However, these databases use a variety of logical structures and file formats to store their data. Examples of different logical structures include the Japan MAFF's database of JMS AMEDAS data (Japan MAFF 1999), which stores data in a single table of hourly data values, and a single daily data table, and the New Zealand National Climate Database CliDB (Penney 1997), which stores each element, or related group of elements in a separate table. Many file based systems store each year's data for a particular station in a single file eg (Arizona Cooperative Extension 1999).
MetBroker provides a consistent interface to these disparate weather data for web-based applications.
Easy addition of new databases - the object-oriented design of MetBroker facilitates the addition of new online databases.
Insulation - changes in the structure or implementation of any underlying database will not affect applications. The changes can be accommodated by MetBroker in the relatively small amount of database-specific code.
Performance - MetBroker maintains information about the data available from each station, speeding some types of queries.It updates this knowledge each time it queries for data from a station.
Multilingual support - MetBroker supports multilingual station names (eg kana or romaji in the case of Japanese).
One query, many stations and many databases - With spatial
queries, MetBroker will query all the relevant database and merge the
resulting data into a single answer.
However,any remote database accesses from Java have similar security issues. MetBroker may migrate to CORBA in the future, allowing access for non-Java clients.
MetBroker takes some learning
Using MetBroker requires learning about the classes used for queries
and results.
However, you don't need to know the details of either the particular database structure, or the database vendor's implementation of JDBC.
Two network links instead of one
Direct connections to a single database are likely to be faster than
working through MetBroker. However, MetBroker's knowledge of each station's
data holdings may result in similar overall performance by avoiding queries
for non-existent data.
References
Arizona Cooperative Extension, S., Water and Environmental Science Dept (1999). AZMET : Arizona Meteorological Network, University of Arizona. http://ag.arizona.edu/azmet/
Japan MAFF (1999). Amedas Data, Japan Ministry of Agriculture, Forestry and Fisheries. http://db.cc.affrc.go.jp/npdb/amedas/amedas.html (in Japanese, and restricted to MAFF access)
Penney, A. C. (1997). Climate database (CLIDB) user's manual. Wellington, New Zealand, The National Institute of Water and Atmospheric Research.