Swedish forestry is undergoing large changes in order to meet increasing demands of renewable raw material as a substitute for fossil raw material. Increased forest production must go hand in hand with environmental and social aspects, which raises the question how production can be maximized in a sustainable way.

There are several procedures to choose from to increase Swedish forest production, of which persistent long perspective silviculture is the traditional way, including:

  • soil preparation method
  • regeneration strategy (e.g. method, plant species and genotype, spacing)
  • cleaning and thinning
  • growth increasing management such as fertilizing and enhanced draining
  • efficient harvesting systems and sustainable methods.

The knowledge about how the above-mentioned means of silviculture vary with terrain and climate is well known since long, together with logistics and demand. Low resolution maps and coarse scale planning have often increased the harvesting area sizes and has caused a schematic choice of regeneration method as well as nature conservation areas rather than adjusted to terrain variation. In a long perspective this can result in missing value of forest production and biodiversity.

Aiming for full potential with more precise planning tools

Taking natural landscape variation into account requires a thorough management plan. This has traditionally been based on several maps, knowledge about local production potential and frequently updated inventories and field mapping, all contributing to a functional plan of manageable area sizes. Management areas have traditionally been based on large scale characteristics, e.g. by stand age and rotation period, tree species and possible conservation planning. For practical reasons the management units have measured ca 2-15 hectares, depending on ownership, property size and locality, which is often considerably larger than the actual landscape variation. But a finer resolution has not been practical with traditional maps and planning tools, which has resulted in immense generalizations of complex landscape variation.

Large volumes of data are becoming increasingly available about forest and forest soil with the potential to increase the resolution in planning and silvicultural measures. Examples of available data are high resolution digital terrain models (DTM) from landscape laser scanning, and data from every single tree harvested as collected by the harvester machines and -drivers (HPR-data).

Our hypothesis is that a large portion of Swedish productive forest can produce more timber and biomass than today, by understanding and revealing its full small-scale potential and managing the forest accordingly. To facilitate this challenge in planning, we use digital tools that handle the spatial variation in topography as well as the documented variation in timber production.

Topography works well in predictions

In this study we investigate how the site index varies with topography and consequently combine harvester data and topography to facilitate a higher resolution in forest planning in order to more precisely make use of the full growth potential in the forest.

Skogforsk has earlier presented the planning tool Plantbeställning based on harvester data to create a suggestion to forest planners on plant species and density based entirely on harvester data, which means that the silvicultural practice of the previous rotation period had a high influence on the suggested regeneration. The resolution in that tool was 0,5 hectares, which were in most cases aggregated into larger units of land area.

This study reveals a high potential for topography as a predictor of spatial variation of site index (SI) in the landscape combined with freely available canopy height model data.

Using the tree height information from the harvester data can increase the ability to generalize sample-based site index data (e.g. from the harvester database) onto the landscape level. Moreover, spatial resolution can be tailored for various situations as the method offers the possibility to optimize the model predictions at different spatial resolutions and to specific objects.

Before pursuing further modeling advancements and practical applications it is crucial to present this tool to forest planners and receive their thoughts and ideas about its practical applicability. In the end it is always a matter of comparing costs to gain.

It is crucial to the authors of this project to emphasize that this planning tool should be considered a support to the existing extensive knowledge about forestry and about the land owners often long experience about their own forests.