20th International CODATA Conference
Session: Integrating Heterogeneous and Very Large Datasets for Global Water Cycle Studies: a Case Study for GEOSS

Virtual Integration of Heterogeneous Legacy Weather Databases and Sensor Network Data

Seishi Ninomiya (snino@affrc.go.jp), Takuji Kiura (kiura@affrc.go.jp) and

Masayuki Hirafuji (hirafuji@affrc.go.jp)

National Agriculture and Food Research Organization, Japan

There are several meteorological databases available through the Internet. They are, however, heterogeneous in access methods, data formats, available items etc., so that clients applications are needed to be tightly linked to particular databases. To solve this issue, we developed a data broker named MetBroker that provides client applications consistent accesses to such heterogeneous databases, hiding the heterogeneity and virtually integrating distributed databases. Now, it covers more than 22,000 weather stations of 25 databases, including the database of sensor network nodes called Field Server developed by our group. We have also developed several client applications that obtain meteorological data via MetBroker mainly for agricultural decision supports.

In MetBroker, a wrapper application for each database translates the local heterogeneity of the database to standardize the data for clients. In the original version of MetBroker, the translating procedure is hard-coded in each wrapper, causing static translation that makes the system inflexible. Newly implemented Intelligent MetBroker solved this inflexibility, using Web ontology that allows us to handle semantic heterogeneity. Intelligent MetBroker uses a meta-database that consists of an OWL file for standard vocabulary defined based on the WMO metadata core profile, OWL files for local vocabulary and RDF files for station descriptions. All the services by MetBroker are provided as Web services.

Keywords: agricultural decision support, data Grid, middleware, Web ontology, Web service.