There are over 500 million trees in Switzerland – far too many to examine individually. However, sufficient information can be obtained with random sampling. For this purpose, a 1 km grid was established over Switzerland for NFI1 (1983–1985), and it was determined whether the intersections of this grid were located in the forest. Where this was the case, sample plots were established.
For NFI2 (1993–1995), the mesh size of the sampling grid was extended. Since then, the sample plots have been arranged at intervals of 1,4 km. The sampling grid thus still comprises about 6,600 forest sample plots, slightly more than half of the original number. To compensate for the larger sampling error resulting from the reduction in the number of sample plots, the data have since been evaluated using a two-phase estimation procedure. In NFI2 and in NFI3 (2004–2006), this involved using supporting information from an aerial photo taken using a denser sampling grid (first phase) to stratify the terrestrial sample (second phase). Since NFI4 (2009–2017), an NFI vegetation height model has been used for this purpose.
All terrestrial sample plots are permanently set up so that changes in the individual sample plots can be observed. In order to find the sample plots again for each survey, the plot centres are measured precisely (also using GPS since NFI4) and marked with an aluminium bar buried in the ground.
For NFI1 to NFI3, the surveys were carried out periodically, with all plots surveyed over three years in a given NFI. Since NFI4, however, the surveys have been carried out continuously, with one-ninth of the sample plots – systematically distributed across the whole of Switzerland – being surveyed each year, leading to all plots being surveyed over nine years.
The survey and evaluation methods used by the NFI are continuously being further developed and optimised. For more information please refer to the publications, in particular those on the methods and models used in NFI4.