Improved Boreal Forest Height Estimation by Fusion of TanDEM-X InSAR and ICESat-2 Data Based on Sub-aperture Decomposition

TanDEM-X twin-satellite interferometric system is widely used to estimate forest height. However, TanDEM-X acquires single-polarization data in standard mode, which makes it difficult to provide sufficient observation information to support the Random Volume over Ground (RVoG) model invertible. To overcome these problems, this paper proposes a boreal forest height estimation framework based on sub-aperture decomposition, which extracts the surface phase through time-frequency analysis and realizes the wall-to-wall extinction coefficient acquisition driven by Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) canopy height. Specifically, we explored three strategies: 1) TanDEM-X coherence alone; 2) TanDEM-X complex coherence + surface phase; 3) TanDEM-X complex coherence + surface phase + ICESat-2 canopy height. We then tested and validated the results in a Spanish boreal forest test area. The validation results with airborne LiDAR forest height showed that the TanDEM-X forest height inversion performance can be effectively improved by providing the surface phase through sub-aperture decomposition. The third strategy achieved the best consistency by further providing the extinction coefficient of the RVoG model, which improved the forest height estimation accuracy from 4.39 m (Strategy 1) to 2.42 m at 30 [m] spatial resolution.