One of the first forest inventories in the world based on statistical sampling was carried out in Finland in the beginning of 1920's. The first NFI was planned and conducted by professor Yrjö Ilvessalo. After that, inventories have been repeated in about 5-10 year's cycles. The latest inventory (NFI10) was completed in 2008, and the new NFI11 was started in the summer 2009.
NFI1 1920 - 1924
NFI2 1936 - 1938
NFI3 1951 - 1953
NFI4 1960 - 1963
NFI5 1964 - 1970
NFI6 1971 - 1976
NFI7 1977 - 1984
NFI8 1986 - 1994
NFI9 1996 - 2003
NFI10 2004 - 2008
NFI11 2009 -
The first four NFIs were carried out as line surveys. The field crews walked through the whole country along inventory lines, which run from south-west to north-east, and observed and assessed forest stands that occurred on the line. In the NFI5, inventory lines were disconnected and measurements were done only on rectangle shaped sample plots. However, the volume of the growing stock was still assessed from all forest stands along the inventory line.
In the NFI6 and later inventories, all field measurements and assessments were done in sample plots. A new inventory method, a cluster sampling, was introduced. The size of a cluster of sample plots was planned so that the cluster could be measured in one working day. The sampling design, i.e. the distance between clusters and the distance between sample plots, was a compromise between the accuracy of results and resources available. In Southern Finland, the sampling design remained the same until the NFI9.
In North-Finland, the inventory method in the NFI5, NFI6 and NFI7 was two-phase sampling for stratification. Black and white aerial photographs were used in the first sampling phase. A field sample was allocated to part of the plots in each photo stratum, and field measurements were then generalized for all plots in the stratum.
During the NFI8, a new inventory method based on satellite images was introduced. The NFI sample plots were used as ground truth in the estimation of forest variables according to the spectral values in the images. The aim was to produce information on forest resources also for small areas, and to generate digital forest maps. The new multi-source inventory method did not displace the field inventory: the regional and national statistics of forest resources are still calculated from the field measurements.
The digital volume map covering the whole country, produced by the multi-source NFI8, was utilized when the sampling design of NFI9 was planned. Because the volume of the growing stock was known, it was possible to simulate different sampling designs and calculate the error from the sampling. In addition to the accuracy requirement, the time constraint was also used in simulations. Because forest structure varies between regions in Finland, different sampling designs were found to be efficient in different parts of the country.
Traditionally, the NFI sample plots have been temporary. In the NFI9, some of the sample plots were established as permanent ones. Re-measurements of permanent plots provide additional information about the changes taking place in forest nature, for example about growth, drain and forest damages. The share of the permanent plots was about 20%, and their location was systematic so that one in four clusters consisted of permanent sample plots. However, the number of plots was different in temporary and permanent clusters.
The NFI is continuously developed in order to better meet the changing information needs of the forestry sector. For example, new measurements on the biological diversity of the forests were included in the NFI9. The volume of dead wood and area of rare habitats are examples of variables that indicate biodiversity and could be estimated from the NFI data.