Mapping Issues

Mapping Approach

The Sensors Team was unable to identify an existing comprehensive ontology for sensors used by the marine science community. A substantially complete sensor vocabulary has been compiled at GCMD, and this was chosen as our basis vocabulary. This list includes many definitions and links to examples. For our purposes, we needed to reorganize it into a more heirarchical and consistent structure.

The top-level classes in the GCMD sensors vocabulary were altimeters, cameras, in situ sensors, laboratory sensors, lidars/lasers, magnetic sensors, photometers, positioning/navigation devices, radars, scatterometers, seismic sensors, sonar/acoustic sensors, sounders/profiles, spectral/radiation sensors, and telescopes. This mixture of domain and measurement type seemed problematic for use in an ontology, and we noted that many sensors would fit into more than one class.

The Sensors team decided that the most logical way to organize sensors is by method of measurement. The top level classes in the newly created sensors ontology are acoustic, electrochemical, electromagnetic, electromechanical, and optical. Some of the GCMD categories can be recast as sensor attributes, such as domain (seismic, or positioning) , or sensor use location (laboratory, or in situ).

With this scheme, any specific sensor (ie acoustic rain gauge) would appear in only one class, although some types of instruments (ie current meter) would be found in multiple classes. Current meters might be acoustic or electromagnetic, rain gauges can be optical or electromechanical. The use of attributes and the availability of search tools for ontologies should make it easy to find sensors that measure a specific parameter, in whatever classes it exists.

Further, for the purpose of our ontology, we needed to define our terms somewhat narrowly. The team decided to consider sensors and samplers as two classes of the superclass instrument. An instrument would be defined as anything that collects data. Samplers are instruments that actually collect physical specimens, so this class would include nets and bottles. A sensor is defined as an instrument that measures, analyses and/or records a physical observable, but does not return any sampled material.

Other potential issues involve the relationship between platforms, instruments and sensors. We define a platform as an entity that carries instruments, and an instrument as an entity that carries sensors. The distinction between sensors and instruments is more complex when intermediate measurements are made, however.

For example, a CTD is considered an instrument because it has multiple sensors that return data that can be used independantly. An instrument that measures multiple parameters to generate a single useful output parameter, such as longwave radiation, is considered a sensor because its intermediate parameters are of no use independantly.

Unresolved Issues

The distinction between instruments and platforms remains to be defined. Using the example of a CTD again, some trigger water samplers, as well as carrying sensors; some also have velocities and location coordinates that are not defined by another platform. These definitions need to be part of the ongoing work of this team or of another group.

Other issues

• enhance our Sensor ontology by adding definitions to the terms
• further discuss/resolve lingering issues like "measurement" (in the instrument) vs. "positional" (in the platform) accuracy
• re-map our Sensor ontology to the BODC Instrument vocabulary
• work with GCMD-SWEET to merge/adopt/adapt our Sensor ontology