Describing 3-D structure of tropical forests from radar acquisitions: potentials, challenges, and links to ground and lidar measurements

Remote sensing systems have been demonstrated to be a powerful source of information for monitoring tropical forest ecosystems. In particular, Synthetic Aperture Radar (SAR) and LIght Detection And Ranging (lidar) systems provide measurements sensitive to 3-D forest structure parameters at high spatial and temporal resolutions. This is of critical importance given the high spatial heterogeneity and complexity of tropical forests. In particular, SAR acquisitions in tomographic mode enable the estimation of 3-D radar reflectivity profiles. A framework for qualitative and quantitative forest structure estimation from these reflectivity profiles has been recently proposed. It is based on a pair of complementary indices for the characterization of horizontal and vertical structure. In order to ensure the physical interpretation and ecological relevance, the indices have been derived in correspondence to structure indices already established in forestry and ecology studies. This two-index framework has been shown to provide: 1) the systematic interpretation of the 3D radar reflectivity in terms of physical forest structure. This is not always an intuitive task because of the dependency of radar reflectivity on system parameters as well as on the geometric and dielectric properties of the forest. 2) the direct comparison of forest structure estimates from field inventory data to estimates derived from radar and even lidar measurements. In this presentation, potentials and challenges of the application of this framework to tropical forests are discussed. Results from the recent AfriSAR campaign, in which time-coincident SAR and lidar data were acquired over several tropical forest sites in Gabon were acquired, will be presented. The correspondence between the structure descriptors calculated from remote sensing data and ground measurements, together with the behaviour of the obtained forest structure estimates as a function of the spatial scale will be addressed