© © design and /4/ analysis and interpetation of test results. 
Although the first step is an obvious one, in practice it 
is often quite difficult to get general agreement upon 
what a system is expected to accomplish operationally. The 
second phase, the collection of accurate and unbiased 
ground data, is perhaps the most vital in the evaluation 
of a remote sensing system. All the ground data must re- 
present a sample of the population present in an evalua- 
tion criterion. The sample is the base on which all train- 
ing, testing and evaluation of.the system is accomplished. 
The third phase belongs to the scientific methodology. At : 
| this time, variables defined initially during the objecti- 
e e ve phase are quantitatively tested. In the fourth phase 
the mathematical model is tested using statistical methods, 
and then the interpretation of the results for operatio- 
nal decisions oan be made. | 
In practice the operational applicability of a remote 
sensing system is many times reduced or has little value, 
because often assumptions must be made during the research. 
Such assumptions are required for the analysis but which 
also reduce the applicability of the test results. 
Operational classification algorithms /still under re- 
vision and modification/, usable with or without interac- 
tion are: /1/ clustering, /2/ likelihood ratio, /3/ table 
@ ] & lookup, and /4/ sequential methods. 
.6.0RGANIZATION : 
The recognition of the limits of the earth's natural 
resources has inereased world-wide the research efforts 
to use the environment more effectively. From the point 
of view of this primary objective, the organizational as- 
pect of modern remote sensing research is similar all 
over the world, however, the administrative approach to 
the problem's solution may vary widely. 
6.1. INTERNATIONAL ORGANIZATIONS 
In these organizations, international programs are 
coordinated and evaluated.