A multi-scale based method for estimating coniferous forest aboveground biomass using low density airborne LiDAR data

Sammanfattning: Forest biomass acts as an important indicator of carbon resources in terrestrial system. Estimation of forest biomass enables a straightforward measurement of carbon storage and provides initial values for process-based carbon cycle models to simulate carbon dynamics. LiDAR (Light Detection and Ranging) remote sensing is increasingly used to estimate forest biomass because of its ability to detect the structure of forest. However, it is still not adequately studied from a viewpoint of multi-scale. This study is a new attempt for the application of multi-scale theory in forest aboveground biomass (AGB) estimation based on low density LiDAR data (less than 1 point/m2). The study area is located in Krycklan catchment which is approximately 50 km northwest of Umeå, Sweden. A method based on local maximum height point identification and downscaling calibration is provided. By implementing local maximum elevation extraction and visualization of aerial images, an algorithm directly based on point cloud data is designed. This algorithm retains more details of the LiDAR data and therefore provides better results. Two calibration look-up tables are provided to approximate the forest AGB derived from low density LiDAR data to the forest AGB derived from high density LiDAR data (more than 1 point/m2). The error of downscaling calibration in the test sample plot is of 0.28%, which proved validity of the method applied.