(5) Aerial imagery should be considered for use in obtaining subsample data 
extending the available "ground sample" information for some parameters. 
These data can be very valuable for building models to stratify satellite 
remote sensing data into classes of interest, to verify the accuracy of those 
strata, and for direct use in the estimation of area and other forest 
parameters. 
(6) Pursue change detection and labelling research, especially to support 
classification of more subtle change (c.g., thinning, certain types of forest 
damage) using remote sensing data. Key topics include band calibration 
methodology, forest canopy/spectral reflectance modelling, image 
segmentation for change, simultaneous use of different sources of data for 
the same location, ground measurements related to change phenomena seen 
in images, and efficient procedures for producing forest parameter change 
estimates with a remote sensing-aided system. This work should improve 
performance and understanding in all areas of the system. 
(7) It will be important to join with the Canadians and others in evaluating 
the added information content and practical applicability of radar data. 
(8) The utility of all data and information is likely to be enhanced through 
the use of geographic information systems. For example, cover type or 
damage assessment results derived through the Swedish NFI could be 
improved in identification detail and/or spatial accuracy when combined 
with National Board of Forestry compartment map data, or when combined 
with private company stand management data. 
(9) A significant field data set will be required to support this work. In this 
context, the existing NFI data set represents an extremely valuable and rare 
resource. It should be supplemented in a properly designed way, if possible 
through field data collected by other organizations in order to minimize cost. 
(10) It will be important to listen to, cooperate with, and, in some cases form 
well-coordinated partnerships with other groups in order to develop a 
flexible and effective remote sensing-aided Swedish NFI. 
CONCLUDING REMARKS 
The program to incorporate remote sensing in the Swedish NFI is scheduled 
to begin during the summer of 1990. A project will be initiated to evaluate 
clear-felling monitoring techniques for NFI use based on satellite data. The 
project will also prepare a program for the longer term effort to implement 
remote sensing for the fuller range of topics discussed above. It is expected 
that the complete program will be implemented in stages, and will take at 
least 10 or more years to be realized. 
AKNOWLEDGEMENIS 
The authors would like to thank the faculty and staff of the the Swedish 
Faculty of Forestry, Departments of Forest Survey and Biometry and Forest 
Management for their review and input during the Project. Similar thanks 
also goes those at the Swedish Committee on Forest Remote Sensing, the 
Swedish Space Corporation, the University of Helsinki Department of Forest 
Mensuration and Management, the Finnish Forest Research Institute, the 
Danish Bureau of Land Survey, and the Danish Technical University of 
415