Research On The Action-Driven 3D Spatio-Temporal Modelling And The Distributed Index

GIS data model,a bridge between the virtual world and real world,has always been the core and foundation of geographic information science.With the rapid development of technologies jn cloud computing,Internet of Things,and big data,a three-dimensional and dynamic environment is required in the field of smart city for information management,scheme simulation,and disposal decision-making,which can further cope with the challenge of climate change,environmental degradation,land degradation,coastal changes,disease transmission,and some other problems.At the same time,with the rapid progress of space information acquisition technologies(e.g.tilt photogrammetry,laser scanning,satellite remote sensing,GNSS and so on)spatial data is becoming smaller in granularity,more dynamic in time,and more diversified in semantic features.Besides,spatial data also tends to be of dynamic linkage.Integrative modeling combining time,space,and semantics is,urgently needed by GIS in the new era.Data model is essentially a simplified representation of the real world,withspatiotemporal cognition being the premise of spatiotemporal modeling.However,because the real world is complicated,human cognition of space-time is generally heterogeneous and multiple.Currently,a normalized and universal expression model in geo-ontology cognition,3D geometric expression,and spatiotemporal data modelling is not yet available.Further,most of the existing data models merely focus oncertain aspects of geometric expression or spatiotemporal process,thus ignoring connections between different research directions,such as geometric expression,spatiotemporal model,and data indexing.In order to overcome the shortcomings of current GIS data models,following the principles of geo-ontology,this thesis explored methods of granularity-based three-dimensional geometric modelling,including the spatiotemporal modelling driven by actions,and the peer-to-peer networks-based distributed spatiotemporal indexing.Finally,the proposed methods were validated through practical applications.The main research contents of this thesis are as follows:1.Actions in geo-ontology.Based on the research progress of geo-ontology,this thesis makes a comparative analysis of the shortcomings of existing geo-ontology,and presented a novel geo-ontology:O(28)(27)S_t,A_t,Action(t _i)(29),which is spatially and temporally integrated model considering comprehensively the basic tasks of GIS,the objective rules of space-time,and the cognitive process of space-time.Starting from the multiple connotations of geo-ontology,this thesis discussed the definition,connotation and classification of actions,and illustrated the similarity between geo-ontology and spatiotemporal modelling.The logical relationship between the three core problems of spatiotemporal modelling was then discussed and a unified theoretical basis and a methodology guidance for further research were provided.2.3D modelling based on semantic granularity.Firstly,in light of the constraint conditions in 3D geometric modelling,this thesis demonstrated that the modelling methods which only consider geometric representation cannot meet requirements of real applications.By analysing extensively the characteristics of geometry,semantics,and cognition of 3D geometric modelling,a multi-level 3D modelling method was proposed based on perceptual space,and the advantages and shortcomings of different 3D geometric modelling were also discussed.Then,the aggregation semantics of'i S-part-of'were extended,with five operation modes(composition,attachment,dissection,inclusion and projection)being built.By using graph theory,the semantic topology diagram and topological string were constructed,which upgrades the 3D modelling from mathematical modelling to pattern descriptions.3.Spatiotemporal modelling driven by actions.In light of the constraints of spatial-temporal modelling,such as multiple scales of time and space,modelling unit diversification,and cognition diversification,this thesis discussed methods of spatial-temporal modelling step by step from the aspects of time ontology and spatiotemporal actions.Subsequently,the advantages of such model in terms of spatiotemporal integration expression,multiple spatiotemporal processes,and multi-dimensional space and self-expansion were highlighted.4.Research on distributed spatiotemporal index.To adapt to the distributed and parallel computing environment,this thesis proposed a distributed space-time index based on Skip Index in the P2P environment,which improved time range retrieval and data skew through time-line index.Such index can search features within spatiotemporal spaces efficiently.Moreover,it is simple,efficient,and stable,and can meet the requirements of spatiotemporal objects in terms of autonomously discovering or retrieving related spatial objects,thus lays a foundation for"activated"spatiotemporal modelling.By introducing actions,this thesis innovatively developed a spatiotemporal geo-ontology,based on which major research contents were performed.The granular3D modelling on the basis of perception space and aggregation semantics,the generalized and normalized spatiotemporal data model driven by actions,as well as the distributed spatiotemporal index were systematically discussed in thisthesis.This research enriches the connotation of geo-ontology and provides a feasible method of integrated modelling of time,space,and semantic facing to multi-element and multi-scale complex space-time,which would effectivelymeet space-time applications in the scene of smart city.