Forest structure monitoring by means of multi-baseline SAR configurations

Abstract

The penetration capability of microwaves into and through vegetation layers allows scattering interactions across the whole vertical vegetation extend. This, combined with the ability of multi-baseline Synthetic Aperture Radar (SAR) techniques to reconstruct the 3D radar reflectivity opens the door to the use of air- and space-borne SAR configurations to explore and map forest structure parameters on large (global) scales with high spatial and temporal resolution. In this paper we discuss the potential of multi-baseline SAR configurations to estimate (or map) physical forest structure parameters. For this, first the link between 3D radar reflectivity and 3D forest structure has to be established. Accordingly, a framework for qualitative and quantitative forest structure interpretation and estimation from 3D radar reflectivity reconstructed by multi-baseline SAR measurements is introduced, reviewed and validated against experimental data (including Lidar and inventory meassurements) acquired over/in different temperate and tropical forest conditions. The role of system frequency, implementation and spatial resolution is discussed. The dependency of the obtained / possible structure estimates on the spatial scales is addressed and through the analysis on real and simulated data it is demonstrated that a single (spatial) scale does not always ensure a complete characterization of forest structure. The potential of multi-baseline SAR configurations to monitor dynamic effects as the temporal variations of forest structure induced by natural or anthropogenic changes is considered. Results achieved in the framework of actual air-borne SAR (F-SAR) temperate and tropical forest campaigns and/or experiments at different frequencies and/or configurations are critically discussed. Finally, the role of future multibaseline SAR spaceborne configurations / missions as Tandem-L in the context of global forest structure and structure change mapping is discussed.