Research On The Key Problems In Massive Remote Sensing Data Parameter Product Automatic Production System

With the development of remote sensing technology, remote sensing image analysis results have been used in the areas of global climate change, environmental assessment, land surface change monitoring, water resources management and other aspects. Remote sensing data service has become an important developing direction of remote sensing technology in the future. The remote sensing images processed by some special procedure are called parameter products, which can be used into different science analysis. So the efficient production of remote sensing parameter products is the core problem for future large scale remote sensing data services.According to the requirement of massive remote sensing parameter products automatic production, this paper researches on the key problems of the production system. Research results provide technical support for remote sensing parameter products automatic, continuous and large-scale production. The main innovation points of this paper includes:(l)Establish the domain ontology model of remote sensing parameter products automatic production. Considering of the complicated relationship of the remote sensing images and the requirement of the automatic production, the paper chooses ontology technology to build up the domain metadata model. The domain Ontology model RSPPPDO is presented,which divides the domain ontology into concept ontology and produce ontology. Concept ontology includes satellite ontology, sensor ontology, remote sensing image ontology and computing resource ontology. Produce ontology contains produce plan ontology,produce flow ontology, process type ontology and process component ontology. The ontology structure and relationship are described in the paper. The paper also analyzes the effects of ontology model for automatic production.(2)Propose a mixed pattern based produce flow modeling scheme. Aiming at the continuos production, based on the research of dynamic service composition theory, the paper raised a new produce flow modeling method named PFCMBMM. When design the produce flow,the model allows choosing the process node by process type and by process component, which is called hybrid pattern. This kind of modeling method reduce the difficulty of flow design and provide a guarantee for continuous production. Aiming at transfer PFG to PJG, an algorithm called PCSAPM is put forward which can determine the appropriate process component during produce job flow generating.(3)Put forward a mechanism for automatic production of remote sensing parameter products. Facing the requirement of automatic and efficient production, a hierarchical mechanism is presented which include produce plan layer, produce job layer, produce task layer and produce execution layer. Start with produce plan destination, the system decomposes the produce plan into concurrent execution produce jobs, then the produce jobs split into many produce tasks, which can be distributed to produce execution layer to execute.This mechanism reduces the coupling among produce tasks, ensures high concurrency production, which lays the foundation for massive remote sensing parameter products production. Some key algorithms include PJCCA, PTCADD, PTSAED are raised during the automatic production.(4)Research and develop the prototype system of remote sensing parameter product automatic production. On the basis of key problem research, an prototype system is designed, which is based on the ontology base and composited with user interface subsystem, data catalogue subsystem, produce model management subsystem, produce management subsystem and produce execution subsystem. The system execution process is designed directed by the hierarchical production driven mechanism, which ensures the automatic production. The hybrid patten produce flow modeling schema is used in the produce model management subsystem, which supporting the flexible produce flow definition and The reliability of the production. PJCCA algorithm and PTCADD algorithm are applied in produce management subsystem,which realize the concurrence execution of production. The cloud computing environment provides the distributing production environment and brings good scalability for the system. In the prototype system, parameter products can be produced driven by the produce plan automatically, reliably and efficiently. The test results show that the research findings aiming at the key problem are feasible and reasonable.