Navigate Our Web Site Solid Earth Geophysics Geothermal Gravity Habitat Magnetics Natural Hazards Topography -------- People Products -------- What's New Data Tools SEG Index SEG FTP site World Data Center for SEG -------- NOAA's National Data Centers NGDC Home Paleoclimatology Marine Geology & Geophysics Solar-Terrestrial Physics

Data Fusion, What is it?

There are many ways to describe Data Fusion. Here are a few working definitions:

1. A simple definition: Data fusion is the seamless integration of data from disparate sources. The data have been integrated across data collection "platforms" and geographic boundaries, and blended thematically, so that the differences in resolution and coverage, treatment of a theme, character and artifacts of data collection methods are eliminated. At present, this is a desirable but unattainable goal.

2. The opposite of data fission: Data fission could be considered the result of developing separate data sets from a single source of data, such as the NOAA Polar Orbiting Environmental Satellites. The data are "separated at birth" for storage in different locations. An example of data fission desceribes legacy data processing from satellite imgry. Data fusion would be the process of re-joining, or integrating, these data. Ideally, the ultimate process would re-evaluate the original separation of the data, and revise the processing technique to keep the data integrated, or to facilitate the re-integration of the data.

Data Fusion vs. On-Demand Mapping..

Note that data fusion may be performed by a data supplier or by the user. A "data supplier" approach may oppose another embryonic area of investigation, called on-demand mapping. In the latter, source data are kept in as close to their original state as possible. Methods for correcting and integrating the data, and for deriving subsidiary products (such as Sea Surface Temperature from AVHRR data) are served up for users. A few discussion points:

With on-demand mapping:

• As techniques advance, the user is given updated processing techniques.
• If a variety of approaches is available, this variety would be made accessable to the user.
• The user is empowered with the latest choice in processing techniques (rather than what was used some time previously by a database developer/supplier.)
• In many cases, the source data and processing methods may often be downloaded more conveniently from source archives than are fully re-processed data.
• Although current technology would place a heavy workload on the user, successful implementation of on-demand mapping may actually ease the user's task, as much post-processing currently needed to adapt data to user requirements would be eliminated.

Thus the location of the data fusion effort (at the server's site, or at the user's site) partially determines the approach to data fusion.

Here are some sites on the World Wide Web that discuss different aspects of data fusion, found by various WWW-based search engines:

• imgry data fusion as discussed by the Electrical Engineering Department of Georgia Tech University
• Coastal Ocean Sensing and Data fusion discussed by the U. S. Navy
• Here's information on the 1st Australian Data Fusion Symposium.
• Hyperspectral data fusion research project at the U. S. Air Force's Air University
• Data Fusion Specialists discussed by HRB Systems
• Data fusion modeling product + services
• Home page of the IEEE/GRSS Data Fusion Technical Committee at the University of Manitoba

Submit your favorite definition of data fusion! Send us your favorite link to discussion on data fusion! Comments are welcome: