Canopy Height Estimation in Boreal Forest via Subaperture Decomposition of Single-Baseline TanDEM-X InSAR Data

To address the underdetermined problem caused by insufficient observations in forest canopy height estimation using single-baseline TanDEM-X InSAR data, this paper proposes an improved estimation method specifically adapted for boreal forests. First, sub-aperture decomposition is employed to generate multiple sublook interferograms from the original single-look complex (SLC) images. Based on the Random Volume over Ground (RVoG) model, the ground phase is then retrieved by fitting a coherence line to the complex coherence coefficients derived from these sublook images. Furthermore, based on an analysis of the geometric distribution of the sublook coherences within the complex plane unit circle, a forest height inversion approach is established within a nonlinear iterative framework by jointly utilizing observations from different sublook interferograms. Two boreal forest study areas (Krycklan and Remningstorp) were selected for validation. The results demonstrate that the sub-canopy topography derived from the estimated ground phase effectively reduces the forest-induced bias present in InSAR-derived DEMs, achieving an accuracy improvement of 60.43% and 49.53%. The forest height inversion yielded RMSEs of 2.51 m (rRMSE = 16.8%) and 3.39 m (rRMSE = 19.7%), confirming the effectiveness of the proposed method in boreal forest scenarios.