Computer processing: facilities and training 
The recent increase in the availability of computer-compatible re- 
mote sensing data, both from airborne scanners and the LANDSAT MSS, has 
generated wide interest in computer processing. To date, however, most 
of the development and application of computer classification techniques 
has been done by large research organizations; we are now facing the 
problem of how to transfer this technology to individuals and organiza- 
tions with operational responsibilities. There is a need for: (1) op- 
portunities for "hand-on" training in computer processing techniques for 
resource managers, (2) demonstration of cost-effectiveness of computer 
processing for real problems in a "real-world" context rather than on a 
research scale, and (3) better dissemination of information on costs and 
expected results using systems which are available to resource managers 
on a production basis, or costs required for organizations to develop 
their own computer processing capabilities. In summary, resource man- 
agers need an opportunity to see for themselves what computer processing 
can and cannot do for them, and what the costs are in both time and money. 
Until this occurs there is really no way to accurately determine the ex- 
tent of the practical uses of computer processing techniques, or to direct 
the remote sensing community toward a solution of the most pressing prob- 
lems. 
Improved output products 
Principal products produced from classification of multispectral 
scanner data include map overlays and tabular statistics in a variety 
of formats. Although techniques are available to output classification 
results geometrically corrected to 1:24,000 on a line printer, or to a 
user specified scale on a film recorder, these techniques still are not 
widely used nor are they understood bv the user community. Map overlays 
that are not geometrically corrected, are at the wrong scale, utilize a 
nonstandard classification scheme, and are not compatible with the users 
existing map base are common products but have very little value to a 
user faced with a specific management problem. 
The remote sensing community has the technical capability to im- 
prove the quality and user acceptance of map overlay products. These 
capabilities should be focused on producing low cost colored classifi- 
cation map overlays at user specified scales. Often map overlays are 
used in management briefings or planning more intensive sampling efforts. 
Map overlay products with management or administrative units, watershed 
boundaries, or other user specified strata boundaries overprinted on the 
classification results are repeatedly asked for by natural resource sci- 
entists. Again, the technical capabilities exist but have not yet gained 
widespread use. 
There are several major problems confronting the remote sensing 
community with respect to generating tabular statistics after classifi- 
cation has been completed. In addition to information relative to the 
areal extent of resources depicted by map overlays, the resource man- 
ager needs tabular estimates of the area (e.g., acres or hectares) of 
each resource class and an estimate of the accuracy of the figures. 
Tabular statistics generally are extensively reported for training fields 
and test fields often with questionable statistical justification in