423 
which covers the whole country each year. Tree and stand models, and simulation, 
can be applied to update the data between two image acquisition dates. Other remote 
sensing data such as radar data or different kinds of airborne data can be utilized later. 
3) keep the geographically localized information in a digital form and to transfer it 
easily into the databases of the users; and 
4) estimate the time and spatial variation of variables more reliably than before. 
The new ground sampling design, which includes permanent sample plots, will support 
this task. 
References 
1. Besag, J.E. 198b. On the statistical analysis of dirty pictures (with Discussion). 
J.R.Statist.Soc. B, 48, 259-302. 
2. Tomppo, E. 1987. An application of a segmentation method to the forest stand 
delineation and estimation of stand variates from satellite images. Proceedings of 
the 5th Scandinavian Conference on Image Analysis. Stockholm, .lime 1987. pp. 
253-260. 
3. Tomppo, E. 1989. Comparisons of some classification methods in satellite image 
aided forest tax class estimation. In: SC1A ’89 Proceedings of The 6th Scandinavian 
Conference on Image Analysis. Oulu, Finland, June 19 - 22, 1989. Vol 1. 150-157. 4 
4. Thomas, R. W. 1990. Some Thoughts on the Development of Remote Sensing-Aided 
National Forest Inventory in Sweden. To appear in: Proceedings of the SNS/1UFRO 
Remote Sensing and Forest Inventory Workshop in Uinea, Sweden, February 26 to 
28,1990.