601 
distributed data analysis centres, such a data volume seems unreasonable as a basis for 
routine environmental monitoring. A sampling strategy is suggested, both in space and in 
time, in order to reduce the sheer scale of the problem to tractable levels [Specter and 
Raney, 1989]. However, for active global monitoring, there is no escaping the fact that a 
large amount of data will be required, and with a variety of resolutions, scales, and spectral 
properties, even for one thematic subject such as tropical forests. 
Once acquired, data processing is needed to transform the data into usable 
information. Processing, interpretation, and information extraction must be accomplished 
relatively routinely for environmental data bases, or else the point will be missed. Active 
monitoring requires rapid and reliable transformation of data into usable information on 
which environmental management decisions can be based. 
For global applications, large area coverage is required, as noted above. As a 
consequence, variations in pixel size on the order of 1000:1 are inevitable, stretching from 
the AVHRR resolutions for the global vegetation index [Choudury and Tucker, 1987] to the 
finer resolutions available from SPOT or RADARSAT [Raney, 1990a]. Information is 
gained as resolution is improved, but of course the nominal size of the data base grows with 
the square of resolution ambitions [e.g. Townshend and Justice, 1988]. Both image data and 
tabular files are required, and cross-referenced, in both pixel specific and polygonal classes. 
For SAR (synthetic aperture radar) data in particular, area norms rather than single pixel 
classifiers are appropriate. SAR data, either logically analyzed (for spatial spectral 
signature, for example) or as image files, will need to be merged with information from 
other sensors, or with non-image GIS (geographic information system) records. Finally, for 
many if not all environmental issues, the variation of localized geophysical parameters over 
time is the most critical quantity to monitor. All of these issues are matters of current GIS 
research, although not necessarily vigorously at the present time [e.g. Nagy and Swain, 1987]. 
But these problems are relatively tangible, and substantial progress may reasonably be 
expected over the next decade. 
Availability of the large amounts of data required for global monitoring rather than 
data transformation emerges as the key issue. Thus we turn to considerations of remote 
sensing data policy as it impacts data availability. 
3. REMOTE SENSING IN THE MARKET PLACE 
Current data policy is based on a market place philosophy, for which the watershed 
development in North America was the Remote Sensing Act of Congress of 1984 (PL 98- 
365). This Act set the stage for transfer of operational (land) remote sensing out of NASA, 
with the objective to reach commercial viability for remote sensing within an aggressively 
short period of time, thus removing the burden of support or subsidy from the Federal Go 
vernment. The arguments considered by the U. S. Congress at that time were almost 
exclusively ones of domestic market assessment [e.g., Office of Technology Assessment 1984]