Three-Baseline Approach to Forest Tree Height Estimation

Abstract:
In this paper we discuss a new approach to tree height estimation from repeat-pass PolInSAR data. We propose to use jointly three interferometric pairs to constrain a physical model that links the volumetric-temporal decorrelation to forest parameters. The adopted model is the random-motion-over-ground (RMoG) model, here extended from the single-baseline scenario to the three-baseline scenario so that each interferometric pair is characterized by its own level of temporal decorrelation. The parameter estimation problem is formulated by balancing the RMoG model equations with two coherence observations for each pair. We illustrate the three-baseline approach in contrast with the singlebaseline one using L-band E-SAR data collected by the German Aerospace Center during the BIOSAR-2008 airborne campaign.