'basic theory-method of data expression model-model realization-application practice',the related theories and methods of spatio-temporal data model and expression method are discussed in detail,and finally try to solve the problem that the existing models can not meet the LCC spatio-temporal expression,and the relationship between microscopic changes and macroscopic changes in land cover.The model proposed in this paper is helpful to the innovation and development of the method on spatio-temporal data management and expression modeling of areal geographic entities,which provides the basis for further study of spatio-temporal process simulation and change pattern mining of areal geographic entities.