Research On Semantic Modeling And Inference Under The Geospatial Sensor Web Environment

With the changes of Geospatial environment, in order to adapt to the development of geospatial-national conditions montoring in Smart City, the demands of quantity and complexity of geospatial observation resources increase. The sensor and observation data in Geospatial Sensor Web(GSW) show diversity, and require the interoperability. Therefore, the GSW with semantics has become inevitable. In order to describe the semantics of observation resources in Sensor Web, and to improve the application performance and interoperability, the efficient management and accurate discovery are the major scientific problems under the Geospatial Semantic Sensor Web environment.The observation resources in GSW include sensors and observation data, as well as other information, which are the main source of geospatial data. However, the observation resources have different types and observation mechanisms in the existing model. It means the description of sensors and observations is closed and isolated. Due to the lack of the description model of observation data along with sensor, the associative retrieval can not be achived. It will lead to incomplete query and inefficient retrieval. Therefore, this paper aims to establish observationprocessing-centered semantic description model of observation resource. The model could manage the heterogeneous multi-source resource, provide the protection of semantic registion, and is the model basis of efficient management and accurate discovery.According to the characteristics of observation resources under GSW and the problems of inefficient disocovery for diverse applications, this paper points out that mining the internal relations between observation resources and applications is a major mean for achieving accurate discovery. The paper carries out the semantic modeling and inference from the research of semantic modeling framework of observation resources.Firstly, the paper starts with the summary to the existing coding model of observation data and metadata, and the analysis of semantic requirements under GSW. Through the expansing and resuing the existing observation metadata, the paper proposes semantic modeling framework of Sensor Web resources with spatiotempral-spectral semantics. Based on the SMOF semantic metamodel theory, the paper defines the semantic description model architecture for sensors and observation data, establishes the semantic metamodel and semantic objects of observation resources, which plays a basis of semantic service of observation.Secondly, in order to achieve semantic registry of observation resources, the paper presents observationprocessing-centered GSW ontology. Based on the tope level ontology, the paper established the core classes of sensors and observations, the observation classes for observation process and capability. It could achieve the semantic modeling of observation resources, breaking the restrain of nonstandard and insufficiency description. The GSW ontology is a basis model of semantic registry under GSW environment.Thirdly, in order to achieve the association of spatiotempral-spectral characteristics beween application and sensors, the paper propses spatiotempral-spectral inference method of GSW. By defining the spatiotempral-spectral semantics of geospatial entity, and the semantic features of observation metadata, the inferences rules for geospatial spatiotempral-spectral semantics and observation metadata are described. Then, the rules improve the existing ontology model. Thus, when breaking the limitations of key match in discovery, the inference method lays the feasible method for efficient retrieval.Finally, this paper builds the Geospatial Semantic Sensor Web service platform framework. By using the soil moisture monitoring observation resources in Hubei Province as the experimental resources, the application of propsed semantic modeling and inferences has been examplifies. The results indicate that metadata retrieval with a spatiotemporal-spectral-enhanced method can efficiently achieve fine-grained discovery of qualified observation metadata and obtain soil moisture monitoring information from sensor images. In summary, the spatiotemporal-spectral semantics in the proposed method demonstrate the use of observation metadata, improving the efficiency and accuracy of EO metadata discovery.The innovation of this paper is as follows:1) semantic modeling framework with spatiotemporal-spectral characteristics under Sensor Web environment is built, in order to break the limitations of incomplete observation metamodel description and semantic redundancy. It is the basis to achieve efficient management of GSW.2) The observationprocess-centered GSW ontology is proposed, against the inadequate description. It is the basic model of semantic registry in GSW.3) the spatiotemporal-spectral semantic inference method for’application -observation resources’is established. It could improve the efficiency of observation resources, provide the efficient access and query, and promote the reasonable found of multi-source observation resources under the Internet environment.