in their national forest inventory (c.g., Tomppo 1990) and similar forest 
environments. The Scandinavian remote sensing and inventory groups also 
have an existing cooperative program of research and information 
exchange supported by the Cooperation Council for Nordic Forestry 
Research. This cooperation provides an existing framework for productive 
research exchange between the Department of Inventory and other Nordic 
institutions working on related problems. 
Cooperation will be a key to reducing cost and enhancing available 
knowledge and skills, and thereby enhancing overall progress. Cooperation 
should also result in access to a more adequate range of experimental data 
and greater appreciation and support from those who will stand to benefit 
from this effort in the future. 
RECOMMENDATIONS: TECHNICAL 
The following are considered key technical recommendations for long term 
success of the program to implement remote sensing in the Swedish NFI. 
Additional breadth and detail related to findings and recommendations is 
summarized in Thomas (1990a). The full report is provided in Thomas 
(1990b). 
(1) The constellation of Swedish NFI remote sensing-related estimation 
objectives should be seen as an integrated whole. Data collection and 
analysis methodologies should be encouraged that can cost-effectively 
provide information products for more than one objective, or which can 
provide key inputs to producing other information products (e.g., forest 
masks for clear-felling estimation being also used in cover type 
stratification). 
(2) Cloud cover could prevent successful application of satellite remote 
sensing technology in some areas of Sweden. In fact, for this reason, and for 
reasons of data acquisition and analysis cost, it is recommended that the 
annual use of satellite remote sensing in the Swedish NFI be planned on an 
adaptable sampling basis (e.g., nominal checkerboard pattern giving 
complete coverage on a five year cycle, adjusted to completely contain 
communes where possible). Results could be statistically linked to complete- 
area image mosaics, when available. Provision should be made to substitute 
current map or aircraft remote sensing data where needed satellite data is 
not available. 
(3) A successful application of the remote sensing technology in the 
Swedish NFI, particularly for obtaining more localized statistics, will require 
some modification or supplementation of the current NFI ground plot system. 
Modification could take the form of obtaining additional data on the 
representativeness of the plots relative to the surrounding forest stand(s). 
Supplementation could take the form of using forest industry forest plot and 
stand data, and National Board of Forestry forest compartment maps. The 
highest priority will be obtaining accurate location for each plot to enable 
accurate co-location with remotely sensed data. 
(4) The most effective use of satellite remote sensing data will likely require 
segmenting (automatically, manually, or both) the satellite imagery into 
forest stands. This approach, in some situations achieved with relatively 
little cost, appears to improve estimation results as well as provide 
information for use in local forest management. 
414