Mikko Kaasalainen, abstract

Big forest data and inverse problems: new-generation 3D/4D forest models

Mikko Kaasalainen1, Pasi Raumonen1, Ilya Potapov1, Sanna Kaasalainen2, Raisa Mäkipää3, Risto Sievänen3, Jari Liski4
1TUT, Tampere, Finland, 2FGI, Masala, Finland, 3Luke, Helsinki, Finland, 4SYKE, Helsinki, Finland

Modern growing demands on forest information for multiple ecosystem services cannot be met by the simple and limited trunk volume and canopy size estimates currently in use. Laser scanning techniques have brought about the possibility to map trees and forests efficiently in 3D detail. These quantitative structure models (QSMs) contain any desired geometric, volumetric, and topological properties of the trees. With the advent of lightweight and mobile scanners (ubiquitous laser scanning), this will, for the first time, allow the fast and precise 3D mapping of entire forests from billions of data points. We expand this scheme to 4D (growth predictions) by modifying theoretical plant growth algorithms to have stochastic components that produce the characteristic structural properties for each species.

The measurements are made by a large domestic and international collaboration network that also develops new types of instruments, such as the hyperspectral lidar that allows the identification of the surface material (chlorophylll, moisture, the condition of the tree, etc.) in addition to the laser scanning point cloud. This approach allows the mapping of forests with an unprecedented detail and quality. We have shown by field experiments that, with our modelling, the volumetric accuracy of tree biomass estimates is 10 % or better, while other contemporary methods cannot reach a better than roughly 50 % accuracy and they lack the topological and geometric off-trunk information.