Research On Dynamic Three-dimensional Geospatial Information Service Supported By Stream Computing

As the rapid development of Web technologies and Earth observation technologies,geographical information services(GIServices)become three-dimensional(3D),dynamic,and more intelligent.3D GIServices are gaining investments in the geospatial domain since 3D geospatial data are very commonly used in application.The Open Geospatial Consortium(OGC)defines Web 3D Service(W3DS)and 3D Portrayal Service(3DPS)for the 3D visualization of geospatial data(e.g.,landscape models,city models,textured building models,vegetation objects,and street furniture).Another driving force for this changing trend is the need to solve real world problems.In recent years,human face an increasing number of natural hazards and human-generated emergency situations,such as water pollution,urban waterlog,hurricanes,earthquakes and heavy traffic jams,which erode the living environment and threaten people health and lives.The solving of these problems firstly requires the dynamic simulation of these disasters to help find the causes,control the situations and predict the trends.Sensor Web and high performance computing(e.g.stream computing)technologies are often used to achieve the dynamic 3D simulation of highly time-sensitive geographical processes.Depending on these technologies,3D GIServices supporting on-demand geographic simulation in distributed environments turn into a reachable goal.Currently,3D geographic information systems and services mainly focus on three-dimensional rapid modeling and dynamic visualization.There have been a lot of remarkable achievements in the 3D modelling of indoor/outdoor spaces,automatic consolidation of urban textures and 3D navigation.However,there are several limits in the current 3D GIServices.Current service cannot support(1)a rapid query method for both 3D static archives and dynamic data;(2)a real-time computing method for 3D dynamic variables in Sensor Web environment;and(3)a systematic architecture for constructing 3D geoprocessing services.Against the above three scientific issues,this thesis presents a complete set of solution to build stream computing based 3D GIServices that support on-demand processing.Specific research tasks are performed as follows.Regarding to the first limit,we adopted two different frameworks,Apache Hadoop and Spark framework,to achieve the rapid query of static 3D datasets and analyzed their performance on querying.Stream computing technologies are used in publishing and querying dynamic 3D data.Against the second limit,the paper employs stream computing technique to publish and process sensor observation data.We applied the approach in the urban waterlog case where we define six 3D variables to model precipitation data.For the third limit,the paper presents an architecture of dynamic 3D GIServices that can provide efficient processing capability.A service interface is composed to render 3D scenes using both archived data and live-streaming data.Finally,a prototype system is developed including sensor observation data management module,stream data publication module,data query module,on-demand computing module and online service module.The system can support real-time or near real-time 3D scene rendering in the urban waterlog use case.In conclusion,the contribution of this paper includes:(1)a high-performance query method for 3D static data,and real-time publication and query of stream data;(2)an on-demand computing method for dynamic 3D variables,which utilizes stream computing technique in publishing and processing real-time sensor observation data;(3)a system architecture and interoperable interface for 3D GIServices in the Sensor Web environment,which can support access,query,process,and publishment of real-time sensor data.The proposed methods and architecture are implemented in a prototype system to provide GIservices for dynamic simulation of geographical events.