More information from high-density laser data and harvester data

In recent decades, airborne laser scanning (ALS) has become a source of information for the forestry sector. Lantmäteriet (the Swedish mapping, cadastral and land registration authority) has laser-scanned the whole of Sweden, and SLU has produced maps containing estimated forest variables, which the Swedish Forest Agency publishes on its website. The scanning is to be repeated approximately every eight years, making it possible to update the maps and also to produce information on growth and site quality, meaning the production capacity of the land. The laser data collected by Lantmäteriet has 1–2 points per m² and is used to estimate statistical forest measurements, such as volume per hectare and average height.

By scanning at a higher density, it is possible to analyse individual trees. This has been done in a research context and can be used to extract information on tree species and diameter distribution, for example. At national level, the current plan is to continue scanning at a similar density to that used so far. However, several forestry companies have chosen to scan all, or parts of their holdings at a higher density in order to gain more detailed information.

In previous research, we have developed methods for the segmentation of individual trees, and for the estimation of properties such as height and volume of each tree, as well as for the classification of tree species for each individual tree. We have also generated information on both standing and fallen dead trees from high-density laser data.

In August–September 2024, two areas within Mistra Digital Forest were laser-scanned, around Vindeln in Västerbotten, and around Tönnersjöheden in Halland. The scanning was carried out at a point density high enough to enable extraction of information on individual trees, but with the consideration that the cost of covering large areas would remain reasonable, at approximately 25 points per m².

The laser-scanned area around Vindeln measures 1,250 km² and covers both the Svartberget experimental forest, as well as land owned by Holmen and Sveaskog. Five prototype harvesters are operating within the area, recording information about the logs they fell in so-called HPR files. The area around Tönnersjöheden measures 625 km² and covers both the Tönnersjöheden experimental forest, and land owned by the company Silvestica. Currently Södra has a harvester that saves HPR files whilst operating on Silvestica’s land.

For every log the harvesters cut, dimensions such as diameter, length and volume are recorded. Each log is assigned an identity that can be linked to the tree from which it was cut, and details such as tree species are also saved for each individual tree. In addition, the harvesters record the position of the felled trees. Four of the harvesters in the area around Vindeln, and the harvester in the area around Tönnersjöheden, measure tree positions using crane head control, which ensures a high degree of accuracy.

Data from harvesters enables the potential development of models for log properties based on individual trees, segmented from ALS data. Laser data can provide information on the shape and structure of trees, which can be linked to trunk shape and branches. The models can be applied across the entire laser-scanned area to estimate the properties of every tree. Data is collected across large areas, enabling the models to be tested and evaluated for different types of forest.

In addition, the harvester operating in the area around Tönnersjöheden is also equipped with a laser measurement system from Nordic Forestry Automation (NFA). The laser measurement system continuously records data on the trees surrounding the machine. This data can be used to estimate position and diameter, and potentially also tree species, stem shape and branch structure.

Measurements from the NFA system can be used as reference data in order to model the same characteristics, based on individual trees segmented from ALS data. This opens up new possibilities for the estimation of more detailed tree properties. The NFA system also makes it possible to obtain measurements of the trees that remain after thinning. It can be used for forest management planning, as well as for forecasting in order to make decisions regarding future operations.

Mistra Digital Forest is also developing a traceability model for tracking forest products from the forest to the industry. The model is based on a combination of database technology and digital tagging, and on tracking technology using positioning and other sensors. The aim is better management, more efficient industrial processes and an improved working environment. The principle behind the new model is to collect data from harvesters, forwarders, lorries and industrial measurements in one central database.

One possible further development would be to link ALS data to the traceability model. This would make it possible to link laser data from the trees in the forest all the way to measurement data from the industry. This provides the basis for the creation of models for predicting industrial data from laser data, models that can be applied across the entire laser-scanned area. The traceability model will primarily handle data for groups of trees in timber stacks, rather than individual trees. To use laser data effectively, one solution would be to segment individual trees from the laser data, and aggregate information about them within a stand, in order to link them to the traceability model.

Eva Lindberg

Associate Senior Lecturer at SLU and work package leader for Forest Information Systems (WP1) in Mistra Digital Forest