Forest Height Estimation and Inversion of Satellite-Based X-band InSAR Data

Objectives Based on TerraSAR/TanDEM-X spaceborne single-polarization interferometric synthetic aperture radar (InSAR) data, this paper investigates the impacts of algorithm selection and coherence coefficient calculation methods on forest height estimation across different scales using InSAR technology.

Methods The DSM（digital surface model）-DEM（digital elevation model） differential algorithm and the sinc model are used for forest height inversion. The effects of X-band penetration on the forest height estimation results are analyzed based on the DSM-DEM differential algorithm, and the effects of the traditional coherence calculation method and the phase-only coherence calculation method on the estimation results are analyzed based on the sinc model. The effects of different scales on the forest height estimation results of the above two estimation algorithms are also clarified.

Results The experimental results show that the DSM-DEM differential algorithm underestimates the forest height, and the coherence calculation method has a significant effect on the forest height estimation results of the sinc model. The estimation results derived from the coherence calculated by the traditional method overestimate the forest height, while the estimation results derived from the coherence calculated by the coherence-only method are in good agreement with the light detection and ranging acquired canopy height model. The two forest height estimation algorithms show a steady improvement in the estimation accuracy with the increa-sing scales.

Conclusions The uncertainty of forest height estimation results can be significantly affected by the adjustment of methods, parameters, and the choice of parameter calculation methods. Both methods achieve reliable forest height estimates, but the coherence-based sinc model has broader practical value because it does not require real data calibration or high-precision DEM. Furthermore, the phase-only coherence calculation yields higher accuracy and is more suitable for forest height inversion in the sinc model.