The ambiguity in forest profiles and extinction estimated from multibaseline interferometric SAR

Abstract

This paper demonstrates by simulation that in the estimation of vegetation profiles from multibaseline interferometric synthetic aperture radar (InSAR), the peak extinction coefficient is poorly determined for typical interferometric coherence and phase accuracies. This coefficient determines overall density and affects the relative density profiles estimated from interferometry. This paper shows that a given radar power profile gives rise to a family of vegetation density profiles, depending on the peak extinction assumed. It is further demonstrated that estimating the peak extinction requires coherence accuracies of better than 0.1% and phase accuracies of better than a few tenths of a degree, both of which exceed the performance of typical or envisioned SAR systems. Two recommended approaches to profile production with InSAR are 1) use the radar power profile instead of the vegetation density profile for biomass estimation and other ecosystem characterization (in analogy to LIDAR power which is most frequently used for lidar studies of biomass) or 2) apply external information to establish the extinction characteristics needed for vegetation density profiles.