An ontological framework for ocean sensor data interoperability using semantic web technologies for weather applications

Abstract

Ocean and Land based satellite observation system comprises of various sensors and configurations. Sensor data is the major source for any weather-related researches considered by various researchers and domain experts. Government and public organizations started to publish these datasets as open data to the users through climatic web portals. Rapid growth of observation sensors results in copious satellite data which makes information retrieval and decision making a difficult task. Research addressing decision support system for managing the possible impacts on human communities by climatic change is insisting. The traditional knowledge repositories work on static models but challenging in dynamic fields. The intricacy in ocean observing community is to achieve semantic heterogeneity, interoperability and interpretation, leading to a high-end information retrieval system. Heterogeneity among sensor data can be handled by facilitating the data to reach semantically structured through a knowledge base which integrates the data into a single platform for knowledge retrieval and exchange. This research presents a four-phase weather data model through deploying semantic technologies namely; (1) pre-processing of heterogeneous satellite data, (2) developing a ocean knowledge-base through ontology, (3) semantic web processing for integrating real-time database with the domain ontology and, (4) implementing semantic query engine for efficient ontology-based information retrieval system. In first phase, Heterogeneous Geospatial Climatic Data to Resource Description Framework (Hetero-GCD2RDF) approach is presented to extract the records of satellite data and represent it as Linked Data (LD) namely RDF. The ocean data is recorded in various file formats namely; Comma Separated Values (CSV, *.csv), Excel (*.xls), Totals (TUV, *.tuv) and Network Common Data Format (NetCDF, *.nc) that are converted into Resource Description Framework (RDF). Real-time data along south-eastern coastal areas of India is taken as a typical example to execute the proposed approach. The proposed work uses, approximately 170 files of about 650 MB in memory containing 1,278,000 records that are converted into RDF. Compared to conventional methods the proposed method saves nearly 38.15% of time on average in comparison with generic tabular model and saves 61.93% of time approximately when compared to the annotated tabular model. Thus, Hetero-GCD2RDF approach is recognized to be efficient, reliable and suitable for semantic web. newline newline newline