IN SITU OBSERVATIONS: SYSTEMS AND MANAGEMENT
221
¾ Metadata management: Metadata describes the data. Certain
classes of metadata (variable names, units, coordinates, etc.) are
mandatory to any utilization of the data, and must be tightly bound
to data transport as an integral part of the delivery protocols. Other
types of information, such as descriptions of measurement and
analysis techniques, help to place the data in context and are
essential to the overall understanding and usefulness. To be able to
share data among a network it is mandatory to have a common
vocabulary. Some international groups are working together to build
such norms: FGDC and ISO19115 are the most common for
geospatial data. As a lot of system pre-exist to Ocean US, it is
mandatory to develop translation mechanism to build metadata
catalogues that will be used by the Data discovery system.
¾ Data discovery: it is the way to locate data of interest for the user.
This search is done by scanning the metadata catalogue. Depending
on the possibilities that the system wants to offer the user, the
metadata data stored in the catalogues can be more or less precise.
This use of metadata is comparable to the indexing of catalogue
records within a library to help users to locate books of interest. The
common data discovery systems typically allow selecting the
available data for a set of parameters, on a geographical area, within
a period of time. In future, "data mining" techniques will offer
search on semantic criteria ("I want a cloud free AVHRR image of
SST over this area in March 2004 together with SST from drifters
acquired in same area at same time").
¾ Data transport protocols: these are protocol between a user or a
system who wants data and a data repository that stores the data. It is
in this field that significant improvements have been made with
Internet revolution and the increase in network speed. These
protocols are mainly based on available technologies ("web
services", cgi, scripts, etc) based on current transfert protocols
(HTTP, FTP, etc). Each data provider needs to serve its data and
metadata through a common access interface, which can be achieved
with existing softwares such as OpeNDap (alias DODS) Live
Access software, etc. Direct access to these interfaces may require
also specific software or libraries for the user. Although the datasets
are distributed through various nodes within the system, setting up a
centralized query system that will redirect the user requests to the
relevant node can hide this to users.
221
¾ Metadata management: Metadata describes the data. Certain
classes of metadata (variable names, units, coordinates, etc.) are
mandatory to any utilization of the data, and must be tightly bound
to data transport as an integral part of the delivery protocols. Other
types of information, such as descriptions of measurement and
analysis techniques, help to place the data in context and are
essential to the overall understanding and usefulness. To be able to
share data among a network it is mandatory to have a common
vocabulary. Some international groups are working together to build
such norms: FGDC and ISO19115 are the most common for
geospatial data. As a lot of system pre-exist to Ocean US, it is
mandatory to develop translation mechanism to build metadata
catalogues that will be used by the Data discovery system.
¾ Data discovery: it is the way to locate data of interest for the user.
This search is done by scanning the metadata catalogue. Depending
on the possibilities that the system wants to offer the user, the
metadata data stored in the catalogues can be more or less precise.
This use of metadata is comparable to the indexing of catalogue
records within a library to help users to locate books of interest. The
common data discovery systems typically allow selecting the
available data for a set of parameters, on a geographical area, within
a period of time. In future, "data mining" techniques will offer
search on semantic criteria ("I want a cloud free AVHRR image of
SST over this area in March 2004 together with SST from drifters
acquired in same area at same time").
¾ Data transport protocols: these are protocol between a user or a
system who wants data and a data repository that stores the data. It is
in this field that significant improvements have been made with
Internet revolution and the increase in network speed. These
protocols are mainly based on available technologies ("web
services", cgi, scripts, etc) based on current transfert protocols
(HTTP, FTP, etc). Each data provider needs to serve its data and
metadata through a common access interface, which can be achieved
with existing softwares such as OpeNDap (alias DODS) Live
Access software, etc. Direct access to these interfaces may require
also specific software or libraries for the user. Although the datasets
are distributed through various nodes within the system, setting up a
centralized query system that will redirect the user requests to the
relevant node can hide this to users.