KEY RECOMMENDATIONS: ORGANIZATIONAL AND INSTITUTIONAL 
(1) Scale-up to operational application should proceed through an orderly 
series of development stages and trials. In an initial small project, a group 
should be established to define an outline of the longer term program and 
details of its near-term elements. 
A phased, integrative program of technique development should be adopted 
in this program. All objectives, cover type as well as wood removals, 
regeneration, etc. to which remote sensing will contribute, can be seen and 
managed programmatically in terms of their relative maturity. That is, each 
information production capability should be developed through a sequence 
of research, system engineering, and finally, operational application. 
Capabilities can come "on-line" when technically ready, and when an 
infrastructure of expertise and facilities has been properly prepared and 
tested. 
(2) An expanded base of experience and skills required for implementation 
of remote sensing technology must be developed in the Department of Forest 
Survey. In order to accomplish this without unnecessary loss of time, it will 
be important to place people experienced in these problems in leadership 
roles from the beginning of the effort. These people can help guide the 
development effort and train other department personnel in this 
technology. 
Working in this way, the contribution of current NFI personnel to the 
development of the remote sensing capability can be made most effective. 
They understand the inventory problem and details of peculiarities 
particular to given areas as seen on the ground and in the statistics. Such an 
understanding can be invaluable in engineering robust procedures. The 
team of personnel experienced in remote sensing (and geographic 
information systems) and the current staff can together best see 
opportunities for linkage of the current procedures and data processing 
system with the new spatial data analysis technology. 
(3) The in-house computer system supporting this effort is likely to be 
workstation-oriented, linked by a network ultimately capable of high data 
rates. A large "number-cruncher" may find use in this network for pushing 
through the big jobs in a timely fashion. Software, while tailored to the 
needs of the NFI, should be "industry standard" wherever possible in order to 
facilitate comparability of results with other groups and a reasonable 
programmer staff budget. 
(4) To accomplish this phased, integrative program, a set of alliances and/or 
contractual relationships with research and development partners should be 
developed. These include firstly the Remote Sensing Laboratory of the 
Department of Biometry and Forest Management of the Swedish Faculty of 
Forestry, and the Swedish Space Corporation. The former can provide in- 
house skills and training, an already existing physical facility, and an 
existing research and development capability. The latter can provide access 
to the large amounts and kinds of satellite data required in the development 
process, and expertise in system application and engineering. Both of these 
groups can in turn benefit by the possibilities for development of practical 
techniques and applications for remote sensing data of value beyond the NFI. 
Another key group of research cooperators or partners can be institutions 
in Finland who have similar goals with respect to the use of remote sensing 
413