The goal of this project was to develop a method for forestry mapping to enable better accuracy and efficiency in forest inventory, based on airborne laser scanning (lidar) and hyperspectral imaging. Lidar data consists of a point cloud in 3D describing tree heights, crown structures and underlying terrain with high accuracy. Hyperspectral images give highly detailed colour information as visible and infrared light is split into hundreds of spectral channels. With correct processing we can thus offer information on the vegetation's biochemical composure such that e.g. wood species, diseases and forest age can be identified. Combining these two methods we have examined new possibilities for forestry mapping to reduce the need for field work and manual data classification.
The present situation in the forest industry in Norway is difficult due to reduced timber prices and high labour cost. The project will use new technology to reduce costs and improve the accuracy of forest inventory mapping.
Currently, forest inventory mapping is based on separate acquisitions of airborne laser scanning data (lidar) and colour images with four broad bands (red, green, blue, and near-infrared). We will develop a new forest mapping service which improves the quality of forest inventory, while reducing the amount of manual work. The new service is based on highly automated processing of simultaneously acquired hyperspectral and lidar data. Hyperspectral data, i.e., images with hundreds of narrow spectral bands, will provide information for species classification and vegetation condition monitoring not currently available in lidar data and broadband colour images, and reduce the amount of manual work.
The combination of hyperspectral and lidar data will allow for automatic, detailed and accurate species classification not possible from other airborne data. This will in turn lead to better timber volume estimates. Manual stand delineation may be done faster, cheaper and with better accuracy using innovative visualisations of the combined hyperspectral and lidar data. The information contained in hundreds of spectral channels also have the potential to reduce the need for field work, thereby reducing cost and speeding up data delivery.
The project will create an updated business model for TerraTec. Instead of offering only basic geodata in the form of point clouds, digital terrain models and multispectral images, TerraTec will now be able to offer an advanced, beyond state-of-the-art forestry mapping service.