Predicting Vegetation Structure using Lidar and Multiangle Remote Sensing Data Fusion

Several recent studies have highlighted the importance of lidar in three dimensional vegetation structure characterization. Multiangle remote sensing data, which captures the anisotropic behavior of vegetation reflectance, is also being increasingly used to obtain vegetation structure. Fusion of sparse but accurate lidar data with commercially available remote sensing data is a promising approach towards obtaining seamless forest structure at various scales. This study is an attempt to predict forest structure with statistical fusion of lidar and multiangle Airmisr data. Statistical methods are used to explore the efficacy of Airmisr in predicting lidar measured forest metrics such as canopy height and canopy cover. Several study sites, including Howland Forest, ME and Harvard Forest, MA are compared to evaluate the robustness of techniques and to evaluate the change in relative importance of Airmisr variables for prediction with changing landscape. The results of this study are expected to enhance lidar fusion studies with multiangle as well as other remote sensing data.