SATELLITE IMAGE-BASED NATIONAL FOREST INVENTORY 
OF FINLAND 
ISPRS, COMISSION VII, Mid-Term Symposium 
Global and Environmental Monitoring, Techniques and Impacts 
September 17-21, 1990 Victoria, British Columbia Canada 
Erkki Tomppo 
The Finnish Forest Research Institute 
Unioninkatu 40 A 
SF-00170 Helsinki, Finland 
1. INTRODUCTION 
The National Forest Survey (NFI) data of Finland has been based until today on 
field measurements only. Systematic cluster sampling has been applied in the last four 
inventories in Southern Finland. It has been considered generally that the area for which 
the results can be computed reliably enough is about 150 000 ha. The inventory of the 
whole country takes about 10 years. This implies that a part of the data concerning 
the amount of growing stock and the level of growth is out of date all the time. Other 
drawbacks of the recent inventory are: 1) cutting and mortality can not be estimated 
reliably enough by means of the temporary sample plots; 2) diseases and damages can 
not be detected continuously; 3) time variation of the growth can not be estimated 
reliably in different, regions of the country; and 4) results are not in an easy form to use 
by other organizations. 
Using ground measurements only to remove these drawbacks and to get localized 
information is too expensive. Therefore, The Forest Research Institute has started to 
develop a new inventory system. The method exploits satellite image data and digital 
map data as well as other geographical data, for example, meteorological data as well 
as ground measurements. 
2. DATA 
2.1. Satellite image data 
The possible image data are at this moment Landsat TM and Spot images (and 
MOS-1 images). Of these, TM will be used in the first phase. The reasons are: 1) One 
TM scene covers a larger area (about 180A’m x 1800??) than a Spot image (fiOAm? x fi()A*m ). 
This is of importance at the beginning, when a reliable image calibration system is not 
available. This makes it also more possible to obtain clomlfree images covering the whole 
country. 2) I'lie price of TM data is about 0.01 Fmk/ha compared with 0.025 Fmk/ha 
of Spot -data. However, according to our hypothesis, the best results would be obtained 
if both TM and Spot imagery could be used because of t he better spatial resolution of 
Spot. This approach will be used after the computer capacity of The Institute becomes 
large enough. The satellite images will be rectified with a pixel size of 25m x 25m to 
the base map coordinate system. 
419