Lidar point cloud representation of canopy structure for biomass estimation

Abstract:

Laser mapping systems (lidar) have become an essential remote sensing tool for determining local and regional estimates of biomass. Lidar data (possibly in conjunction with optical imagery) can be used to segment the landscape into either individual trees or clusters of trees. Canopy characteristics (i.e. max, mean height) for a segmented tree are typically derived from a rasterized canopy height model (CHM) and subsequently used in a regression model to estimate biomass. The process of rasterizing the lidar point cloud into a CHM, however, reduces the amount information about the tree structure. Here, we compute statistics for each segmented tree from the raw lidar point cloud rather than a rasterized CHM. Working directly from the lidar point cloud enables a more accurate representation of the canopy structure. Biomass estimates from the point cloud method are compared against biomass estimates derived from a CHM for a Juniper savanna in New Mexico.